Neurophysiologie Clinique/Clinical Neurophysiology 51 (2021) 291—302

Disponible en ligne sur

ScienceDirect
www.sciencedirect.com

ORIGINAL ARTICLE

Posterior-superior insular deep transcranial

magnetic stimulation alleviates peripheral
neuropathic pain —– A pilot double-blind,
randomized cross-over study
Liu Dongyang a, Ana Mércia Fernandes a,
Pedro Henrique Martins da Cunha a, Raissa Tibes a, João Sato a,
Clarice Listik a, Camila Dale a, Gabriel Taricani Kubota a,
Ricardo Galhardoni a, Manoel Jacobsen Teixeira a,
Valquíria Aparecida da Silva a, Jefferson Rosi a,
Daniel Ciampi de Andrade a,b,∗

a
LIM-62, Pain Center, Department of Neurology, University of São Paulo, Av. Dr. Enéas de Carvalho Aguiar,
255, 5th Floor, P.O. Box: 05403-900, São Paulo, SP, Brazil
b
Pain Center Instituto do Câncer Octavio Frias de Oliveira, University of São Paulo, Avenida Dr. Arnaldo
251, P.O. Box: 01246-000, São Paulo, SP, Brazil

Received 27 April 2021; accepted 9 June 2021

Available online 24 June 2021

KEYWORDS Abstract
Insula; Objectives. — Peripheral neuropathic pain (pNeP) is prevalent, and current treatments, includ-
Neuropathic pain; ing drugs and motor cortex repetitive transcranial magnetic stimulation (rTMS) leave a
Neuronavigation; substantial proportion of patients with suboptimal pain relief.
Peripheral Methods. — We explored the intensity and short-term duration of the analgesic effects produced
neuropathy; in pNeP patients by 5 days of neuronavigated deep rTMS targeting the posterior superior insula
Transcranial magnetic (PSI) with a double-cone coil in a sham-controlled randomized cross-over trial.
stimulation Results. — Thirty-one pNeP patients received induction series of five active or sham consecutive
sessions of daily deep-rTMS to the PSI in a randomized sequence, with a washout period of at
least 21 days between series. The primary outcome [number of responders (>50% pain intensity
reduction from baseline in a numerical rating scale ranging from 0 to 10)] was significantly higher

∗ Corresponding author at: Divisão de Clínica Neurocirúrgica do Hospital das Clínicas da FMUSP, Instituto Central, Av. Dr. Enéas de Carvalho

Aguiar, 255, 5◦ Andar, Sala 5084, Cerqueira César, 05403-900, São Paulo, Brazil.
E-mail address: ciampi@usp.br (D. Ciampi de Andrade).

https://doi.org/10.1016/j.neucli.2021.06.003
0987-7053/© 2021 Elsevier Masson SAS. All rights reserved.
 L. Dongyang, A.M. Fernandes, P.H.M. da Cunha et al.

after real (58.1%) compared to sham (19.4%) stimulation (p = 0.002). The number needed to treat
was 2.6, and the effect size was 0.97 [95% CI (0.6; 1.3)]. One week after the 5th stimulation day,
pain scores were no longer different between groups, and no difference in neuropathic pain
characteristics and interference with daily living were present. No major side effects occurred,
and milder adverse events (i.e., short-lived headaches after stimulation) were reported in both
groups. Blinding was effective, and analgesic effects were not affected by sequence of the
stimulation series (active-first or sham-first), age, sex or pain duration of participants.
Discussion. — PSI deep-rTMS was safe in refractory pNeP and was able to provide significant
pain intensity reduction after a five-day induction series of treatments. Post-hoc assessment of
neuronavigation targeting confirmed deep-rTMS was delivered within the boundaries of the PSI
in all participants.
Conclusion. — PSI deep-rTMS provided significant pain relief during 5-day induction sessions
compared to sham stimulation.
© 2021 Elsevier Masson SAS. All rights reserved.

Introduction to antinociceptive responses [31]. In experimental models

Neuropathic pain affects up to 10% of the general popula-
tion, and despite recent advances in its control, up to 40%
of patients remain symptomatic [7,44]. In the last fifteen
years, noninvasive brain stimulation (NIBS) has emerged
as a therapeutic option for NeP control [5]. Most studies
have focused on repetitive transcranial magnetic stimula-
tion (rTMS) aimed at the precentral gyrus. Currently, rTMS
over the primary motor cortex (M1) is considered a thera-
peutic option for NeP and ranks in societal recommendations
worldwide [28,34]. It is known from animal and human
studies that M1 stimulation is able to induce changes in
brain areas distant to the stimulation site, and its analgesic
effects depend on opioidergic [15] and glutamatergic effects
[11,35].
Also, rTMS to M1 is able to restore defective GABA-A
dependent intracortical inhibition in NeP patients, which is
associated with pain intensity improvement [30]. Despite
these effects, it has been reported in experimental [43] and
clinical studies [35] that not all individuals respond to M1
rTMS, which fails to improve pain in up to 50% of patients.
This has led to the quest for new NIBS approaches for pain
relief. New stimulation patterns have been tried, such as
theta-burst stimulation [16,29], and another line of research
has focused on the quest for new stimulation targets.
The idea behind the quest for new targets for NIBS in
chronic pain relies on the idea that since M1 stimulation
influences several distant areas within the neuroaxis [37],
one could obtain more focused or intense pain relief by
stimulating these secondary brain areas directly, bypass-
ing the precentral gyrus. This might provide pain relief
for patients not responding to classical M1 NIBS since the
analgesic effects of these extra-motor areas would rely on
different mechanisms of action [1,19]. One of these areas
is the posterior insula. The posterior granular part of the
insula is one of the main recipients of somatosensory affer-
ents reaching the cortex; it is implicated in the central
integration of the sensory-discriminative aspect of pain. It
is consistently active in different neuroimaging studies of
acute and chronic pain [38,39].
Intraoperative stimulation of the human insula in patients
under stereo-EEG [18] was reported to induce antinocicep-
tive effects, and its inhibition in experimental studies led
of peripheral neuropathic pain in rats, deep brain stimula-
tion of the posterior insula led to antinociceptive effects
dependent on endogenous opioids and cannabinoids [1,19].
Recently a method to stimulate the insula non-invasively by
deep rTMS using neuronavigation has been reported [10,24]
and tested by other groups [31]. Here we performed a
pilot exploratory double-blind cross-over trial assessing the
safety and the magnitude and temporal profile of the poten-
tial analgesic effects of five consecutive days of d-rTMS to
the PSI compared to sham stimulation in pharmacologically-
refractory peripheral neuropathic pain.
Methods
Patients
Our Ethics Review Board approved this study
(#28659714.1.0000.0068) and registered it on clinical-
trials.gov (NCT04279548). The present study is the first
step of a two-step larger trial: in step one (present study),
patients received noninvasive PSI and sham in a cross-over
design. In step 2, ten responders were invited to enroll in
a double-blind randomized study of PSI deep brain stim-
ulation (i.e., PSI targets used in step 1 d-rTMS study will
be used as a target for implanted electrodes for long-term
treatment of neuropathic pain by deep brain stimulation).
All patients provided informed consent to participate in
the study. Patients were recruited from outpatient clinics
geographically near the outpatient pain clinics of the
Hospital das Clínicas, University of São Paulo, between
March 2018 and March 2020. Inclusion criteria were: age
over 18; presence of defined chronic peripheral neuropathic
pain according to current guidelines [23]; presence of
neuropathic pain refractory to drug therapies [12] persis-
tent for longer than six months, with an average intensity
score above 40/100 mm on a visual analogue scale (VAS).
Exclusion criteria were contraindication for rTMS [i.e., past
head trauma, current epilepsy, intracranial ferromagnetic
components, pacemaker, implanted microprocessors (i.e.,
cochlear implants), pregnancy], known major psychiatric
disorders (as assessed by the DSM-V), history of substance
abuse or work litigation issues. At the inclusion visit, we col-

