www.thelancet.com/neurology Vol 9 April 2010 373
Lancet Neurol 2010; 9: 373–80
March 4, 2010
See Refl ection and Reaction
Departments of Neurology,
Epidemiology, and Population
Health, and the Montefi ore
Headache Center, Albert
Einstein College of Medicine,
Bronx, NY, USA (R B Lipton MD);
Department of Neurology,
Mayo Clinic Arizona, Phoenix,
AZ, USA (D W Dodick MD);
Department of Neurology, and
the Jeff erson Headache Center,
Thomas Jeff erson University
Hospital, Philadelphia, PA, USA
(S D Silberstein MD); Michigan
Head Pain and Neurological
Institute, Ann Arbor, MI, USA
(J R Saper MD); Swedish Pain
Center, Seattle, WA, USA
(S K Aurora MD); Armstrong
Atlantic State University,
Savannah, GA, USA
(S H Pearlman PhD); Neuralieve,
Sunnyvale, CA, USA
(R E Fischell ScD); Innovative
Analytics, Kalamazoo, MI, USA
(P L Ruppel MPH); and
Headache Group, Department
of Neurology, University of
California, San Francisco, CA,
USA (P J Goadsby MD)
Richard B Lipton, Department of
Neurology, Albert Einstein College
of Medicine, 1300 Morris Park
Avenue, Rousso Building, Room
332, Bronx, NY 10461, USA
Single-pulse transcranial magnetic stimulation for acute
treatment of migraine with aura: a randomised,
double-blind, parallel-group, sham-controlled trial
Richard B Lipton, David W Dodick, Stephen D Silberstein, Joel R Saper, Sheena K Aurora, Starr H Pearlman, Robert E Fischell, Patricia L Ruppel,
Peter J Goadsby
Background Preliminary work suggests that single-pulse transcranial magnetic stimulation (sTMS) could be eff ective
as a treatment for migraine. We aimed to assess the effi cacy and safety of a new portable sTMS device for acute
treatment of migraine with aura.
Methods We undertook a randomised, double-blind, parallel-group, two-phase, sham-controlled study at 18 centres in
the USA. 267 adults aged 18–68 years were enrolled into phase one. All individuals had to meet international criteria for
migraine with aura, with visual aura preceding at least 30% of migraines followed by moderate or severe headache in
more than 90% of those attacks. 66 patients dropped out during phase one. In phase two, 201 individuals were randomly
allocated by computer to either sham stimulation (n=99) or sTMS (n=102). We instructed participants to treat up to
three attacks over 3 months while experiencing aura. The primary outcome was pain-free response 2 h after the fi rst
attack, and co-primary outcomes were non-inferiority at 2 h for nausea, photophobia, and phonophobia. Analyses were
modifi ed intention to treat and per protocol. This trial is registered with ClinicalTrials.gov, number NCT00449540.
Findings 37 patients did not treat a migraine attack and were excluded from outcome analyses. 164 patients treated at
least one attack with sTMS (n=82) or sham stimulation (n=82; modifi ed intention-to-treat analysis set). Pain-free
response rates after 2 h were signifi cantly higher with sTMS (32/82 [39%]) than with sham stimulation (18/82 [22%]),
for a therapeutic gain of 17% (95% CI 3–31%; p=0·0179). Sustained pain-free response rates signifi cantly favoured
sTMS at 24 h and 48 h post-treatment. Non-inferiority was shown for nausea, photophobia, and phonophobia. No
device-related serious adverse events were recorded, and incidence and severity of adverse events were similar between
sTMS and sham groups.
Interpretation Early treatment of migraine with aura by sTMS resulted in increased freedom from pain at 2 h
compared with sham stimulation, and absence of pain was sustained 24 h and 48 h after treatment. sTMS could be a
promising acute treatment for some patients with migraine with aura.
Migraine is a primary headache disorder aff ecting about
18% of women and 6% of men in the USA and western
Europe,1,2 and it consists of two major forms—migraine
without aura and migraine with aura.3 Migraine with aura
is characterised by reversible focal neurological features
that usually precede the onset of headache.3 Usually,
symptoms of visual aura include spots of light, zigzag
lines, or regions of visual loss (scotomas); non-visual
symptoms include somatosensory features (tingling and
numbness) and language or motor eff ects (weakness).4
Migraine with aura aff ects about 20–30% of patients.
The presumed substrate of migraine with aura is
cortical spreading depression.5,6 This eff ect consists of a
wave of excitation followed by a wave of inhibition of
both neurons and glia, which spreads across the cortical
mantle. Work in animals suggests that cortical spreading
depression depolarises meningeal nociceptors giving rise
to migraine pain,7 and that transcranial magnetic
stimulation (TMS) disrupts this depression wave.8
TMS is a non-invasive technique that applies a brief
magnetic pulse to the scalp and underlying cortex,
changing the pattern of neuronal fi ring. Since its discovery
in 1985, it has been used for anatomic brain mapping and
for diagnostic and therapeutic purposes.9 TMS was tested
in individuals with migraine based on the hypothesis that
a fl uctuating magnetic fi eld delivered by the device, when
applied to the back of the head, would induce electrical
current and disrupt cortical spreading depression. In view
of its ability to disrupt the depression wave, TMS has
promise as a non-invasive, non-pharmacological, and safe
treat ment for migraine.10 We aimed to assess the effi cacy
and safety of a novel, portable, single-pulse TMS
(sTMS) device for acute treatment of migraine with aura.
