Therapeutic Devices for Epilepsy

Text-only Preview

Therapeutic Devices for Epilepsy
Robert S. Fisher, MD, PhD
Therapeutic devices provide new options for treating drug-resistant epilepsy. These devices act by a variety of
mechanisms to modulate neuronal activity. Only vagus nerve stimulation (VNS), which continues to develop new
technology, is approved for use in the United States. Deep brain stimulation of anterior thalamus for partial epilepsy
recently was approved in Europe and several other countries. Responsive neurostimulation, which delivers stimuli to
1 or 2 seizure foci in response to a detected seizure, recently completed a successful multicenter trial. Several other
trials of brain stimulation are in planning or underway. Transcutaneous magnetic stimulation (TMS) may provide a
noninvasive method to stimulate cortex. Controlled studies of TMS are split on efficacy, which may depend on
whether a seizure focus is near a possible region for stimulation. Seizure detection devices in the form of shake
detectors via portable accelerometers can provide notification of an ongoing tonic-clonic seizure, or peace of mind
in the absence of notification. Prediction of seizures from various aspects of electroencephalography (EEG) is in early
stages. Prediction appears to be possible in a subpopulation of people with refractory seizures, and a clinical trial of
an implantable prediction device is underway. Cooling of neocortex or hippocampus reversibly can attenuate
epileptiform EEG activity and seizures, but engineering problems remain in its implementation. Optogenetics is a
new technique that can control excitability of specific populations of neurons with light. Inhibition of epileptiform
activity has been demonstrated in hippocampal slices, but use in humans will require more work. In general, devices
provide useful palliation for otherwise uncontrollable seizures, but with a different risk profile than with most drugs.
Optimizing the place of devices in therapy for epilepsy will require further development and clinical experience.
ANN NEUROL 2012;71:157-168
Devices do not cure epilepsy, but they may help to duced by the patient or the bed,1 video algorithms
control otherwise refractory seizures. This overview
detecting rhythmic movements,2 software to recognize
will briefly summarize selected therapeutic devices being
EEG patterns associated with seizures,3 monitoring of
investigated as treatment for epilepsy. The review will
sounds made during a seizure,4 and analysis of heart rate,
not cover diagnostic devices related to electroencephalo-
rhythm, or regularity.5 Video detection methods can rec-
graphic (EEG), magnetic resonance imaging (MRI), or
ognize regular (rhythmic) movements across pixels of a
positron emission tomography scans, chemical delivery
digital image, but such recognition fails with patients
vehicles for drugs, gene therapy, or radiation therapy, nor
under covers. Shake detectors only detect seizures with
will it attempt to be comprehensive. With the exception
repetitive physical movements, but these are among the
of the vagus nerve stimulator, none of the devices dis-
most worrisome, and sudden death is more likely in peo-
cussed in this article is presently approved for use in the
ple with uncontrolled tonic-clonic seizures.6
United States.
Movement-based devices utilize accelerometers that
can measure movement in real time, coupled with soft-
ware. A ``Multi-Modal Intelligent Seizure Acquisition Sys-
Seizure Alert System
tem' developed in Denmark7 detected simulated seizures.
Family members of a person with epilepsy may worry
Cuppens et al8 documented detection by accelerometers
that a seizure is happening out of the range of observa-
on the leg in 3 patients during the special case of frontal
tion, particularly while family members are asleep. Sei-
lobe seizures with pedaling movements. Sensitivity was
zure monitoring devices are being developed to notify of
92%, and specificity was 84%. In 2 separate papers, Nij-
an ongoing seizure. Seizure detection can be done by a
and colleagues9,10 reported
variety of methods, including analysis of shaking pro-
detection of tonic, myoclonic, tonic-clonic, and partial
View this article online at DOI: 10.1002/ana.22621
Received Jun 30, 2011, and in revised form Aug 19, 2011. Accepted for publication Aug 29, 2011.
Address correspondence to Dr Fisher, Stanford University School of Medicine, Department of Neurology, Room A343, 300 Pasteur Drive, Stanford,
CA 94305-5235. E-mail: [email protected]
From the Department of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA.
C 2012 American Neurological Association

ANNALS of Neurology
seizures with positive predictive value >50%. Other
TABLE 1: Methods to Predict Seizures
authors2,8 reported motion detection of seizures in neo-
nates and in children with nocturnal frontal lobe seizures.
Recent seizure in patients whose seizures tend to
One study of a Medpage ST-2 bed shaking monitor1
detected 5 of 8 tonic-clonic seizures, but at a price of
Known seizure precipitants
269 false-positive alarms.
Patient awareness
Three portable devices recently have been developed
Blood flow changes
to directly detect limb or body shaking by a patient hav-
EEG component frequency analysis
ing a tonic-clonic seizure. EpDetect (http://www.epde- uses accelerometers built into many
EEG nonlinear `chaos theory'
smart phones to detect shaking in the 5Hz frequency
Energy bursts
range, lasting for at least 10 seconds. After detection, the
phone sounds an audible alarm, gives time to press a
EEG synchronicity and correlation
cancel button, and if not canceled, automatically phones
an alarm to up to 3 predetermined caregivers. Smart-
EEG high-frequency oscillations
Watch comprises a wrist device with a miniature 3-
Multiple unit response
dimensional accelerometer sensor programmed to detect
rhythmic movements. A study of 6 patients with tonic-
EEG 1/4 electroencephalographic.