292
 Neurophysiologie Clinique/Clinical Neurophysiology 51 (2021) 291—302
lected the patient’s demographic and clinical information:
age, gender, duration of pain, lesion and painful region,
current medications, and performed physical examination
to confirm neuropathy and the presence of peripheral NeP.
Experimental design
This was an exploratory randomized, double-blind, sham-
controlled, pilot cross-over trial to investigate the analgesic
effects of active versus sham deep (d) repetitive (r) tran-
scranial magnetic stimulation (TMS) of the posterior superior
insula (PSI) in patients with peripheral neuropathic pain.
All participants received one sham and one active series
of d-rTMS. Participants were randomly allocated (using
www.randomizer.org) in blocks of four into one of two
possible sequences of series of stimulation: one sequence
commencing by active-PSI-rTMS, or another sequence start-
ing by sham-PSI-rTMS series. In both sequences, participants
were switched to the respective comparator series after
the first one was completed and a wash-out period was
observed. Each series comprised stimulations for five con-
secutive days. Clinical and pain assessments were performed
at baseline (D0), daily during treatment (D1−5) (from 15 to
45 min after the end of stimulation), and one week after the
last stimulation session (D12, i.e., one week after the 5th
day of stimulation) session in each series. A washout period
of at least 21 days was performed between the active and
sham series. Pain intensity had to return to >80% pain inten-
sity seen at baseline for the second series of stimulation to
initiate.
Determination of the rTMS target
Patients underwent magnetic resonance imaging, which is
used to define the stimulation target using the neuronavi-
gation technique. Neuronavigation was used exclusively to
determine the stimulation target, but not to monitor the coil
position during stimulation procedures. A 3D image of the
head was obtained using volumetric T1-weighted magnetic
resonance imaging (MRI) for frameless stereotaxic neuron-
avigation (BrainsightTM, Rogue Research). We located the
insula and selected the upper-posterior sub-quadrant as the
rTMS target, using the technique previously described [10]
which included the identification of the insular limen after
‘‘peeling’’ the lateral series of 3D image until the insular
cortex is detected. Then, a parallel line is drawn, parallel
to the coronal plane. Its midpoint is used to draw a sec-
ond line, this time orthogonal to the first one, running from
anterior to posterior. This procedure divided the insula into
4 quadrants. The upper-most and posterior-most quadrant
was them identified, and this quadrant was further divided
into four new sub-quadrants based on two new orthogonal
lines departing from the midpoints of the posterior and ante-
rior lines delineating this quadrant. This provided four new
(sub)quadrants withing the upper-posterior original quad-
rant. The mid-point within the upper-posterior sub-quadrant
is the target marked on the MRI. Then the orthogonal pro-
jection from this quadrant to the scalp was identified with
the neuronavigation software. Using the system pointer with
infrared reflectors, the orthogonal scalp projection was
identified as the d-rTMS target and its location was marked
293
over the skin of the scalp with a skin-marking pen, so that
the point could be identified in subsequent sessions.
Anatomical confirmation of the insular target
Similar to others [31], we chose to perform a post-hoc
reconstruction analysis to ensure that the visually-guided
identification of the targets were actually confined within
the boundaries of the PSI. These analyses were indepen-
dent of the TMS procedure; they were performed after
the trial ended and were not used to monitor potential
dislocations of the coil during the TMS procedure. Thus,
an off-line assessment of the location of each patient’s
stimulation target was conducted. Each participant’s insu-
lar target/region of interest (ROI) used for PSI d-rTMS was
traced into the MNI152 Space [21] for each axis: hori-
zontal (x), anteroposterior (Y) and superior-inferior (Z).
This was made by locating the position of the anterior
commissure — posterior commissure (ACPC) line using the
BrainsightTM software (Rogue Research) and identifying the
respective MNI space coordinates. These coordinates were
then compared with the probability maps of the six insu-
lar subdivisions in the same space (the three short gyri, the
anterior inferior cortex, and the two long gyri) [22]. After
this comparison, it was possible to show, in a normalized
brain space (i.e., MNI152), that the PSI d-rTMS was tar-
geted close to the most posterior and superior aspect of
the anterior insular long gyrus (ALG), the place previously
reported by our team [10] (Fig. 2). Finally, each patient’s ROI
was marked and placed into a 3D Brain Image model and
exported as NIfTI files into Mango brain visualization soft-
ware (http://ric.uthscsa.edu/mango/). These points were
marked into an anatomical T1 brain template (Colin27 T1
seg MNI.nii available from http://www.brainmap.org/ale/).
Transcranial magnetic stimulation — rest motor
threshold determination
A double-cone coil (cooled D-B80 coil, MagVenture, Farum,
Denmark) connected to a MagProX100 machine (MagVenture)
was used to measure rest motor threshold (RMT) with the
main phase of the biphasic induced electric current deliv-
ered from the posterior to the anterior (PA) direction. RMT
was determined before each series, with patients positioned
on a reclining chair, relaxed, in a sound-attenuated room,
using single pulse stimulation over the primary motor cor-
tex representation of the tibialis anterior muscle on the
same body side as the patient’s pain using surface electrodes
(Natus, Middleton, WI, USA). RMT were considered as the
lowest pulse intensity required to evoke a motor evoked
response of at least 50 ␮V in size in 5 out 10 successive
pulses. RMT were recorded as the maximal stimulator output
(%).
Transcranial magnetic stimulation — d-rTMS
procedure
The double-cone coil (cooled D-B80 coil, Magventure) was
placed on the scalp and centered over the scalp target for
stimulation. During all stimulation sessions a second coil
 L. Dongyang, A.M. Fernandes, P.H.M. da Cunha et al.