We did a randomised, sham-controlled, double-blind,
parallel-group study in two phases at 18 centres in the
USA. Patients were eligible for the study if they were
www.thelancet.com/neurology Vol 9 April 2010
aged 18–70 years and met International Classifi cation of
Headache Disorders (second edition) criteria for migraine
with aura (classifi cation code 1.2.1).3 They had to have
between one and eight migraine episodes per month,
with aura preceding migraine for at least 30% of episodes,
followed by moderate or severe headache in 90% of
attacks. Individuals undergoing randomisation (in phase
two) had to have at least one migraine with aura episode
during the study’s 1-month lead-in phase (phase one).
Key exclusion criteria were: aura that lasted more than
60 min; presence of metal implants; headaches due to
underlying pathology or trauma; and overuse of drugs
for headaches or use of drugs that could confound
interpretation of study results.
We undertook the study in accordance with recognised
Good Clinical Practice, International Conference on
Harmonization guidance, and US Code of Federal
Regulations requirements related to the protection of
human subjects, and investigational device exemptions.
All study centres provided institutional review board
approval before study initiation, and all participants gave
written informed consent.
Randomisation and masking
Before the study was initiated, an unmasked statistician
(Innovative Analytics)—who had no involvement with
the remainder of the study—developed a randomisation
scheme (using PROC PLAN, SAS version 8) to assign
sTMS or sham treatment devices to sequential numbers.
Every device used for this study was identifi ed by a serial
number, which corresponded to a patient’s number in
the randomisation schedule. The manufacturer of the
sTMS device (Neuralieve, Sunnyvale, CA, USA) also
developed an identical sham stimulator (see Procedures).
To maintain blinding, both stimulators buzzed and
vibrated for 2 s when administering the active (or sham)
pulse. To further keep the study blind, the appropriate
number of sTMS and sham devices was sent to every
participating site. To assess masking, patients recorded
their perception of treatment allocation as either sTMS
or sham at baseline (before application of the study
device) and immediately after use. Patients, investigators,
and study personnel were unaware of treatment
assignment until the database was locked and the study
In phase one (lead-in phase), we recruited study
participants, trained them to use an electronic diary
(personal digital assistant [PDA]; on average, training
took less than 5 min), and asked them to keep a headache
diary for 1 month. This phase was undertaken to confi rm
the individual’s ability to use the PDA and to verify
prospectively the diagnosis of migraine with aura. In
phase two (treatment phase), we randomly allocated
eligible individuals in a 1:1 ratio to either active sTMS or
sham stimulation, in a double-blind manner.
The sTMS device (Cerena Transcranial Magnetic
Stimulator; webappendix p 1) used in the study is a
handheld, non-invasive, portable (uses an AC/DC
adapter), monophasic machine that delivers brief,
individual, magnetic pulses. Magnetic fl ux fl ows into the
cortex and induces an anticlockwise fl ow of current
during the rising pulse. The device is 32·5 cm long and
weighs 1·54 kg. The sham device is identical in
appearance to the sTMS device.
We gave assistance to every study participant on how
to use the device, with training sessions taking less
than 5 min. Patients applied the machine to the occiput,
just below the occipital bone (webappendix p 1), and
administered treatment consisting of two pulses about
30 s apart. They pressed a “charge” button to recharge
the capacitors in the device and, when the capacitors
were fully charged, a “treat” button was lit. Patients
then pressed the treat button to generate a single
magnetic fi eld pulse of nominally 0·9 T peak (measured
1 cm from the device surface) with a rise time of roughly
180 μs and a total pulse length of less than 1 ms. Patients
heard a brief sound when the magnetic pulse (or sham
pulse) occurred. The patient then pressed the charge
and treat buttons for a second time for the device to
deliver the second pulse. We instructed participants not
to use the study device to treat an episode in which aura
had lasted more than 60 min. We asked patients to
contact the study coordinator at their doctor’s offi ce if
they had any diffi culties or concerns about their
We instructed patients to begin treatment as soon as
possible after aura began and always within 1 h of aura
onset. We permitted participants to treat up to three
attacks. We asked all individuals to return to the study
site at the end of 3 months for the study completion or
early termination visit, irrespective of the number of aura
episodes that had been treated.
Participants recorded baseline and follow-up features
with the PDA. During the treatment phase, they noted all
outcomes on their PDA at baseline and 30 min, 1 h, 2 h,
24 h, and 48 h after treatment. They also reported baseline
pain and associated symptoms within 1 h of onset of aura
immediately before treatment.
monitored use of rescue drugs after treatment. At the
end of the 48 h assessment period after treatment,
participants recorded their global assessment of pain
relief as excellent, very good, good, fair, or poor.