clonic seizures occurring in an epilepsy monitoring unit11
detected 7 of 8 seizures, with 1 seizure missed because of
battery depletion. One false detection occurred during
Prediction by EEG Features
sleep. Numerous false-positive detections during waking
An early attempt21 to use EEG pattern recognition to
could be canceled by the patient. A watch called Epi-
predict seizures was successful at detecting seizures, but
Lert,12 evaluated in 15 patients in an epilepsy monitoring
impractical because of a high false-positive rate. Synchro-
unit, correctly identifying 20 of 22 seizures with 8 false
nization among different EEG channels tends to decrease
alarms in 1,692 hours of monitoring.
just before a seizure,22 and bursts of high-frequency
activity appear.23
Early seizure prediction models24 utilized a new
Seizure Prediction
field called nonlinear dynamics or chaos theory to study
Having a seizure would not be nearly so troublesome if
the interictal-ictal transition. A standard EEG plot of
time of seizure occurrence were predictable.13 A series of
voltage versus time does not display information about
international workshops has reviewed goals and problems
whether 1 time segment of the EEG is nonrandomly
with seizure prediction.14 Table 1 lists methods employed
related to another. Using nonlinear dynamics, Iasemides
to predict seizures.
et al25 asserted that ` ...the next seizure can be predicted
91.3% of the time, about 91 min prior to its onset, with
the issue of 1 false warning every 8.27 h.'' Although this
Patient Awareness
level of accuracy has not always been replicated, several
Temporal distribution of seizures is not random.15 Some
other researchers have successfully employed nonlinear
seizures cluster. Patients report precipitating factors or
dynamic methods for seizure prediction.26-29 One
premonitory sensations for seizures16 that can serve as
study30 of 21 patients with implanted mesial temporal
predictors of likely seizures. Haut and colleagues17 docu-
electrodes recording 88 seizures over 582 continuously
mented that 21% of 57 patients could predict seizures,
recorded hours concluded that predictive ability was bet-
with specificity of 90% and sensitivity of 37%. A study
ter than random, but sensitivity was only 21 to 42%.
of 83 patients undergoing video EEG monitoring18
Several EEG characteristics other than nonlinear
observed a probability of having a seizure of 0.15 with
dynamics have been used to attempt seizure prediction,
including accumulated energy,31 wavelets,32 and baseline
voltage crossings.33
One of the earliest observations on seizure predic-
tion was that of Wilder Penfield,19 who observed blood
Seizure Prediction and Devices
flow changes in advance of clinical seizures. Weinand
A warning of an impending seizure has little value unless
et al20 noted a 20-minute preictal increase of blood flow
something is done to forestall the seizure or minimize its
in the temporal lobe giving rise to a seizure.
impact. At the simplest level, a warning of a seizure
Volume 71, No. 2

Fisher: Therapeutic Devices
could enable a person to withdraw to a safer or less
embarrassing environment for a seizure. The time hori-
zon for the warning has to allow for such actions, but
not be so long as to produce ongoing anxiety.
A collaborative European effort called EPILEPSIAE
(Evolving Platform for Improving Living Expectation of
http://www. is collaborating with Micromed S.P.A.
( to develop a database of
ictal events to test prediction algorithms.13 A second goal
has been development of a portable EEG recorder using
low power Bluetooth transmission technology. Clinical
trials of seizure prediction using this device are in plan-
ning. In another effort, the National Institutes of Health
has funded the International Epilepsy Electrophysiology
Database, a platform based on cloud computing, to share
large archives of expertly annotated human and animal
intracranial EEG for collaborative research and bench-
marking device performance. This database, promoted by
the International Seizure Prediction Group, is working to
share resources with EPILEPSIAE.
NeuroVista has developed a seizure advisory system,
validated in canines with epilepsy,34 that has progressed
to clinical trials in 3 hospitals in Australia (http://clinical
Recording strips of 16 electrodes are implanted subdur-
ally, in and around the seizure onset region. These
electrodes connect to a subclavicular unit that transmits
EEG to a belt-worn device. The external device continu-
ously analyzes intracranial EEG, and uses algorithms to
FIGURE 1: (A) A Peltier cooling device (cool side down) is
determine the likelihood of seizure onset. The user is
situated over a left frontal seizure focus in an anesthetized
informed of seizure probability by a display of colored
rat. The seizure focus was induced by placement of the c-
lights--blue for low likelihood, white for moderate likeli-
aminobutyric acid antagonist, bicuculline methiodide (2mM).
(B) The upper panel shows spikes at 378C. The middle panel
hood, or red for high likelihood--and is provided with
shows attenuation of spikes with cooling of cortex. The
an audio alert when the seizure likelihood changes.
lower panel shows reversible return of spiking with rewarm-
Once seizure prediction can be shown to have rea-
ing to 378C. [Color figure can be viewed in the online issue,
which is available at]
sonable sensitivity and specificity, then the opportunity
will be present for incorporation into closed-loop thera-
peutic devices. Such devices might stimulate brain or
metal plates of different conductivity produces heat on 1
vagus nerve to reduce seizure likelihood, apply antiseizure
side and cold on another. Figure 1 shows a Peltier device
drugs to regions of the brain or elsewhere, engage brain
placed on a seizure focus in an anesthetized rat (Fisher,
cooling devices, or instruct a patient to take a quick-act-
unpublished data). Epileptiform spikes recorded from
ing (for example, nasal spray) medication dose.