(figure-of-eight B-65 coil, MagVenture) was placed orthog-

onally to the double-cone coil. This second coil was used for
stimulation in cases of sham stimulations, while the double-
cone coil remained turned off. For active stimulations, the
stimulation was delivered by the double-cone coil, while the
figure-of-eight coil was left in place, turned off. This pro-
cedure has been used in several studies and has provided
proper blinding even in cross-over designs (Fig. 1A and B)
[2,16]. Both coils were fixed on two independent adjustable
mechanical arms, and their positions were monitored sys-
tematically during the rTMS session by visually inspecting the
location relative to the target. During sessions the handle of
the double-cone coil was pointed backwards, so that the first
phase of the biphasic electromagnetic field induced electric
current was in the anterior-posterior direction and the elec-
tric current from its main phase was the posterior-anterior
direction (Fig. 1A and B) [2,3,24]. The PSI contralateral to
the pain side was used. In cases of bilateral pain, the PSI
contralateral to the most painful side was stimulated. The
PSI-rTMS parameters used were similar to those showing
antinociceptive effects in patients and healthy volunteers
[23,30]: 3000 pulses (30 trains of 10 s each, inter-train inter-
val of 20 s) delivered at a frequency of 10 Hz and an intensity
set at 80% of the resting leg motor threshold using a Mag-
ProX100 machine (MagVenture).

Pain and related assessments

Clinical and pain assessments were performed at baseline
(D0), daily during treatment, and one week after the last
stimulation (D12) in each series (active and sham).
The primary endpoint of the study was the number of
patients reaching significant pain relief (>50% pain intensity
reduction) assessed by visual analogue scale (VAS) ranging
from 0: no pain to 100 mm: maximal pain imaginable one
week after the 5th stimulation session compared to the
baseline assessment.
Figure 1 Stimulation montage.
Secondary outcomes were: (a) neuropathic pain symp-
tom inventory (NPSI). This questionnaire quantifies the mean
intensity of 10 neuropathic symptoms and their combina-
tion (dimensions), assessing five different dimensions over
the last 24 h on an 11-point (0−10) numerical rating scale. It
also generates a total score (sum of each of the 10 individual
scores, scored out of 100) [6,14]; (b) the seven items of pain

Figure 2 Anatomical confirmation of the insular target.

Location of posterior superior insula deep repetitive transcranial magnetic stimulation targets within the left (A) and right (B)
insulae according to the coordinates as described by Faillenot et al. [22]. See Table 2.

294
 Neurophysiologie Clinique/Clinical Neurophysiology 51 (2021) 291—302
interference of the brief pain inventory (BPI), rated from 0
(does not interfere), to 10 (complete interference), mea-
suring the impact of pain on general activity, mood, walking
ability, normal work, relations with other people, sleep and
enjoyment of life.
Adverse events report
During each treatment session, a side effect assessment was
conducted to monitor any study-related complications using
a dedicated questionnaire. Patients were asked to report
any potential neurological side-effects related to the treat-
ment, such as headaches, nausea, sleepiness, and local pain.
Blinding assessment
Researchers who delivered d-rTMS were blinded to all other
assessments except the type of TMS being applied and
had no other role in the study. Care was taken not to set
patients’ appointments simultaneously so that waiting-room
conversations could be avoided and to mitigate possible
jeopardizing of blinding integrity. The blinding assessment
was performed at the end of the study (i.e., end of the
trial, on D12 from the last series of stimulation) as previ-
ously reported by asking all patients [42] (1) if they were
able to tell which sequence of treatment they were allo-
cated to (i.e., yes or no); (2) which sequence they think
they actually received (i.e., active-first or sham-first); (3)
if they would like to maintain the sessions of d-rTMS for a
longer period, should this option be offered to them (i.e.,
yes or no).
Statistical analyses
Statistical analyses were conducted with SPSS version 22
(SPSS Inc, Chicago, IL). Results were expressed in aver-
ages and standard deviation. The Kolmogorov-Smirnoff test
accessed normality of the data. In all cases, P values <0.05
were considered significant. The Cohen’s d, defined as the
difference between the means of the 2 groups divided by the
pooled standard error, was used to calculate effect sizes.
To evaluate intervention efficacy, the percentage of respon-
ders (>50% pain intensity reduction from baseline) measured
on the last day of stimulation were calculated, and a
comparison between groups was performed by using the
McNemar test. Due to the pilot nature of the trial, we per-
formed a modified intention to treat (mITT) analysis where
only patients undergoing at patients who received at least
three stimulation sessions in both treatment sequences were
included. Since the Kolmogorov—Smirnov test revealed that
secondary outcomes such as BPI and NPSI scores did not have
a normal distribution, the differences between groups were
compared using a non-parametric test (Wilcoxon’s test). The
sample size was calculated based on the effect size achieved
by a previous d-rTMS trial, considering a repeated-measures
ANOVA approach and using the software G*Power 3.1.9.2
for Windows (California, USA). Since this is a pilot study
using PSI d-rTMS for peripheral neuropathic pain for the
first time, we calculated the expected sample size based
on a previous study using deep rTMS for the same type of
295
pain [36]. Thus, based on this study’s effect size (0.8), a
two-tailed ˛ error level probability of 0.05, and a power
of 0.99, the estimated sample size needed would be 13
subjects per treatment sequence, thus providing 26 par-
ticipants. We have added six more participants beginning
with each of the two treatment sequences to account for
potential dropouts. The effects of the sequence of the series
of stimulation (active-first or sham-first) were explored for
the active and sham stimulations comparing data according
to sex, and side of stimulation (Wilcoxon test) for active
and sham series. The correlation between pain relief (pain
relief at D5 compared to D) and pain duration in years was
compared with the Spearman’s test. Also, the effectiveness
of the washout period was assessed by comparing base-
line pain intensity before each sequence of stimulation in
the active and sham series of stimulation (Wilcoxon’s test).
Blinding assessment was composed of three questions cited
above. Responders were classified as being ‘‘real respon-
ders’’ response to active but not to sham stimulations)
or not, and according to the sequence of treatment they
received (active-first, or sham-first). Proportions were com-
pared by the Chi2 test. Fisher’s exact test was used when
n < 5. For all these instances significance was also set at
p < 0.05.
Results
Sample description
Forty-six patients were screened with peripheral neuro-
pathic pain for this trial, and 38 were randomized (Fig. 3).
Nineteen patients started with active treatment and 19 with
sham. Table 1 shows baseline demographic and clinical char-
acteristics of the 31 patients that received the allocated
interventions and were analyzed.
Anatomical confirmation of the insular target
Fig. 2 (A and B) shows the neuronavigated PSI location’s
target location in respect to the MNI coordinates in both
hemispheres. Stimulations were performed on the scalp
orthogonal projection of the insular target. Table 2 (and
Supplementary Table S1) showed that all targets were con-
fined within the boundaries of the anterior long gyrus of the
insula, as defined by Faillenot et al., 2017 [22].
Pain assessment — primary outcome
Thirty-one patients were analyzed in a modified intention-
to-treat approach, including patients who received three
stimulation sessions in both treatment sequences. Eighteen
(58.1%) were responders (>50% pain intensity reduction) to
real-PSI-dTMS and six (19.4%) responded to sham PSI-dTMS
(p = 0.002) (Table 3). The number needed to treat was 2.6,
and the effect size was 0.97 [95% confidence interval (0.6;
1.3)]. Fourteen patients (45.1%) were real responders, hav-
ing >50% pain intensity reduction after real PSI-dTMS and no
response to sham PSI-dTMS. The sequence of the stimulation
series was not significantly different between responders
and non-responders for both the >50% and >30% pain inten-
 L. Dongyang, A.M. Fernandes, P.H.M. da Cunha et al.