We allowed all patients to continue use of their usual
and necessary medical treatments, including stable doses
of migraine preventive drugs (including, but not restricted
to, β adrenergic antagonists, selective serotonin-reuptake
inhibitors, or calcium-channel blockers). We did not
permit use of analgesics, antiemetics, triptans, ergots, or
other drugs that could confound assessment of the
effi cacy or safety of study treatment within the 12 h period
before treatment. Rescue drugs were permitted 2 h after
See Online for webappendix
www.thelancet.com/neurology Vol 9 April 2010 375
If adverse events arose with treatment, we asked
patients to call the study site to report them within 48 h.
All adverse events were recorded on case report forms at
every site and further noted at the study completion or
early termination visit.
The primary effi cacy outcome was the proportion of
patients who were pain free at 2 h post-treatment for the
fi rst treated attack; although we permitted patients to
treat up to three attacks, this outcome is the usual primary
endpoint in pharmacological studies of acute migraine.
Co-primary effi cacy outcomes were the proportions of
patients who had photophobia, nausea, or phonophobia
at 2 h post-treatment for the fi rst treated attack. For these
associated symptoms, we used a non-inferiority
comparison. We selected the non-inferiority design for
two reasons. First, very little preliminary data exist to
show that sTMS relieves associated symptoms;
nonetheless, we needed to assess whether treatment
might exacerbate associated migraine symptoms. Second,
in a very early treatment design, rates of associated
symptoms are expected to be low,11 limiting power to
indicate a signifi cant diff erence between active and
placebo treatments. Primary and co-primary outcomes
were all prespecifi ed in the protocol.
Secondary effi cacy outcomes were the proportion of
patients who had mild or no pain 2 h after treatment of
the fi rst aura episode, the proportion with a sustained
pain-free response 24 h and 48 h after treatment
(defi ned as pain free at 2 h, no recurrence, and no use
of rescue drugs), the proportion who needed rescue
drugs during every migraine attack, the consistency of
pain-free episodes in two of three aura episodes, the
patient’s global assessment of pain relief, and the
proportion with vomiting. We also assessed the effi cacy
of treatment in patients who used migraine preventive
drugs and response to treatment based on pain severity
at time of treatment.
On the basis of fi ndings of previous studies,12,13 we
expected that about 20% of participants in the sham group
and about 45% of those in the sTMS group would be
pain free from migraine headache 2 h after treatment of
the aura episode. About 80 patients per treatment group
were needed to detect this magnitude of eff ect, with a
two-sided χ² test, α of 0·05, and 90% power. Assuming a
dropout rate of 20%, we needed to enrol about
200 individuals into the treatment phase of the study to
ensure that 80 in every group completed the study.
We used the Cochran-Mantel-Haenszel test to measure
superiority of pain effi cacy outcomes, stratifi ed by centre.
We tested signifi cance of sTMS versus sham treatment
with the statistic for general association (α=0·05) and
assessed centre disparity with the Breslow-Day test for
homogeneity of strata (α=0·10).
To assess non-inferiority, we compared the associated
migraine symptoms of photophobia, nausea, and
phonophobia at 2 h post-treatment in the sTMS and
sham groups for the fi rst treated attack. That is, we
wanted to establish that no signifi cant loss of benefi t
had happened in relief of associated symptoms with
sTMS versus sham stimulation. We calculated the
diff erence (with 95% CI) in the proportion of patients
with every associated symptom 2 h after active versus
sham sTMS and termed this value the symptom
diff erence score. We defi ned clinically meaningful loss
of benefi t as a 5% reduction in benefi t of active
stimulation in relation to sham stimulation. Accordingly,
no loss of benefi t was established if the upper limit of
the one-sided 95% CI on the symptom diff erence score
was less than this clinically meaningful non-inferiority
margin of 5%. We assessed the three associated
symptoms by univariate and multivariate analysis,
adjusting for important baseline covariates.
We accomplished the prespecifi ed multivariate logistic
regression modelling by backward elimination of
non-signifi cant covariates. We used the fi nal logistic
regression model for photophobia, nausea, and
phonophobia to generate the probability of occurrence of
these three symptoms for every patient, adjusting for
important covariates. We ascertained the upper one-sided
95% CI for the diff erence in covariate-adjusted treatment
means from the symptom probabilities for patients in
each treatment group. We ordered all primary hypotheses
Figure 1: Trial profi le
sTMS=single-pulse transcranial magnetic stimulation.
267 enrolled (first phase)
201 randomised (second phase)
99 allocated sham
Safety analysis set
17 did not treat 20 did not treat
11 not per protocol 12 not per protocol
102 allocated sTMS
82 treated one or more attacks
82 treated one or more attacks
71 treated one or more attacks
70 treated one or more attacks
66 dropped out
37 did not treat migraine with aura
29 protocol deviation
9 screening failures
www.thelancet.com/neurology Vol 9 April 2010
tested in this study in a hierarchy to control for multiplicity
with closed-form hierarchical testing.
We used three analysis sets. First, the safety analysis
set included all patients randomly allocated to a treatment
group. We used this set for assessment of adverse events.