bone screws in rat after administration of the c-aminobu-
tyric acid (GABA) antagonist bicuculline methiodide
(BMI; 2mM) are displayed at the top of the figure (top
2 traces). Spikes decline with cooling cortex to 30C
Cooling to temperatures <27C can reversibly block syn-
(middle 2 traces), and return after rewarming to 37C
aptic transmission, and with it, epileptiform bursting in
(bottom 2 traces). The effects of cooling with a Peltier
animal models.35,36 Cooling therefore might provide a
device linked to software detection systems have been
reversible means to inactivate a clinical seizure focus at
studied extensively by Rothman and associates.37
the time of a seizure. In 1834, Jean Charles Peltier dis-
In the clinical arena, cooling of the brain has been
covered that electrical current passing through 2 adjacent
February 2012

ANNALS of Neurology
hypothermia was said to be useful in the treatment of
status epilepticus in 6 patients refractory to medica-
tions.38 Sourek and Travnicek39 used air and ice cooling
via skull burr holes in 23 patients with refractory epi-
lepsy. Ommaya and Baldwin40 interrupted status epilepti-
cus in 1 patient and suppressed epileptic activity in 6
other patients by cooling the brain with iced saline intro-
duced by catheters placed bilaterally over the cortex and
into ventricles. One study of 4 patients with refractory
status epilepticus showed benefit of endovascular cooling
in terms of seizures, but 1 of the patients died of sepsis,
possibly related to hypothermia.41 A cooling microprobe
to be inserted into tissue has been developed.42 The
major barriers to a practical cooling device are cooling
sulci and deep brain tissue, dissipating heat, and diffi-
culty in obtaining an adequate portable power supply.
Optogenetics, pioneered by Boyden and Deisseroth,43
controls brain cell activity with visible light.44 Certain
single-celled organisms that live in hostile environments
utilize a protein similar to rhodopsin in our retina to reg-
FIGURE 2: (A) In an organotypic hippocampal culture, a yel-
ulate ion concentration in response to available light.
low-orange light (marked by bar) activates channel rhodop-
sin, opens chloride channels, hyperpolarizes transfected
These microbial opsin proteins act as ion channels or
neurons, and inhibits firing. (B) Stimulus train-induced burst-
pumps so as to increase excitation or inhibition. Channel
ing in hippocampal culture slices is inhibited by halorhodop-
rhodopsin-2 produces immediate neuronal excitation in
sin-mediated response to orange light (marked by bar,
middle panel). Region of expanded insets is marked by
response to blue light, whereas halorhodopsin (NpHR)
dotted rectangles. Adapted with permission from Tonnesen
produces inhibition in response to yellow-orange light.
et al.45
Rhodopsins are injected via viral vectors into a brain
structure of interest. Transfection occurs in a subset of
light in hippocampal neurons transfected by NpHR. In
neurons and glia near the injection site. Additional mo-
the lower traces, epileptiform bursting induced by tetanic
lecular biological techniques can confer light sensitivity
afferent stimulation is blocked reversibly by orange light.
to specific subpopulations of cells. For example, after len-
The ability to selectively activate or inactivate cer-
tovirus delivery of the inhibitory halorhodopsin eNpHR3
tain specific neuronal populations likely will have game-
to hippocampus, use of a calmodulin kinase promoter,
changing impact on our understanding of functional
CaMKIIa, leads to relatively selective light sensitivity of
neuroanatomy and neurophysiology in health and dis-
pyramidal neurons and dendrites. The Cre-lox recombi-
ease. Whether it will lead to a practical treatment for epi-
nant technique identifies specific stretches of DNA and
lepsy is less clear. Barriers to applying optogenetic techni-
splices it using Cre-recombinase. Lines of transgenic
ques for clinical epilepsy include the need for a
mice expressing Cre-recombinase in specific cell subtypes
conveniently implantable light stimulation system, dem-
have been linked to rhodopsin genes, rendering only
onstration of safety of viral transfection of these ion
Cre-activated cell types light sensitive.
channels into human brain, and verification of long-term
Reduction of neuronal firing by light-induced hy-
neural control, without producing unexpected changes in
perpolarization should be adaptable to seizure control in
brain function.
suitable systems. Tonnesen and colleagues45 injected a
construct of lentivirus NpHR CaMKIIa into hippocam-
pus of rat pups. Hippocampal slices prepared subse-
Local Drug Perfusion
quently showed pyramidal cell hyperpolarizations to or-
The 16th-century physician Paracelsus commented that
ange light. Light also was able to interrupt epileptiform
the only difference between a drug and a poison is dose.
stimulus train-induced bursting. Figure 2 shows hyperpo-
With the possible exception of cancer chemotherapy
larization and interruption of neuronal firing by orange
agents, this observation is nowhere as true as with
Volume 71, No. 2

Fisher: Therapeutic Devices
TABLE 2: Laboratory Studies of Drug Perfusion for Seizures
Epilepsy Model
Botulinum neurotoxin E
Chronic seizures
Decreased spontaneously
from kainic acid
recurrent seizures
into rat hippocampus
R,S-4-phosphonophenylglycine, Auditory seizures in
a metabotropic glutamate
genetically prone rats
sound-induced seizures
Dean &
Bicuculline (disinhibition)
Maximal electroshock Rostrolateral
Inhibited seizures
Gale 198948
in rat
Gallego 200949 GABA
Amygdala kindling
Elevated threshold
in rat
for generalized seizures
Gasior 200750
Omega-conotoxin N-type
Prolonged attenuation
calcium cannel blockers
seizures in rats
of seizures
Myhrer 200651 Ethanol, propofol,
Soman seizure
Seizure inhibition
muscimol, diazepam
in rats
nigra or area
Myhrer 201052 Various metabotropic
Soman seizure
Increased latency to
glutamate antagonists
in rats
or protected against
Sayyah 200753
All-trans retinoic acid
Inhibits kindled
seizures in rats
Motor cortex,
Reduced photosensitive
seizures in baboon
substantia nigra
Smith 199355
Bicuculline in rat
Reduced duration
and severity of seizures
GABA 1/4 c-aminobutyric acid.
antiepileptic drugs that have narrow therapeutic margins.