Figure 3 Study flow-chart.

80 Pain assessment — secondary outcomes

P=0.002 active d-rTMS
70 sham PSI-dTMS
* Secondary outcomes measured on the 7th day after the end
VAS (0-100)

60 of stimulation (D12) showed that the pain intensity differ-

ence between groups was no longer significant. There were
50
no differences between groups concerning pain interference
40 in daily activities, DN-4 or NPSI scores (Table 5).
The average wash-out time between the two-stimulation
30 series was 28.6 days (CI95%: 23.8; 33.4) in the active-first
D0 D1 D2 D3 D4 D5 D12
group and 27.9 days (CI95%: 22.4; 33.6) in the sham-first
sequences of stimulation series (p > 0.2). Pain was unilat-
Figure 4 Pain intensity effects of PSI d-rTMS. eral in all instances and the PSI contralateral to the pain
PSI d-rTMS: posterior superior insula deep repetitive tran- side was targeted. Accordingly, the left insula was stimu-
scranial magnetic stimulation; whiskers over bars represent lated in 54.8% of patients and the right in 45.2%. Active
standard error *p < 0.05. d-rTMS provided pain relief (NRS measured at D5 sub-
tracted from baseline) of −39.2 ± 31.8 after left and of
sity reduction thresholds (p > 0.2). Pain intensity went from −23.0 ± 23.9 after right PSI stimulation (p = 0.08). After
68.07 ± 12.91 and 65.92 ± 11.41 at baseline to 33.96 ± 27.11 sham d-rTMS, right side stimulation provided −13.1 ± 22.6
and 51.45 ± 24.09 on 5th day of stimulation after the active pain relief, while left sham decreased pain by −16.7 ±
and the sham d-rTMS series, respectively (p = 0.002) (Fig. 4). 26.2; (p = 0.80). Pain intensity before active and sham
One week after stimulation (D12), differences between arms series did not differ according to the sham-first or active-
were no longer present (p > 0.2). Table 4 shows pain intensity first series of stimulation (Fig. 5). Concerning sex, active
reduction from D1 to D12 in both groups. PSI provided pain reduction {D5-D0 pain intensity, [aver-

296
 Neurophysiologie Clinique/Clinical Neurophysiology 51 (2021) 291—302

Table 1 Demographical profile and baseline characteristics of subjects included in the study.

Total (n = 31)

Age (years)a 46.28 ± 11.44 (20−67)

Sex, n (%) Female 4 (12.9)
Male 27 (87.1)
Etiology of neuropathic Root avulsion 22 (71.0%)
pain, n (%) Trigeminal neuralgia 3 (9.7%)
Traumatic nerve injury 1 (3.2%)
Postherpetic neuralgia 3 (9.7%)
Radiculopathy 2 (6.4%)
Pain side Left 14 (45.2%)
Right 17 (54.8%)
Duration of pain (years)a 10.22 ± 9.31 (1−38)
Current neuropathic pain Gabapentinoids 27 (87.1%)
medication, n (%) Antidepressants 21 (67.7%)
Opiates 24 (77.4%)
Dipyrone 14 (45.1%)
Neuroleptics 15 (48.4%)
Anticonvulsants 12 (38.7%)
Muscle relaxants 8 (25.8%)
BPIa Worst pain 8.65 ± 1.25 (6−10)
Least pain 4.75 ± 1.80 (0−9)
Average pain 6.81 ± 1.47 (3−10)
Now pain 6.13 ± 1.80 (2−9)
Interference with general activity 6.77 ± 2.53 (0−10)
Interference with mood 7.06 ± 3.1 (0−10)
Interference with walking 5.03 ± 3.7 (0−10)
Interference with work 6.94 ± 2.6 (0−10)
Interference with relations with others 6.19 ± 2.9 (0−10)
Interference with sleep 5.94 ± 3.4 (0−10)
Interference with enjoyment of life 7.81 ± 2.3 (1−10)
Pain intensity index 6.58 ± 1.28 (3.7−9.5)
Pain interference daily activity score 6.53 ± 2.24 (1−10)
NPSIa Burning 6.19 ± 3.7 (0−10)
Pressure 5.10 ± 4.3 (0−10)
Squeezing 4.65 ± 4.15 (0−10)
Electric shocks 4.61 ± 4.01 (0−10)
Stabbing 4.65 ± 4.15 (0−10)
Evoked by brushing 3 ± 4.05 (0−10)
Evoked by pressure 2.55 ± 3.67 (0−10)
Evoked by cold stimuli 3.48 ± 4.4 (0−10)
Pins and needles 4.39 ± 3.89 (0−10)
Tingling 4.71 ± 3.91 (0−10)
Total score 43.32 ± 22.51 (6−94)
DN4a Total score 7.03 ± 2.04 (2−10)
a Values are presented in: mean ± SD (minimum and maximum). BPI: brief pain inventory, NPSI: neuropathic pain inventory, DN-4:
Douleur neuropathique-4 questionnaire.

age ± standard deviation (minimum—maximum values)]} of
−29.66 ± 61.40 (−70 −41) in women, and −28.76 ± 24.28
(−76 −6) in men (p = 0.21), while sham d-rTMS brought
about pain reduction of 0.00 ± 13.22 (−15 −10) in women
and −13.04 ± 24.14 (−70 −28) in men (p = 0.24). Correlation
analyses between pain duration (in years) and percentage of
pain decrease after stimulation (% pain intensity decrease)
were non-significant: Spearman’s rho = −0.01, p = 0.92 for
active and Spearman’s rho = 0.14, p = 0.43 for the sham
series.
Dropouts and adverse events report
Seven patients dropped out of the study. One of them was on
the second series (drop-out due to a transportation issue),
and six were on the first series of treatment: four left due
to lack of analgesic effects (two after the 2nd active and
two after the 3rd sham daily sessions) and three due to
transportation issues. None of these patients had treatment-
related adverse events as the cause of drop-out. Concerning
adverse events, two patients had headaches after active and

297
 L. Dongyang, A.M. Fernandes, P.H.M. da Cunha et al.

Table 2 Comparison between the polled posterior superior insula target coordinates intervals and insular ALG interval coordi-
nates in MNI152 space according to Faillenot et al. [22].