Second, the modifi ed intention-to-treat analysis set
included all participants who underwent randomisation
and who treated at least one migraine episode. We used
this set for primary superiority hypothesis testing of the
pain outcome because it avoids overoptimistic estimates
of effi cacy that can result from a per-protocol analysis.
Finally, the per-protocol analysis set included all
individuals with at least one treated migraine with no
missing assessments, no use of rescue drugs before the
2 h assessment, no major protocol deviations, and no
changes in concomitant treatment during the study that
might potentially aff ect response. We used this analysis
set for non-inferiority testing of migraine symptom
outcomes because it is conservative for these hypotheses.
We did not do any post-hoc analyses.
This trial is registered with ClinicalTrials.gov, number
Role of the funding source
This study was designed by RBL, PLR, and a group of
external medical advisers, who also served as investigators
(SKA, DWD, PJG, JRS,
Y M Mohammad). Data analysis was planned by RBL
and PLR and implemented by PLR. Neuralieve (study
sponsors and developers of the portable sTMS device
tested herein) managed the study and provided study
data as requested for preparation of the report. REF is an
SDS, S Diamond,
employee of Neuralieve, had the idea for use of sTMS to
treat migraine, and designed the sTMS device used in
the study. All authors had complete access to the statistical
report prepared by Neuralieve. RBL and PLR had full
access to all data and made the fi nal decision to submit
for publication, in collaboration with the other authors.
Between August, 2006, and February, 2008, 276 patients
from 16 centres in the USA (two centres did not enrol any
patients) were screened for inclusion in phase one of the
study. Figure 1 shows the movement of individuals through
the study. 267 participants were enrolled, of whom 201
were randomly allocated into phase two (safety analysis
set). Of this population, 164 (82 sTMS and 82 sham) treated
at least one aura episode (modifi ed intention-to-treat
analysis set). Three patients in the sham group discontinued
the study, for absence of a treatment response (n=1),
non-compliance (1), and dropout with no reason given (1).
In the sTMS group, seven individuals discontinued, for
exclusion criteria (1) and protocol (1) violations, a change in
migraine prophylaxis drug (1), withdrawal of consent (1),
and loss to follow-up (3). No patients discontinued because
of adverse events from treatment.
Table 1 shows baseline characteristics of patients.
Participants were predominately women. Treatment
groups were similar with respect to frequency of migraine
attacks and use of preventive treatments. Of the
201 patients who underwent randomisation, 196 were
taking at least one concomitant drug at enrolment, of
which the most typical were aspirin, paracetamol,
paracetamol and aspirin and caff eine, rizatriptan,
sumatriptan, almotriptan, eletriptan, paracetamol and
hydrocodone, ibuprofen, and naproxen.
In the modifi ed intention-to-treat analysis set, the
proportion of patients achieving a pain-free response 2 h
after treatment for the fi rst attack (primary outcome) was
39% (32/82) for those assigned sTMS compared with
22% (18/82) for those allocated sham stimulation
(therapeutic gain 17% [95% CI 3–31%]; p=0·0179; fi gure 2).
Similar results were recorded in the per-protocol analysis.
More participants achieved the sustained pain-free
secondary outcome (pain-free at 2 h post-treatment with
no recurrence and no use of rescue drug) with sTMS
compared with sham stimulation at 24 h (24/82 [29%] vs
13/82 [16%]; p=0·0405) and 48 h (22/82 [27%] vs 11/82
[13%]; p=0·0327) post-treatment (fi gure 2).
Non-inferiority in symptom prevalence at 2 h was
shown for photophobia, phonophobia, and nausea in the
prespecifi ed multivariate analyses, which were adjusted
for signifi cant covariates (table 2). Associated symptoms
were diminished 2 h after treatment for photophobia
(sTMS 69%, sham 75%), phonophobia (sTMS 51%, sham
62%), and nausea (sTMS 37%, sham 39%). The
prespecifi ed multivariate analysis removed variability
due to important baseline covariates, which resulted in
an adjusted upper one-sided 95% CI that confi rmed
Sham (n=82)sTMS (n=82)
Migraine attacks per month
Use of migraine prophylaxis
Visual plus other aura
Baseline headache intensity at time of treatment for the fi rst attack
Number of treated aura attacks*
Data are number of patients (%) or mean (SD). sTMS=single-pulse transcranial
magnetic stimulation. *In the safety analysis set (sham, n=99; sTMS, n=102).
†Note that patients were asked to treat only three episodes.
Table 1: Baseline characteristics (modifi ed intention-to-treat analysis set)
www.thelancet.com/neurology Vol 9 April 2010 377
non-inferiority for all associated features. In the
per-protocol analysis set, upper one-sided 95% CIs were
–2·69% for photophobia, –5·55% for phonophobia, and
2·39% for nausea (table 2). All were below the prespecifi ed
non-inferiority limit of 5%. Additionally, upper limits for
photophobia and phonophobia were less than zero,
which indicates that superiority in suppression of these
symptoms at 2 h was achieved with sTMS treatment.