Diazepam is not itself a practical drug for perfusion
If we could deliver an antiepileptic drug in high concen-
on human brain, because of its alkalinity and depression
trations specifically to the region of brain involved in
of cardiorespiratory centers. Better candidates for focal
producing seizures, then the therapeutic-to-toxic ratio
brain perfusion to stop seizures include adenosine,58
might be substantially improved.
muscimol,59 and pentobarbital.60 Greater penetration of
Investigators working with animal models of epi-
perfused drug can be achieved by pressurizing the infu-
lepsy have long delivered various putative antiseizure
sion catheter to produce bulk convection.61
medicines to different regions of animal brain, as sum-
Another strategy to distribute infused drug widely
marized in Table 2.
is to deliver it to the ventricular system. Serralta et al62
In 1997, Fisher and associates56 showed that inter-
showed that continuous intracerebroventricular infusion
ictal spikes and seizures produced by the convulsant
of valproic acid in the rat suppressed seizures with mini-
GABA antagonist BMI applied to rat cortex could be
mal side effects. Oommen and the current author63
reduced by local perfusion of diazepam (Fig 3).
showed efficacy against flurothyl-induced seizures after 5
Delivery of diazepam to a BMI-induced cortical
days of gabapentin infusion into ventricles with an
seizure focus based upon an online seizure detection algo-
osmotic pump. A clinical trial of ventricular perfusion of
rithm57 can truncate seizures and prevent other seizures
antiepileptic medication is about to be started by
in a rat.
ICVRx. Safety and efficacy will need to be shown by
February 2012

ANNALS of Neurology
efficacy, but 0.13-milliseconds pulses are less effective.
Stimulation at frequencies <20 per second may allow
increased stimulation of unmyelinated C fibers, with more
autonomic side effects. A controlled study of on-off cycle
durations (DeGiorgio et al76) showed no differences in ef-
ficacy; however, some nonresponders improved when the
on-cycle was later increased. No stimulation parameter set
has yet been shown conclusively to be better than those
used in the pivotal trials, recognizing that individual
patients may respond to various parameter changes.
VNS technology continues to be under develop-
FIGURE 3: (A) The c-aminobutyric acid antagonist drug,
bicuculline methiodide (BMI; 4mM), is placed via cannula
ment. Size of the device is decreasing, such that the Demi-
through a twist drill skull opening onto the left parietal cor-
pulse device is smaller and has improved monitoring of
tex of an anesthetized rat. Bone screws record electroence-
battery life.77 High-field MRI has recently been shown to
phalogram (EEG) at left anterior (LA), right anterior (RA),
be safe, with a 3T GE Signa scanner using a specific T/R
left mid-cortex (LM), right mid-cortex (RM), left parietal (LP),
and right parietal (RP) locations with respect to a reference
head coil,78 but more experience with safety is needed for
(Ref) over the frontal midline sinus. (B) Bipolar EEG at base-
other systems. Externally rechargeable devices are under
line. (C) Epileptiform spikes are seen over the LM electrode
development, so the battery need not be replaced every
after placement of BMI (4 mM). (D) Epileptiform spiking dis-
appears seconds after administration of diazepam via the
few years. Development of remote monitoring and teleme-
catheter. Adapted with permission from Eder et al.56
dicine capabilities for the vagus stimulator is in progress.
The ADNS-300 stimulator system79 can record VNS
clinical trials, and the ideal agent to use is not yet
compound action potentials, affording possible enhance-
ment of understanding of the physiology of vagus stimula-
tion for epilepsy and perhaps better individualization of
stimulation parameters. Recording from the VNS also
VNS is currently the only approved stimulation therapy
holds the possibility of early detection or even anticipation
in the United States, but stimulation has been proposed
of seizures. In rats with pentylenetetrazol-induced tonic
at sites as varied as cerebellum, anterior thalamus, centro-
seizures,80 a measure of energy in the nerve could be used
median thalamus, subthalamic nucleus, hippocampus,
to predict behavioral seizures. Some patients benefit from
caudate, locus coeruleus, corpus callosum, mammillotha-
using a magnet to turn VNS on at the start of a seizure. A
lamic tract, and the cortical seizure focus (see Graves and
trial has begun at Ghent University in Belgium of using
Fisher,64 Lockman and Fisher65 for citations).
ictal tachycardia to trigger stimulation.81
One retrospective study82 found that unilateral
Vagus and Peripheral Nerve Stimulation
interictal discharges, cortical dysplasia, and younger age
VNS was approved in Europe in 1994 and in the United
were predictive of better outcomes. However, most
States and Canada in 1997 for therapy of epilepsy, based
reviews have concluded that it is difficult to predict who
upon pivotal trials in patients with partial and secondar-
will benefit from VNS.70,75 For that reason, external
ily generalized seizures in patients >12 years of age.66-68
stimulation paradigms are of interest as noninvasive
Retrospective studies have reported on 440 patients
screens for whether an implanted stimulating device is
implanted for 2 to 3 years69 and for >10 years.70 VNS
likely to be of value. An auricular branch called the
has been shown to be effective for partial and secondarily
Arnold nerve has been hypothesized to be a potentially
generalized seizures in pediatric populations.71-73 Small
useful test stimulation site prior to device implantation.83
trials74 have suggested efficacy of VNS for some general-
A randomized study of electroacupuncture for pain effec-
ized seizures.