PSI coordinates intervals MNI Insula ALG coordinates interval

Right Left

x +34.86 to +43.37 −45.8 to −32.27 +32 to +43

y −31.84 to −18.72 −29.83 to −13.27 0 (+ or −35)
z +11.68 to +19.82 +8.47 to +18.72 0 (+ or −30)
PSI: posterior superior insula; ALG: anterior long gyrus. Y: values were negative because the target was the most posterior aspect of
the ALG Z: values are positive because the targets were s in the most superior aspect of ALG.

Table 3 Percentage of responders to d-rTMS.

Active-PSI-rTMS Sham-PSI-rTMS p¥¥

50% improvement pain intensity¥

Positive responder, n (%) 18 (58.1) 6 (19.4) 0.002*
Negative responder, n (%) 13(41.9) 25 (80.6)
30% improvement pain intensity
Positive responder, n (%) 19 (61.3) 9 (29) 0.011*
Negative responder, n (%) 12 (38.7) 22 (71)
PSI: posterior superior insula. d-rTMS: deep repetitive transcranial magnetic stimulation.
Data based on the assessment at the 5th day of stimulation (main study outcome).
* Significance p < 0.05.
¥ Study’s main outcome.
¥¥ Chi2 test.

Table 4 Pain intensity reduction.

Active PSI-rTMS Sham PSI-rTMS p

D1 −17.87 ± 19.57 (−80 to 10) −10.70 ± 20.09 (−61 to 43) 0.05

D2 −20.41 ± 23.01 (−80 to 12) −12.66 ± 19.46 (−59 to 17) 0.21
D3 −19.64 ± 27.50 (−80 to 25) −13.50 ± 23.55 (−60 to 27) 0.65
D4 −16.67 ± 25.29 (−80 to 24) −11.37 ± 19.97 (−47 to 21) 0.35
D5 −28.03 ± 27.80 (−80 to 20) −10.54 ± 21.50 (−60 to 28) 0.002*
D12 −1.33 ± 24.31 (−80 to 53) −5.96 ± 24.31 (−80 to 53) 0.20
Pain intensity decrease from baseline [NRS (numerical rating scale; 0−10) at the different timepoints subtracted from NRS at baseline.
D: day.
* p < 0.05.

1 after sham stimulation, none persisting until the next stim-
ulation session, and one requiring analgesics (metamizole)
for symptomatic control. One patient had dizziness persist-
ing for two days after the active stimulation that did not
necessitate treatment interruption, and one had scalp ten-
derness after one of the active sessions. There were no other
major side effects, such as seizures [31]. No side effects
demanded treatment interruption, and none led to patient
drop out from the study.
Blinding assessment
After the study ended, 21 patients (67.7%) said they could
tell which sequence of treatment they received, com-
pared to 10 (32.3%) who did not (p = 0.04). Among these 21
patients, ten (47.6%) were real responders, and 11 (52.4%)
were not (p = 0.82). Twelve (57.1%) received the active-first
sequence of series of stimulation, and nine (42.9%) receive
the sham first-sequence (p = 0.51).
When directly asked to report which sequence, they
thought they actually received, eighteen (58.0%) guessed
it right, and 13 (42%) guessed it wrong (p = 0.36). Eleven
(61.1%) of those who guessed right the sequence of stimu-
lation received the active-first sequence, while 7 (38.9%)
received the sham-first sequence (p = 0.48). Among these
same 18 patients, 10 (55.6%) were real responders, and 8
(44.4%) were not (p = 0.81).
Nineteen patients (61.3%) said they would like to main-
tain the sessions of d-rTMS for a more extended period,
should this option be offered to them. Among these nine-
teen patients, nine (47.4%) received the active-first and

298
 Neurophysiologie Clinique/Clinical Neurophysiology 51 (2021) 291—302

Table 5 Pain assessment one week after the last stimulation.

Active-PSI-rTMS Sham-PSI-rTMS p

Brief pain inventory
Douleur neuropathique-4
Neuropathic pain symptom
inventory
Values are presented in: mean ± SD (minimum and maximum).
Worst pain 8.35 ± 1.4 (5−10) 7.77 ± 2.0 (1−10) 0.09
Least pain 5.23 ± 2.1 (0−10) 4.55 ± 1.9 (0−7) 0.11
Average pain 6.58 ± 1.7 (3−10) 6.06 ± 1.8 (1−9) 0.10
Now pain 6.23 ± 2.4 (0−10) 5.94 ± 2.0 (1−10) 0.32
General activity 6.74 ± 2.5 (0−10) 6.42 ± 3.0 (0−10) 0.70
Mood 6.74 ± 2.8 (0−10) 6.23 ± 3.3 (0−10) 0.56
Walking 5.10 ± 3.4 (0−10) 4.94 ± 4.0 (0−10) 0.92
Work 6.03 ± 3.0 (0−10) 6.19 ± 2.9 (0−10) 0.65
Relations with others 6.26 ± 3.2 (0−10) 5.26 ± 3.7 (0−10) 0.22
Sleep 5.42 ± 4.0 (0−10) 5.61 ± 3.5 (0−10) 0.90
Enjoyment of life 7.68 ± 2.4 (0−10) 6.65 ± 3.3 (0−10) 0.17
BPI pain intensity 6.59 ± 1.76 (2.7−10) 6.18 ± 1.7 (0.7−9) 0.07
BPI interference daily activity score 6.28 ± 2.47 (0−9.4) 5.89 ± 2.8 (0−10) 0.90
Total score 6.73 ± 1.59 (3−10) 6.25 ± 2.12 (1−10) 0.25
Burning 6.29 ± 3.3 (0−10) 5.87 ± 3.62 (0−10) 0.59
Pressure 5.32 ± 4.16 (0−10) 4.65 ± 4.02 (0−10) 0.19
Squeezing 4.97 ± 4.47 (0−10) 4 .00 ± 4.22 (0−10) 0.09
Electric shocks 4.06 ± 3.97 (0−10) 4.19 ± 3.70 (0−10) 0.93
Stabbing 3.58 ± 4.12 (0−10) 4.58 ± 4.09 (0−10) 0.15
Evoked by brushing 3.19 ± 3.95 (0−10) 3.06 ± 3.82 (0−10) 0.86
Evoked by pressure 2.94 ± 3.89 (0−10) 2.13 ± 3.55 (0−10) 0.18
Evoked by cold stimuli 2.52 ± 3.62 (0−10) 2.58 ± 3.70 (0−10) 0.95
Pins and needles 3.87 ± 4.04 (0−10) 3.77 ± 3.80 (0−10) 0.84
Tingling 4.71 ± 3.91 (0−10) 4.71 ± 3.91 (0−10) 0.37
Total score 4.05 ± 1.76 (1.2−7.7) 3.8 ± 2.26 (0−8.6) 0.34