Prevalence of migraine-associated symptoms at baseline
varied with intensity of pain at time of treatment. That is,
nausea, photophobia, and phonophobia were less common
in patients with no or mild pain at time of treatment than
in individuals with moderate or severe pain. When baseline
pain was moderate or severe, treatment with sTMS was
associated with a signifi cant reduction of every migraine-
associated symptom (table 3). When baseline pain was
absent or mild, no diff erence was noted between sham and
sTMS treatment in relief of migraine-associated symptoms
(table 3). The proportions of patients achieving relief with
sTMS were similar in both baseline pain subgroups;
however, the proportions obtaining sham relief were
highest for the subgroup with no or mild baseline pain.
Although the primary outcome and key secondary
outcomes (sustained pain-free response at 24 h and
48 h) diff ered signifi cantly between treatment groups
(fi gure 2), some secondary outcomes did not diff er. These
included headache response at 2 h, use of rescue drugs,
and consistency of pain relief response (table 4).
Use of preventive treatment was identifi ed as a
signifi cant covariate for pain response by logistic
regression analysis. Incidence of pain at 2 h was much
lower in patients allocated sTMS treatment compared
with those allocated sham for the subgroup who used
preventive drugs (30/31 with sham [97%] vs 22/34 with
sTMS [65%]; p=0·0014). The absolute risk reduction was
much greater in those using preventive treatments (32%
vs 8% with no preventive treatments [webappendix p 2]).
To assess masking, patients were asked if they thought
they were assigned to the active or sham group before
and immediately after treatment. Before treatment, 67 of
82 (82%) allocated sham and 66 of 82 (80%) allocated
sTMS thought they were in the sTMS treatment group.
Immediately after treatment, 58 (71%) in the sham group
and 55 (67%) in the sTMS group thought they had
received active treatment.
At the conclusion of the study, participants received a
post-study survey to assess errors with and overall
user-friendliness of the TMS device. Clinical coordinators
reported that patients rarely experienced errors in use of
the device. Participants rated the TMS device an average
of 8 on a 10-point scale for overall user-friendliness.
The sTMS device was well tolerated when used to treat
patients with many episodes of migraine with aura. No
deaths were reported during the study. One serious
adverse event (optic neuritis) was recorded during
migraine treatment in phase two, in one of 102 (1%)
patients randomly allocated sTMS. However, this event
arose before treatment of an aura episode with the device;
thus, it was not deemed by the investigator to be related
to the study treatment. No serious adverse events were
reported in the group allocated sham stimulation.
Prevalence of adverse events in participants allocated
sTMS (safety analysis set) was low (14/102 [14%]) and
comparable with that recorded in the sham group (9/99
[9%]; table 5). No clinically important diff erences were
recorded between treatment groups in either the type or
rate of adverse events during the migraine treatment
phase. No patients in either group discontinued from the
study because of adverse events. The proportion reported
to have had at least one treatment-related adverse event
was low in the group allocated sTMS (5/102 [5%]) and
was similar to that noted in the group allocated sham
treatment (2/99 [2%]). All adverse events reported were
Figure 2: Pain-free response at 2 h, 24 h, and 48 h on active and sham
sTMS=single-pulse transcranial magnetic stimulation. Error bars=SE.
Proportion of patients pain free (%)
Hours after treatment
Adjusted for use of preventive treatment and
Adjusted for use of preventive treatment, baseline
phonophobia main eff ect, and treatment interaction
Adjusted for use of preventive treatment and
sTMS=single-pulse transcranial magnetic stimulation. *Diff erence between the proportion of patients with symptoms at
2 h in each group. †Meets criteria for non-inferiority (<5%) and superiority (<0%). ‡Meets criteria for non-inferiority (<5%).
Table 2: Presence of non-headache migraine symptoms 2 h after treatment (per-protocol analysis set)
www.thelancet.com/neurology Vol 9 April 2010
mild or moderate in severity, with the exception of severe
nausea (n=1 in each group), severe migraine (1, sTMS),
and severe headache (1, sTMS).
Our fi ndings show that sTMS was signifi cantly more
eff ective than sham stimulation for treatment of migraine
with aura, as judged by the primary outcome—the
proportion of patients who were pain free at 2 h
post-treatment for the fi rst treated attack. This pain-free
period was sustained (24 h and 48 h after treatment) in
more participants allocated sTMS versus sham
stimulation. Compared with sham-treated participants,
the pain-free response with sTMS was highest for the
subgroup who used migraine preventive drugs.
Furthermore, we showed that sTMS was non-inferior to
sham stimulation for the associated symptoms of
photophobia, phonophobia, and nausea at 2 h, after
adjustment for the signifi cant covariates of prophylaxis
use and baseline evidence of the symptom. Our fi ndings
are relevant in view of the disabling nature of migraine.
For patients who commonly have aura as a signal of an
impending migraine, treatment with sTMS may abort
progression of the attack and abate disabling pain and
We assessed the size of the treatment eff ect as a
function of baseline pain at the time of treatment. The
eff ect was largest for patients with moderate pain at the
time of treatment compared with those who had no or
mild pain. Pain-free response rates for sham stimulation
diminished as baseline pain increased. However, the
number of participants who reported severe pain before
treatment was too small (sham n=6; sTMS n=4) to draw
fi rm conclusions.