tively used stimulation at this superficial vagal auricular
The clinical VNS device allows a noninvasive pad-
site.84 Transcutaneous stimulation of the left vagus nerve
dle held near the device to program current intensity,
under the tragus of the ear was shown to influence MRI
individual pulse duration, pulse frequency, on-off cycle
blood oxygenation level-dependent signals in the left
time, and intensity and duration of an extra pulse trig-
locus coeruleus, left thalamus, left cingulate, left insula,
gered by a magnet held over the stimulator. Tecoma and
left prefrontal cortex, and bilateral postcentral gyrus.85
Iragui75 reviewed the value of varying these stimulation
DeGiorgio et al86 used superficial stimulation of the
parameters. Pulse width of 0.25 milliseconds may be bet-
supraorbital nerve to identify responders, who were
ter tolerated than those of 0.5 milliseconds, with similar
implanted with a subcutaneous supraorbital nerve
Volume 71, No. 2

Fisher: Therapeutic Devices
stimulator. In an unblinded paradigm, seizure frequency
trends in favor of stimulation, neither seizure frequency
was reduced relative to baseline by 66% at 3 months,
nor EEG epileptiform activity changed significantly with
56% at 6 months, and 59% at 12 months. Efficacy will
need to be validated in a larger controlled, blinded study.
coworkers101 evaluated TMS in 21 patients with localiza-
tion-related epilepsy. TMS was given at 120% of the
Transcranial Magnetic Brain Stimulation
motor-evoked threshold at 1 pulse per second for 15
Electrical shocks to the scalp can activate cortical neu-
minutes twice daily for 1 week. The coil was positioned
rons, but the stimulation tends to hurt87; transcutaneous
over the best estimate of the region of the seizure focus.
magnetic stimulation (TMS) is less painful. Magnetic
Sham stimulation was given with the coil angled away
field-induced brain currents fall off rapidly with distance
from the head. The patients were then observed on a sta-
from the magnetic stimulator coil, so great efforts have
ble drug regimen for 2 months. Neither partial nor gen-
been made to produce coils that can stimulate focally
eralized seizures improved significantly with TMS active
and relatively deeply into brain tissue.88 Figure-of-eight
stimulation in comparison to sham stimulation. A trend
coils are widely in use by virtue of these characteristics.
toward short-term benefit was noted in patients with lat-
Early case series of TMS for epilepsy generally were
eral temporal seizure foci, where magnetic fields would
favorable.89 Nine patients with partial or secondarily gen-
best penetrate the focus.
eralized seizures, 2 from temporal and 7 from extratem-
Experience collectively leaves open the question of
poral regions, were given TMS.90 A round magnetic coil
effectiveness of TMS for epilepsy. One of 3 controlled
stimulated the vertex head region at 1 pulse every 3 sec-
studies showed efficacy. That study targeted stimulation
onds for 2 trains of 500 pulses per day. Weekly seizure
to superficial regions of cortical dysplasia, which may
frequency declined from 10.3 6 6.6 before stimulation
have been a factor in efficacy. Other differences in stimu-
to 5.8 6 6.4, significant at p 1/4 0.048. Subsequent case
lation parameters, such as frequency, intensity, duration
series of TMS showed benefit for seizures in some91-95
of the train, and other factors could have contributed to
and little or no benefit in others.96,97 Positioning of the
different study outcomes. In addition, compared to other
stimulating coil over the seizure focus might be impor-
trials of neurostimulation, TMS trials have stimulated
tant in determination of success, according to 1 study
only during a small fraction of each trial day.
that compared vertex stimulation to targeted TMS.98
Magnetic stimulation is not entirely benign, in that
Three controlled trials of TMS for epilepsy have
it can inadvertently instigate seizures, even with single
been accomplished. In a positive trial, Fregni and associ-
pulses.102 A systematic literature review94 found 16 cases
ates99 targeted TMS to sites of cortical dysplasia in 21
of seizures with TMS. This is a low percentage, given the
patients with medication-resistant seizures. Patients were
thousands of patients exposed to TMS.103 A consensus
subjected to 5 consecutive daily 20-minute sessions of
conference on safety of TMS94 concluded that TMS was
stimulation at 1 per second, using either a figure-of-eight
contraindicated when metallic hardware, such as a coch-
real stimulation coil or a fabricated coil looking and
lear implant or medication pump, was in close proximity
sounding similar to a real coil but delivering no stimula-
to the stimulation site. Special care is required with
tion. The epileptogenic focus was targeted as the site of
untested stimulation parameters, patients with a seizure
stimulation, except in 4 patients with diffuse abnormal-
history or brain lesions or medications that lower seizure
ities, in whom stimulation was delivered to the vertex.
thresholds, or pregnancy or heart disease.