pain refractory to drug therapy [34]. The effect size of the

Figure 5 Pain intensity at baseline according to the sequences
of series of stimulation.
PSI d-rTMS: posterior superior insula deep repetitive transcra-
nial magnetic stimulation; NRS: numerical rating scale; D: day,
whiskers over bars represent standard error *p < 0.05.
10 (52.6%) received the sham-first sequence of treatment
(p = 0.81); eleven (57.9%) were real responders, while 8
(42.1%) were not (p = 0.64).
Discussion
We have shown that PSI noninvasive stimulation has a signifi-
cant analgesic effect in patients with peripheral neuropathic
intervention reached 0.97, and the NNT was 2.6. The stimu-
lation provided significant pain relief in most patients (58.1%
with >50% pain intensity relief), 45.1% being ‘‘real’’ respon-
ders, who had significant relief after real, but not after sham
d-rTMS.
The posterior insula is a key part of the spino-
thalamo-cortical pathway, implicated in the integration of
nociception, thermoreception, and interoception. The pos-
terior insular area alone receives more than 40% of all the
thalamic fibers reaching the insula [25]. It is widely known
that the insula is implicated in both acute and chronic pain
[13,38]. Direct electrical stimulation of the posterior insula
may produce nociceptive sensations, while lesions to this
region reduce pain perception in large areas of the body
[8,26]. Insular glutamate levels correlate with an individ-
ual’s nociceptive threshold, with higher levels correlating
with lower thresholds [45]. Opioids cause depression of exci-
tatory propagation in the insular cortex in rodents, insular
deep brain stimulation led to analgesic effects in a rat model
of peripheral neuropathic pain [19], and opioid receptor
availability in the human insula is positively correlated with
pain relief after motor cortex stimulation [32]. In summary,
the insula possesses a pro-nociceptive role when hyperac-
tive, such as in cases of physiologic acute and chronic pain,
and interventions aimed at its inhibition, whether via local
lesions or via electric stimulation, are believed to tune down
its activity and prompt antinociceptive effects [19].
The disturbance of afferent signals from the posterior
inferior insula due to peripheral etiologies could explain the