Use of concomitant preventive drugs could interfere
with the eff ectiveness of sTMS treatment in some people
with migraine. We noted a reduction in relative risk of
headache pain at 2 h in individuals who used preventive
drugs and were allocated sTMS treatment. Ayata and
colleagues14 proposed that anticonvulsants and β blockers,
and potentially other migraine preventive treatments,
inhibit cortical spreading depression by reduction of
cortical hyperexcitability in animal models of migraine.
Patients with migraine are thought to have a state of
characterised by a lower threshold for initiation of
cortical spreading depression.7,15–17 Cortical spreading
depression might activate headache directly by
depolarisation of meningeal nociceptors7 or indirectly by
activation of descending pain facilitation pathways.18
Additionally, data suggest
dysrhythmia might contribute to a state of cortical hyper-
responsiveness.19 Alternatively, migraine could be related
to an impairment of inhibitory pathways, which can be
altered with sTMS treatment.10,13
Findings of studies suggest that cortical spreading
depression might also arise in migraine without aura;
patterns of hypoperfusion, similar to those seen in
migraine with aura, occur in migraine without aura.20
The eff ectiveness of sTMS in individuals who have
migraine without aura needs additional study.
ShamsTMS Diff erence* Upper one-sided
Moderate or severe pain at baseline
Symptoms at 2 h
Symptoms at 2 h
Symptoms at 2 h
No or mild pain at baseline
Symptoms at 2 h
Symptoms at 2 h
Symptoms at 2 h
sTMS=single-pulse transcranial magnetic stimulation. *Diff erence between the proportion of patients with symptoms
at 2 h in each group. †Meets criteria for non-inferiority (<5%) and superiority (<0%).
Table 3: Eff ect of baseline pain on development of non-headache migraine symptoms 2 h after
treatment (per-protocol analysis set)
Sham (n=82) sTMS (n=82)p
Achieved no or mild pain 2 h
Use of rescue drugs
At 0–2 h
At 0–48 h
Consistency of pain relief
Global assessment of relief
Change from screening
Total disability time (min)
55 (67%)59 (72%) 0·4988
Data are number of patients (%) or mean (SD). sTMS=single-pulse transcranial
magnetic stimulation. MIDAS=migraine disability assessment score.
Table 4: Secondary outcomes
www.thelancet.com/neurology Vol 9 April 2010 379
To date, only a few studies have been done to assess
specifi cally the eff ects of TMS in migraine, with table-top
devices and in-clinic treatment.12 Upton and colleagues13
undertook a small pilot study with a Cadwell stimulator
(model MES-10) to provide TMS pulses to 42 patients
with migraine for treatment of 75 attacks. Pain was
measured on the fi ve-point Likert-type scale. TMS was
associated with a mean decrease in pain of 70% and was
well tolerated. In a second study,21 42 patients with migraine
were treated with sTMS (n=23) or placebo (n=19). About
69% of the participants had relief from headache within
2 h of active treatment compared with 48% in the placebo
group. Treatment was well tolerated.
Although promising, these studies have several
limitations. Previous trials used table-top devices that are
too large and expensive for use outside a clinic; yet, most
migraine attacks are treated outside the medical setting.
Furthermore, previous studies were modest in size, some
were uncontrolled, and, in the controlled trials, blinding
was not shown. To address these limitations, we
developed a handheld device suitable for home use and a
sham device identical in appearance (see Methods). We
undertook a pilot study to assess adequacy of masking of
the sham device to ensure blinding of treatment.22 We
enrolled 60 patients and showed that 50% of those treated
with sTMS and 60% of those administered the sham
device believed they had received genuine sTMS. Results
from this study suggest that the treatment eff ect of TMS
is not due to unblinding of the device after administration
The current study has several potential limitations.
First, in a device trial, unblinding is a potential concern.
In this trial, however, patients who received sTMS and
sham stimulation were equally likely to guess that they
had received active stimulation. This fi nding argues
against the hypothesis that treatment diff erences can
be attributed to unblinding. Second, we have not
explored fully the range of sTMS doses or the optimum
timing of treatment. We studied a pair of magnetic
pulses. Additional work is needed to defi ne the
dose–response curve and the best dose. Third, although
participants were asked to treat during the aura phase,
the time interval varied from onset of aura to treatment,
and pain intensity at the time of treatment diff ered also.
Fourth, although TMS inhibits cortical spreading
depression in animal models, we are not certain that
these mechanisms account for treatment eff ects.
Finally, although the eff ectiveness and safety of
treatment for migraine with aura are established,
optimum patterns of use and cost eff ectiveness remain
to be established.
One of the clinical benefi ts of sTMS is its non-invasive
nature. Treatment can be delivered to a circumscribed
region of the brain, by contrast with drugs that are
delivered systemically. Further, in this study and in
previous reports, sTMS was not associated with
signifi cant adverse events.