By 2, 4, and 8 weeks after stimulation, seizure frequency
was reduced respectively to 72%, 53%, and 58% of base-
Thalamic Stimulation
line, each of which was statistically significant. EEG epi-
The first devices used to treat epilepsy were forms of
leptiform discharges also were reduced. Two other con-
electrical stimulation. Electrical stimulation to map
trolled studies were negative. Cantello and associates100
human brain function may have started in 1884, when
stimulated 43 patients with medication-resistant predom-
the Cincinnati surgeon Robert Bartholow observed con-
inantly focal cortical epilepsies. After a 12-week baseline,
tralateral movements with electrical stimulation of the
TMS was initiated via 2 stacked stimulating coils over
cortex during repair of cranial osteomyelitis.104 Wilder
the vertex. Active treatment was stimulation with the coil
Penfield and Herbert Jasper pioneered the technique of
near the scalp, and sham treatment was stimulation by
mapping cortex with electrical stimulation, and Spiegel
the upper coil distant from the scalp. Stimulation was set
and Wycis of mapping and sometimes stimulating deep
at 2 daily series of 500 stimuli at 0.3Hz, separated by a
structures104; however, these investigators did not use
30-second interval. The stimulus intensity was 100% of
stimulation as treatment. The first therapeutic brain
the motor-evoked threshold. Although the study showed
stimulation efforts were in the field of psychiatry, by
February 2012

ANNALS of Neurology
Heath105 and Delgado et al106 in the early 1950s. Some
of Heath's patients had epilepsy as well as psychiatric
problems, and epileptiform spikes were observed at septal
nuclei and other stimulation sites.105
Deep brain electrical stimulation to reduce seizures
is credited to the New York neurosurgeon Irving Cooper,
who reported improvement in seizure frequency with
stimulation either of the cerebellum107 or the anterior
thalamus.108 Cooper's positive results were qualitative
and uncontrolled, with little detail on individual degrees
of improvement and comorbid conditions. In subsequent
years, about a dozen uncontrolled studies showed benefit
of cerebellar stimulation (DBS) to treat epilepsy, but 2
small blinded studies were negative.109 DBS for epilepsy
FIGURE 4: In the SANTE trial of anterior nucleus stimula-
tion, patients receiving active stimulation (solid line) have
fell out of favor for many years, but interest in it was
fewer median seizures as a percentage of baseline than do
renewed by the success of VNS for epilepsy and DBS for
those receiving sham stimulation (dashed line). Adapted
movement disorders. After the cerebellum, centromedian
with permission from Fisher et al.114 [Color figure can be
thalamus was the primary target of stimulation, pio-]
neered by the Velascos in Mexico City,110,111 but a small
crossover trial was negative.112
A series of studies showed benefit of DBS of ante-
The control group received 5V stimulation at the end
rior thalamus in experimental models of epilepsy.113
of the blinded phase. Seizures declined over the next
Based upon promising animal experimentation and the
2 months to levels encountered in the initially stimulated
early work of Cooper, 6 small unblinded trials of anterior
group. Improvement was sustained, with seizures in the
nucleus stimulation for medication-resistant epilepsy were
terminal 3 months of stimulation at 3 years measuring a
published,64 showing a conglomerate mean 47% reduc-
median 58% reduction compared to baseline. In the
tion in seizures compared to baseline. Uncontrolled stim-
blinded phase, stimulation produced significant reduction
ulation studies are subject to several types of potential
in injuries due to seizures, frequency of complex partial
bias, including placebo effect, regression to the mean,
seizures, seizures originating from the temporal lobes, and
microlesion effects from electrode placement, and other
seizures predesignated as `most severe' by the patient. Re-
unknown confounding factors. Therefore, Fisher and
sponder rates for 50% improvement and quality of life
associates114 performed a randomized, placebo-con-
did not significantly improve during the 3-month blinded
trolled, multicenter trial of anterior nucleus stimulation
phase, but did in the open-label and long-term follow-up
in patients with medication-resistant partial and second-
stages from 1 to 3 years after implantation. In the long-
arily generalized seizures, called SANTE, for stimulation
term phase, 14% of patients became seizure-free for at
of the anterior nucleus of the thalamus for epilepsy. Ran-
least 6 months. Patients who previously had not benefit-
domization was performed on 110 patients either to 5V
ted from VNS or epilepsy surgery had the same favorable
or to 0V (placebo) stimulation of bilateral anterior nuclei
response to DBS as did the overall group.
of the thalamus, at 145 pulses per second, with 0.9-milli-
Complications of stimulation consisted of occa-
second pulses referential to the stimulation case, with
sional chest or other paresthesias, need for repositioning
stimulation on for 1 minute and off for 5 minutes. The
leads, and superficial infections. No symptomatic brain
group had a median of about 20 seizures per month and
hemorrhages were seen, although neuroimaging showed
a mean of 57 seizures per month at baseline. Stimulation
asymptomatic blood in 5 patients. Neuropsychological
was begun 1 month after implantation of the deep brain
tests showed no difference in cognitive or profiles of
leads, and continued for a 3-month blinded phase. Fig-
mood scores, but more stimulated patients reported
ure 4 shows seizure frequency relative to baseline.
symptoms of depression and memory impairment. Five
Seizure frequency declined 20% in the month after
patients had status epilepticus, 2 related to initiation of
implantation prior to initiation of electrical stimulation,
stimulation, and resolving with reduction of voltage.
due either to nonspecific or to microlesion effects. By the
Rates of depression, status epilepticus, depression, sui-
end of the blinded phase, the treated group continued to
cide, and sudden unexpected death in epilepsy all were
improve, to a median level 40.5% less than baseline,
within the expected ranges for a population of people
compared to only 14.5% in the 0V group (p 1/4 0.038).
with refractory epilepsy.