299
 L. Dongyang, A.M. Fernandes, P.H.M. da Cunha et al.
emergence of neuropathic pain symptoms by the Embodied
Predictive Interoception Coding (EPIC) model [25]. The PSI
amplifies inputs coming from the thalamus via its granular
cells and conveys sensory inputs to the agranular anterior
insula (AI), where cells from the deeper layers send sen-
sory predictions back to the PSI, in which a comparison
between prediction (coming from agranular limbic and par-
alimbic areas) and the actual sensory input (coming from
thalamic relays) takes place. The prediction error is then
estimated and sent back to the AI, leading to changes in
sensory gain and behavior modulation to meet inner pre-
diction with the actual sensory experience having as the
main objective the maintenance of homeostasis. In keep-
ing with the EPIC model, peripheral neuropathic pain would
lead to deafferentation of the thalamocortical input, redu-
cing the capacity of this system to ‘‘read’’ the bodily
inputs (e.g., reduced inputs from muscle ergoceptors, from
cold thermoreceptors located within venules walls etc.),
and subsequently jeopardize interoception accuracy, lead-
ing to a mismatch between incoming bodily signals and
the predictions based on the inner inputs conveyed by the
AI/motor limbic system. Prediction error mitigation would
be then threatened, and the precision of the salience net-
work would lose its accuracy to modulate incoming sensory
inputs in an adaptive manner. The final scenario would
be a hyper-activated salience system with disproportionate
somatosensory gain (sensory gain-of-function: hyperalgesia,
spontaneous pain, allodynia), visualized in neuroimaging
studies as a hyperactive insula typical of neuropathic pain
conditions and pain-evoked cortical responses [20,38—40].
Here, we have hypothesized that high-frequency PSI d-rTMS
would allow downregulation of abnormal neuronal hyperac-
tivity by overriding the abnormal PSI firing pattern in the
setting of NeP. Frequency overriding was proposed to occur
in deep brain stimulation, providing results parallel to the
functional turning off of a neuronal network, leading to clin-
ical results analogous to lesions to this same system [27],
thus producing enhanced signal detection and dampening
of entropic noise. It has been shown that excitatory TMS
delivered to hubs of the executive networks disengaged the
tonic influence of this network from the default-mode net-
work and decrease, and consequently affected connectivity
between these two large-scale systems [9].
This pilot study is the second to test the effects of PSI
in neuropathic pain. In a previous study, we tested whether
PSI d-rTMS had analgesic effects in central NeP patients [24].
While no significant analgesic effects were found compared
to sham stimulation, PSI d-rTMS significantly increased heat
pain thresholds, suggesting an effect in nociception (improv-
ing somatosensory discrimination) that was dissociated from
the effect on clinical pain per se. This led us to hypothesize
that PSI stimulation could be useful in a different group of
patients, in which no lesions to the central processing struc-
tures of the spino-thalamo-insulo/parietal network existed,
but where abnormalities of the somatosensory pathways
were restricted to its peripheral portions, thus leading us
to target peripheral neuropathic pain.
Concerning targeting, the focality of both superficial and
deep rTMS have recently been called into question. In the
present study, we used a stimulation coil that has one of
the best depth/focality trade-offs [17], meaning that it can
reach relatively deep cortical or subcortical structures with
300
a reasonable degree of focality. We can infer that several
cortical and subcortical structures are influenced by the
stimulation pulse during deep rTMS. However, one study
specifically addressing this issue compared deep PSI TMS
stimulation against superficial TMS over the same scalp tar-
get using neuronavigation. They found that only deep TMS
changed c-fiber-dependent thermal thresholds [31]. Inter-
estingly, antinociceptive thermal effects were also obtained
after deep rTMS in central neuropathic pain patients [24]
and epileptic patients under stereo-EEG after transcorti-
cal stimulation of the PSI [18]. The finding that an electric
current induced by deep rTMS of the PSI triggers similar
psychophysical effects in healthy volunteers and central
neuropathic pain patients, compared to patients undergo-
ing direct cortical stimulation, is a strong argument in favor
of d-rTMS focality.
While we found that PSI d-rTMS brought about signifi-
cant pain relief in a relatively large proportion of refractory
peripheral neuropathic pain patients, it must be pointed out
that these effects were relatively short-lived, with the dif-
ference between active and sham stimulations vanishing one
week after the last stimulation session, when no significant
effects could be detected in neuropathic pain symptoms and
scores of pain interference with daily activities. The neces-
sity to perform maintenance sessions following an induction
cycle is needed in several other scenarios, including motor
cortex (M1) stimulation for fibromyalgia [33], and complex
regional pain syndrome [41]. It remains to test the possibil-
ity of sustaining the analgesic effects after PSI stimulation
during weekly or fortnightly maintenance sessions.
In the present study, sham stimulation was performed
with the interposition of an active double cone coil and
a sham coil that were present during all sessions in all
patients and were used according to the type of stimulation
patients were supposed to receive. While this strategy has
been used in several previous studies [2,16,24] and has been
shown to correctly maintain participants’ and raters’ blind-
ing, newer technological improvements have advocated the
use of automated blinding systems using pre-recorded cards
or USB keys, in which researchers delivering TMS would also
be blinded to the type of stimulation delivered. While such
advances are welcome and were not used here, data from
recent trials in which both automated and manual sham
deep-rTMS were employed revealed no significant differ-
ences between the sham effects of both types of sham
procedures [24]. In fact, in another study [4], we previously
demonstrated that sham effects are similar in peripheral
neuropathic pain patients even when two completely dif-
ferent techniques of noninvasive brain stimulation were
compared (i.e., transcranial direct current stimulation and
rTMS to M1).
Another limitation of our study is related to its
exploratory and pilot design, primarily performed to assess
safety, feasibility, and temporal profile of pain relief: the
paucity of data related to the quality of life, fatigue, mood,
and sleep assessment. Better appraisal of the temporal pro-
file of the effects of PSI d-rTMS will allow us to design
more comprehensive trials with assessments performed at
targeted time-points, according to the expected effects of
treatment over time.
In conclusion, this pilot study shows that deep-rTMS to
the PSI showed significant analgesic effects in refractory
 Neurophysiologie Clinique/Clinical Neurophysiology 51 (2021) 291—302
peripheral NeP patients and was safe, opening the field for
larger trials testing its use in long-term scenarios with main-
tenance sessions of stimulation.
Funding
Pain Center and LIM62 University of São Paulo, Brazil.
Conflicts of interest
No conflict of interest to be reported.
Appendix A. Supplementary data
Supplementary material related to this article can be found,
in the online version, at doi:https://doi.org/10.1016/j.
neucli.2021.06.003.
References
[1] Alonso-Matielo H, Goncalves ES, Campos M, Oliveira VRS,
Toniolo EF, Alves AS, et al. Electrical stimulation of the pos-
terior insula induces mechanical analgesia in a rodent model
of neuropathic pain by modulating GABAergic signaling and
activity in the pain circuitry. Brain Res 2021;1754:147237.
[2] Andre-Obadia N, Mertens P, Gueguen A, Peyron R, Garcia-