Our fi ndings show that in some patients, sTMS is
eff ective and well tolerated for acute treatment of
migraine with aura. Although the exact mechanisms of
migraine remain under study, administration of sTMS in
people with migraine with aura decreases progression of
the attack in some individuals.
RBL participated in study design and implementation, data analysis,
had full access to data, and helped to write the report. DWD and PJG
participated in study design and data analysis, had full access to data,
and helped to write the report. SDS participated in study design, data
analysis, had full access to data, and helped to write the report. JRS
was a clinical investigator, reviewed the data, and helped to write the
report. SKA participated in study design and, as a study investigator,
had access to data, and helped to write the report. SHP participated in
the complete review and assessment of the clinical trial protocol and
statistical results, had full access to data, and helped to write the
report. REF had the idea for use of sTMS to treat migraine and
designed the sTMS device used in the study. PLR participated in study
design, data collection, data analysis, data interpretation, had full
access to data, and helped to write the report. Clinical data
management and statistical analysis were contracted to Innovative
Analytics (Kalamazoo, MI, USA). Clinical data review and clinical
report writing were contracted to Experien Group (Sunnyvale,
Richard B Lipton (Albert Einstein College of Medicine, Bronx, NY, USA);
David W Dodick (Mayo Clinic Arizona, Phoenix, AZ, USA);
Peter J Goadsby (UCSF, San Francisco, CA, USA); Joel R Saper
(Michigan Head Pain and Neurological Institute, Ann Arbor, MI, USA);
Stephen D Silberstein (Jeff erson Headache Center, Philadelphia, PA,
USA); Sheena K Aurora (Swedish Headache Center, Seattle, WA, USA);
Jerome Goldstein (San Francisco Clinical Research Center,
San Francisco, CA, USA); Yousef M Mohammad (Ohio State University
Medical Center, Columbus, OH, USA); Patricia L Ruppel (Innovative
Analytics, Kalamazoo, MI, USA); Robert E Fischell (Neuralieve,
Sunnyvale, CA, USA); Gary E Ruoff (Westside Medical Center,
Kalamazoo, MI, USA); Brian Grosberg (Montefi ore Headache Center,
Bronx, NY, USA); George R Nissan (Diamond Headache Clinic, Chicago,
IL, USA); Roger K Cady (Clinvest, Springfi eld, MO, USA);
William B Young (Jeff erson Headache Center, Philadelphia, PA, USA);
Todd D Rozen (Michigan Head Pain and Neurological Institute,
Ann Arbor, MI, USA); Benjamin M Frishberg (Neurology Institute,
Oceanside, CA, USA); Troy G Anderson and Eric Eross (Dedicated
Clinical Research, Sun City, AZ, USA); Jack A Klapper (Mile High
Research Center, Denver, CO, USA); John R Kirchner (Kirchner
Headache Clinic, Omaha, NE, USA); James W Banks (Mercy Health
Research, St Louis, MO, USA); Jan L Brandes (Nashville Neuroscience
Group, Nashville, TN, USA); and David W Alway (Innovative Clinical
Research Center, Alexandria, VA, USA).
Patients reporting at least one adverse event
Total number of adverse events reported
sTMS=single-pulse transcranial magnetic stimulation. Data are number of
patients (%). Adverse events listed individually had a frequency (in the total
randomised population) of 2% or more.
Table 5: Adverse events (safety analysis set)
Articles Download full-text
www.thelancet.com/neurology Vol 9 April 2010
Confl icts of interest
RBL has received a clinical research grant from and holds stock options
in Neuralieve; and has consulted for or undertaken research funded by
other manufacturers of drugs and devices for migraine. DWD is an
advisory board consultant to Neuralieve, Coherex, MAP, Merck, Minster,
Ortho-McNeil, Endo, Allergan, and Pfi zer; and is currently undertaking
research supported by Advanced Neurostimulation Systems and
Medtronic. SDS has received a clinical research grant from and holds
stock options in Neuralieve. JRS is a consultant, advisory board
member, or both for Allergan, Ortho-McNeil, Medtronic, Neuralieve,
ANS, Pozen, and GlaxoSmithKline; has received research grants from
GlaxoSmithKline, Merck, Allergan, Cypress, Eli Lilly, ANS, MAP
Pharma, Schwartx, Capnia, Depomed, XLT, Schwartz, and NuPathe; and
receives honoraria from GlaxoSmithKline, Merck, Ortho-McNeil, and
Purdue Pharma. SKA received grant and research support from
Advanced Bionics, Alexza, Allergan, GlaxoSmithKline, MAP
Pharmaceuticals, Merck, Ortho-McNeil, Neuralieve, and Takeda; has
served as a consultant for Ortho-McNeil Pharmaceutical, Merck,
GlaxoSmithKline, Allergan, and Neuralieve; and has received honoraria
from Merck, GlaxoSmithKline, NuPathe, and Ortho-McNeil
Pharmaceutical. SHP served as a paid consultant in the interpretation
of study results and publication of this clinical trial. REF is chief
technology offi cer and board member of Neuralieve, co-invented the
sTMS device used in this study, owns stock in the company, and has
received money for travel expenses; his three sons each own stock in
Neuralieve. PLR was contracted as a paid statistician to provide
consultation and services for the design, operations, analysis, and
reporting of results for this clinical trial. PJG is a consultant to and has
received basic TMS research funding from Neuralieve; and has worked
with or advised other devices companies, including ATI, Boston
Scientifi c, and Medtronic.