Volume 71, No. 2

Fisher: Therapeutic Devices
44% each; at 2 years it was A53% and 55%, respec-
tively. There were statistically significant improvements
in overall quality of life (QOL) and in 9 of 16 Quality
of Life in Epilepsy Inventory-89 scales at 1 and 2 years
after implantation. There was no deterioration in any
neuropsychological measure, and there were statistically
significant improvements at 1 and 2 years postimplanta-
tion in verbal function, visual-spatial processing, mem-
ory, and mood. Stimulation was well tolerated.
At the time of this writing, based upon the SANTE
trial, DBS is approved for clinical use in Europe and sev-
eral other countries, but the US Food and Drug Admin-
istration (FDA) is still evaluating the risk-benefit bal-
ance. The RNS system is under evaluation by the US
FIGURE 5: Diagrammatic view of a responsive neurostimu-
FDA. Is a median 40% improvement in seizures (during
lator implanted in the right temporal skull, with a record-
the blinded phase) sufficient to justify the risks of
ing-stimulating strip over the right frontal region and a
stimulating-recording depth wire in the right occipital
implanted stimulators? Each patient and clinician will
region. With permission, courtesy of Martha Morrell of
ultimately need to individualize this answer. However,
improvements in this range can be clinically meaningful,
especially where all else has failed, and when some
The conclusion of the SANTE study was that stim-
improve markedly with stimulation. Experience is cur-
ulation of the anterior nuclei of thalamus reduced the
rently insufficient to recommend when to use thalamic
number of seizures in patients with medication-resistant
or responsive neurostimulation in relation to VNS. The
epilepsy. Complications were similar to those encoun-
latter clearly is less invasive. Responsive stimulation
tered with DBS for movement disorders, with additional
requires knowing where to place the stimulators. Clinical
concerns raised about possible subjective symptoms of
trials are underway for hippocampal stimulation, and
depression and memory impairment.
testing at other central nervous system sites are in
Responsive Neurostimulation
A second randomized trial of neurostimulation employed
a strategy to stimulate subdural strips or depth electrodes
placed near seizure foci, in response to electroencephalo-
Therapeutic devices to treat epilepsy described in this
graphically detected epileptiform activity.115 A total of
review represent an incomplete list. Some devices may
191 patients with medication-resistant partial or second-
come to the bedside, whereas others will be found inef-
arily generalized seizures were implanted with a respon-
fective, too invasive or too difficult to implement. Clini-
sive neurostimulator (RNS) affixed within a craniotomy,
cal trials of devices are more problematic than are medi-
and connected subcutaneously to a subclavicular stimula-
cation trials. Medical devices are expensive, although not
tor (Fig 5).
necessarily more expensive than chronic therapy with
Patients were randomized to receive active or sham
some drugs. Nevertheless, devices can provide clinically
stimulation. Stimulation was begun 1 month after im-
meaningful benefits in people who have failed less inva-
plantation, and the 3-month blinded test phase began 2
sive therapies. Devices are not curative, and as such, de-
months after implantation. Improvement was similar to
vice therapy will serve as an important bridge to the time
that seen in the SANTE study, with 37.9% mean change
when we can prevent and cure epilepsy.
in seizure frequency relative to baseline for the actively
stimulated group versus 17.3% in the sham stimulated
group, significant by generalized estimating equations at
p 1/4 0.012. In the third month of the blinded evaluation
The author summarizes work that was supported by the
period, the reduction in seizures in the treatment group
James and Carrie Anderson fund for epilepsy research,
reached 41.5%; in the sham stimulation group it was
Susan B. Horngren Fund, CURE Foundation, Epilepsy
9.4% (p 1/4 0.008). The seizure reduction was sustained,
Therapy project, and National Institute of Neurological
and even improved, over time. The median percentage
and Communicative Disorders and Stroke. R.S.F. receives
reduction in seizures and responder rates at 1 year was
no personal support from Medtronic or NeuroPace;
February 2012

ANNALS of Neurology
Stanford received research support from Medtronic to
Haut SR, Hall CB, LeValley AJ, Lipton RB. Can patients with epi-
lepsy predict their seizures? Neurology 2007;68:262-266.
participate in a multicenter trial.
I thank Dr C. Lee-Messer for helpful comments
DuBois JM, Boylan LS, Shiyko M, et al. Seizure prediction and
recall. Epilepsy Behav 2010;18:106-109.
about optogenetics.
Penfield W. The evidence for a cerebral vascular mechanism in
epilepsy. Ann Int Med 1933;7:303-310.
Potential Conflicts of Interest
Weinand ME, Carter LP, Patton DD, et al. Long-term surface
R.S.F. consults with or holds stock options in NeuroVista
cortical cerebral blood flow monitoring in temporal lobe epi-
lepsy. Neurosurgery 1994;35:657-664.
(seizure prediction), ICVRx (CSF perfusion of drugs),
Viglione SS, Walsh GO. Proceedings: Epileptic seizure predic-
Cyberonics (vagus nerve stimulation), and Intelli-Vision
tion. Electroencephalogr Clin Neurophysiol 1975;39:435-436.
(seizure alert).
Mormann F, Andrzejak RG, Kreuz T, et al. Automated detection
of a preseizure state based on a decrease in synchronization in
patients. Phys Rev E Stat Nonlin Soft Matter Phys 2003;67:
Carlson C, Arnedo V, Cahill M, Devinsky O. Detecting nocturnal
Hughes JR, Fino JJ, Patel K, Domarad W. Factors in the interictal
convulsions: efficacy of the MP5 monitor. Seizure 2009;18:225-227.
record predicting an ictal episode: a case study. Clin EEG Neu-
Karayiannis NB, Xiong Y, Frost JD Jr, et al. Automated detection
rosci 2004;35:158-164.
of videotaped neonatal seizures based on motion tracking meth-
Iasemidis LD, Sackellares JC, Zaveri HP, Williams WJ. Phase
ods. J Clin Neurophysiol 2006;23:521-531.
space topography and the Lyapunov exponent of electrocortico-
Gotman J. Automatic recognition of epileptic seizures in the
grams in partial seizures. Brain Topogr 1990;2:187-201.