We thank Richard P Chiacchierini (RPC & Associates, Rockville, MD,
USA) and Carol T Schembri (Neuralieve, Sunnyvale, CA, USA) for their
1 Lipton RB, Bigal ME, Diamond M, Freitag F, Reed ML, Stewart WF.
Migraine prevalence, disease burden, and the need for preventative
therapy. Neurology 2007; 68: 343–49.
2 Stovner LJ, Hagen K, Jensen R, et al. The global burden of
headache: a documentation of headache prevalence and disability
worldwide. Cephalalgia 2007; 27: 193–210.
3 Headache Classifi cation Subcommittee of the International
Headache Society. The International Classifi cation of Headache
Disorders: 2nd edition. Cephalalgia 2004; 24 (suppl 1): 9–160.
4 Kirchmann M. Migraine with aura: new understanding from
clinical epidemiologic studies. Curr Opin Neurol 2006; 19: 286–93.
5 Leao AAP. Spreading depression of activity in cerebral cortex.
J Neurophysiol 1944: 7: 359–90.
6 Eikermann-Haerter K, Mosko witz M. Pathophysiology of aura.
In: Silberstein SD, Lipton RB, Dodick DW, eds. Wolff ’s headache.
Oxford: Oxford University Press, 2008: 121–31.
7 Bolay H, Reuter U, Dunn AK, Huang Z, Boas DA, Moskowitz MA.
Intrinsic brain activity triggers trigeminal meningeal aff erents in
a migraine model. Nat Med 2002; 8: 136–42.
Holland PR, Schembri CT, Fredrick JP, Goadsby PJ. Transcranial
magnetic stimulation for the treatment of migraine aura. Neurology
2009; 72 (suppl 3): A250 (abstr IN4-1.005).
Chen R, Cros D, Curra A, et al. The clinical diagnostic utility of
transcranial magnetic stimulation: report of an IFCN committee.
Clin Neurophysiol 2008; 119: 504–32.
10 Wasserman EM. Risk and safety of repetitive transcranial magnetic
stimulation: report and suggested guidelines from the International
Workshop on the Safety of Repetitive Transcranial Magnetic
Stimulation, June 5–7, 1996. Electroencephalogr Clin Neurophysiol
1998; 108: 1–16.
11 Moskowitz MA, Bolay H, Dalkara T. Deciphering migraine
mechanisms: clues from familial hemiplegic migraine genotypes.
Ann Neurol 2004; 55: 276–80.
12 Mohammad TM, Hughes G, Nkrumah M, et al. Self-administered
transcranial magnetic stimulation (TMS) during the aura phase
improved and aborts headache. Headache 2006; 46: 857.
13 Upton A, Clarke B, Castellanos C. Transcranial magnetic
stimulation (TMS) for migraine: clinical eff ect. Can J Neurol Sci
2005; 32 (suppl 1): S71.
Ayata C, Jin H, Kudo C, Dalkara T, Moskowitz MA. Suppression of
cortical spreading depression in migraine prophylaxis. Ann Neurol
2006; 59: 652–61.
15 Haerter K, Ayata C, Moskowitz MA. Cortical spreading depression:
a model for understanding migraine biology and future drug
targets. Headache Curr 2005; 2: 97–103.
16 Dalkara T, Zervas NT, Moskowitz MA. From spreading depression
to the trigeminovascular system. Neurol Sci 2006;
27 (suppl 2): S86–90.
17 Levy D, Straussman A, Burstein, R. A critical view on the role of
migraine triggers in the genesis of migraine pain. Headache 2009;
18 Lambert GA, Hoskin KL, Zagami AS. Cortico-NRM infl uences on
trigeminal neuronal sensation. Cephalalgia 2008; 28: 640–52.
19 Coppola G, Vandenheede M, Di Clemente L, et al. Somatosensory
evoked high-frequency oscillations refl ecting thalamo-cortical
activity are decreased in migraine patients between attacks.
Brain 2005; 128: 98–103.
20 Barkley GL, Tepley N, Nagel-Leiby S, Moran JE, Simkins RT,
Welch KMA. Magnetoencephalographic studies of migraine.
Headache 1990; 30: 428–34.
21 Clarke BM, Upton ARM, Kamath MV, Al-Haarbi T, Castellanos CM.
Transcranial magnetic stimulation for migraine: clinical eff ects.
J Headache Pain 2006; 7: 341–46.
22 Ruppel PL. Statistical report: Neuralieve protocol NL-2006-002—
eff ectiveness of masking of a sham device compared to sTMS in
asymptomatic migraineurs with and without a history of visual aura
(Aug 7, 2006). Kalamazoo: Innovative Analytics, 2006.