EEG. Electroencephalogr Clin Neurophysiol 1982;54:530-540.
Iasemidis LD, Shiau DS, Pardalos PM, et al. Long-term prospec-
Elzawahry H, Do CS, Lin K, Benbadis SR. The diagnostic utility of
tive on-line real-time seizure prediction. Clin Neurophysiol 2005;
the ictal cry. Epilepsy Behav 2010;18:306-307.
Leutmezer F, Schernthaner C, Lurger S, Potzelberger K, Baum-
Drury I, Smith B, Li D, Savit R. Seizure prediction using
gartner C. Electrocardiographic changes at the onset of epileptic
scalp electroencephalogram. Exp Neurol 2003;184(suppl 1):
seizures. Epilepsia 2003;44:348-354.
Walczak TS, Leppik IE, D'Amelio M, et al. Incidence and risk fac-
Le Van Quyen M, Soss J, Navarro V, et al. Preictal state identifi-
tors in sudden unexpected death in epilepsy: a prospective
cation by synchronization changes in long-term intracranial EEG
cohort study. Neurology 2001;56:519-525.
recordings. Clin Neurophysiol 2005;116:559-568.
Conradsen I, Beniczky S, Wolf P, et al. Multi-modal intelligent
Lehnertz K, Andrzejak RG, Arnhold J, et al. Nonlinear EEG analy-
seizure acquisition (MISA) system--a new approach towards sei-
sis in epilepsy: its possible use for interictal focus localization,
zure detection based on full body motion measures. Conf Proc
seizure anticipation, and prevention. J Clin Neurophysiol 2001;
IEEE Eng Med Biol Soc 2009;2009:2591-2595.
Cuppens K, Lagae L, Ceulemans B, et al. Detection of nocturnal
Navarro V, Martinerie J, Le Van Quyen M, et al. Seizure anticipa-
frontal lobe seizures in pediatric patients by means of accelerom-
tion: do mathematical measures correlate with video-EEG evalua-
eters: a first study. Conf Proc IEEE Eng Med Biol Soc 2009;2009:
tion? Epilepsia 2005;46:385-396.
Winterhalder M, Maiwald T, Voss HU, et al. The seizure
Nijsen TM, Arends JB, Griep PA, Cluitmans PJ. The potential
prediction characteristic: a general framework to assess and com-
value of three-dimensional accelerometry for detection of motor
pare seizure prediction methods. Epilepsy Behav 2003;4:
seizures in severe epilepsy. Epilepsy Behav 2005;7:74-84.
Nijsen TM, Cluitmans PJ, Arends JB, Griep PA. Detection of
Esteller R, Echauz J, D'Alessandro M, et al. Continuous energy
subtle nocturnal motor activity from 3-D accelerometry record-
variation during the seizure cycle: towards an on-line accumu-
ings in epilepsy patients. IEEE Trans Biomed Eng 2007;54:
lated energy. Clin Neurophysiol 2005;116:517-526.
Osorio I, Frei MG, Giftakis J, et al. Performance reassessment of
Lockman J, Fisher RS, Olson DM. Detection of seizure-like move-
a real-time seizure-detection algorithm on long ECoG series. Epi-
ments using a wrist accelerometer. Epilepsy Behav 2011;20:
lepsia 2002;43:1522-1535.
Zandi AS, Dumont GA, Javidan M, Tafreshi R. An entropy-based
Kramer U, Kipervasser S, Shlitner A, Kuzniecky R. A novel porta-
approach to predict seizures in temporal lobe epilepsy using
ble seizure detection alarm system: preliminary results. J Clin
scalp EEG. Conf Proc IEEE Eng Med Biol Soc 2009;2009:
Neurophysiol 2011;28:36-38.
Schulze-Bonhage A, Sales F, Wagner K, et al. Views of patients
Davis KA, Sturges BK, Vite CH, et al. A novel implanted device
with epilepsy on seizure prediction devices. Epilepsy Behav
to wirelessly record and analyze continuous intracranial canine
EEG. Epilepsy Res 2011;96:116-122.
Lehnertz K, Litt B. The First International Collaborative Workshop
Javedan SP, Fisher RS, Eder HG, et al. Cooling abolishes neuro-
on Seizure Prediction: summary and data description. Clin Neu-
nal network synchronization in rat hippocampal slices. Epilepsia
rophysiol 2005;116:493-505.
Balish M, Albert PS, Theodore WH. Seizure frequency in intracta-
Burton JM, Peebles GA, Binder DK, et al. Transcortical cooling
ble partial epilepsy: a statistical analysis. Epilepsia 1991;32:
inhibits hippocampal-kindled seizures in the rat. Epilepsia 2005;
Fisher RS, Vickrey BG, Gibson P, et al. The impact of epilepsy
Hill MW, Wong M, Amarakone A, Rothman SM. Rapid cooling
from the patient's perspective: I. Descriptions and subjective per-
aborts seizure-like activity in rodent hippocampal-entorhinal sli-
ceptions. Epilepsy Res 2000;41:39-51.
ces. Epilepsia 2000;41:1241-1248.
Volume 71, No. 2