Larrea L. Pain relief by rTMS: differential effect of current
flow but no specific action on pain subtypes. Neurology
2008;71(11):833—40.
[3] Andre-Obadia N, Peyron R, Mertens P, Mauguiere F, Laurent
B, Garcia-Larrea L. Transcranial magnetic stimulation for pain
control. Double-blind study of different frequencies against
placebo, and correlation with motor cortex stimulation effi-
cacy. Clin Neurophysiol 2006;117(7):1536—44.
[4] Attal N, Ayache SS, Ciampi De Andrade D, Mhalla A, Baudic S,
Jazat F, et al. Repetitive transcranial magnetic stimulation and
transcranial direct-current stimulation in neuropathic pain due
to radiculopathy: a randomized sham-controlled comparative
study. Pain 2016;157(6):1224—31.
[5] Baptista AF, Fernandes A, Sa KN, Okano AH, Brunoni
AR, Lara-Solares A, et al. Latin American and Caribbean
consensus on noninvasive central nervous system neuromod-
ulation for chronic pain management (LAC2-NIN-CP). Pain Rep
2019;4(1):e692.
[6] Bouhassira D, Attal N, Fermanian J, Alchaar H, Gautron M,
Masquelier E, et al. Development and validation of the neu-
ropathic pain symptom inventory. Pain 2004;108(3):248—57.
[7] Bouhassira D, Lanteri-Minet M, Attal N, Laurent B, Touboul C.
Prevalence of chronic pain with neuropathic characteristics in
the general population. Pain 2008;136(3):380—7.
[8] Bowsher D. Sensory consequences of stroke. Lancet
1993;341(8838):156.
[9] Chen AC, Oathes DJ, Chang C, Bradley T, Zhou ZW, Williams
LM, et al. Causal interactions between fronto-parietal central
executive and default-mode networks in humans. Proc Natl
Acad Sci U S A 2013;110(49):19944—9.
[10] Ciampi de Andrade D, Galhardoni R, Pinto LF, Lancelotti R, Rosi
Jr J, Marcolin MA, et al. Into the island: a new technique of
non-invasive cortical stimulation of the insula. Neurophysiol
Clin 2012;42(6):363—8.
[11] Ciampi de Andrade D, Mhalla A, Adam F, Texeira MJ, Bouhassira
D. Repetitive transcranial magnetic stimulation induced anal-
gesia depends on N-methyl-d-aspartate glutamate receptors.
Pain 2014;155(3):598—605.
301
[12] Ciaramitaro P, Cruccu G, de Tommaso M, Devigili G, Fornasari
D, Geppetti P, et al. A Delphi consensus statement of the
Neuropathic Pain Special Interest Group of the Italian Neuro-
logical Society on pharmacoresistant neuropathic pain. Neurol
Sci 2019;40(7):1425—31.
[13] Coghill RC, Talbot JD, Evans AC, Meyer E, Gjedde A, Bushnell
MC, et al. Distributed processing of pain and vibration by the
human brain. J Neurosci 1994;14(7):4095—108.
[14] de Andrade DC, Ferreira KA, Nishimura CM, Yeng LT, Batista
AF, de Sa K, et al. Psychometric validation of the Portuguese
version of the Neuropathic Pain Symptoms Inventory. Health
Qual Life Outcomes 2011;9:107.
[15] de Andrade DC, Mhalla A, Adam F, Texeira MJ, Bouhassira D.
Neuropharmacological basis of rTMS-induced analgesia: the
role of endogenous opioids. Pain 2011;152(2):320—6.
[16] De Martino E, Fernandes AM, Galhardoni R, De Oliveira Souza
C, Ciampi De Andrade D, Graven-Nielsen T. Sessions of pro-
longed continuous theta burst stimulation or high-frequency
10 Hz stimulation to left dorsolateral prefrontal cortex for 3
days decreased pain sensitivity by modulation of the efficacy
of conditioned pain modulation. J Pain 2019;20(12):1459—69.
[17] Deng ZD, Lisanby SH, Peterchev AV. Electric field depth-focality
tradeoff in transcranial magnetic stimulation: simulation com-
parison of 50 coil designs. Brain Stimul 2013;6(1):1—13.
[18] Denis DJ, Marouf R, Rainville P, Bouthillier A, Nguyen DK.
Effects of insular stimulation on thermal nociception. Eur J
Pain 2016;20(5):800—10.
[19] Dimov LF, Toniolo EF, Alonso-Matielo H, de Andrade DC, Garcia-
Larrea L, Ballester G, et al. Electrical stimulation of the
insular cortex as a novel target for the relief of refractory
pain: an experimental approach in rodents. Behav Brain Res
2018;346:86—95.
[20] Ducreux D, Attal N, Parker F, Bouhassira D. Mechanisms of
central neuropathic pain: a combined psychophysical and fMRI
study in syringomyelia. Brain 2006;129(Pt 4):963—76.
[21] Evans AC, Collins DL, Mills SR, Brown ED, Kelly RL, Peters TM.
3D statistical neuroanatomical models from 305 MRI volumes.
In: IEEE Conference Record Nuclear Science Symposium and
Medical Imaging Conference 3. 1993. p. 1813—7.
[22] Faillenot I, Heckemann RA, Frot M, Hammers A. Macroanatomy
and 3D probabilistic atlas of the human insula. Neuroimage
2017;150:88—98.
[23] Finnerup NB, Haroutounian S, Kamerman P, Baron R, Ben-
nett DLH, Bouhassira D, et al. Neuropathic pain: an updated
grading system for research and clinical practice. Pain
2016;157(8):1599—606.
[24] Galhardoni R, Aparecida da Silva V, Garcia-Larrea L,
Dale C, Baptista AF, Barbosa LM, et al. Insular and
anterior cingulate cortex deep stimulation for central neu-
ropathic pain: disassembling the percept of pain. Neurology
2019;92(18):e2165—75.
[25] Ghaziri J, Tucholka A, Girard G, Boucher O, Houde JC,
Descoteaux M, et al. Subcortical structural connectivity of insu-
lar subregions. Sci Rep 2018;8(1):8596.
[26] Greenspan JD, Lee RR, Lenz FA. Pain sensitivity alterations as
a function of lesion location in the parasylvian cortex. Pain
1999;81(3):273—82.
[27] Grill WM, Snyder AN, Miocinovic S. Deep brain stimulation
creates an informational lesion of the stimulated nucleus. Neu-
roreport 2004;15(7):1137—40.
[28] Lefaucheur JP, Aleman A, Baeken C, Benninger DH, Brunelin
J, Di Lazzaro V, et al. Evidence-based guidelines on the
therapeutic use of repetitive transcranial magnetic stim-
ulation (rTMS): an update (2014-2018). Clin Neurophysiol
2020;131(2):474—528.
[29] Lefaucheur JP, Ayache SS, Sorel M, Farhat WH, Zouari HG,
Ciampi de Andrade D, et al. Analgesic effects of repetitive
transcranial magnetic stimulation of the motor cortex in neu-
 L. Dongyang, A.M. Fernandes, P.H.M. da Cunha et al.
ropathic pain: influence of theta burst stimulation priming. Eur
J Pain 2012;16(10):1403—13.
[30] Lefaucheur JP, Drouot X, Menard-Lefaucheur I, Keravel Y,
Nguyen JP. Motor cortex rTMS restores defective intra-
cortical inhibition in chronic neuropathic pain. Neurology
2006;67(9):1568—74.
[31] Lenoir C, Algoet M, Mouraux A. Deep continuous theta burst
stimulation of the operculo-insular cortex selectively affects
Adelta-fibre heat pain. J Physiol 2018;596(19):4767—87.
[32] Maarrawi J, Peyron R, Mertens P, Costes N, Magnin M, Sindou M,
et al. Brain opioid receptor density predicts motor cortex stim-
ulation efficacy for chronic pain. Pain 2013;154(11):2563—8.
[33] Mhalla A, Baudic S, de Andrade DC, Gautron M, Perrot S,
Teixeira MJ, et al. Long-term maintenance of the analgesic
effects of transcranial magnetic stimulation in fibromyalgia.
Pain 2011;152(7):1478—85.
[34] Moisset X, Bouhassira D, Avez Couturier J, Alchaar H, Conradi S,
Delmotte MH, et al. Pharmacological and non-pharmacological
treatments for neuropathic pain: systematic review and French
recommendations. Rev Neurol (Paris) 2020;176(5):325—52.
[35] Moisset X, de Andrade DC, Bouhassira D. From pulses to pain
relief: an update on the mechanisms of rTMS-induced analgesic
effects. Eur J Pain 2016;20(5):689—700.
[36] Onesti E, Gabriele M, Cambieri C, Ceccanti M, Raccah R, Di
Stefano G, et al. H-coil repetitive transcranial magnetic stim-
ulation for pain relief in patients with diabetic neuropathy. Eur
J Pain 2013;17(9):1347—56.
[37] Pagano RL, Fonoff ET, Dale CS, Ballester G, Teixeira MJ, Britto
LRG. Motor cortex stimulation inhibits thalamic sensory neu-
rons and enhances activity of PAG neurons: possible pathways
for antinociception. Pain 2012;153(12):2359—69.
302
[38] Peyron R, Faillenot I, Mertens P, Laurent B, Garcia-Larrea
L. Motor cortex stimulation in neuropathic pain. Correlations
between analgesic effect and hemodynamic changes in the
brain. A PET study. Neuroimage 2007;34(1):310—21.
[39] Peyron R, Laurent B, Garcia-Larrea L. Functional imaging of
brain responses to pain. A review and meta-analysis. Neuro-
physiol Clin 2000;30(5):263—88.
[40] Peyron R, Schneider F, Faillenot I, Convers P, Barral FG, Garcia-
Larrea L, et al. An fMRI study of cortical representation of
mechanical allodynia in patients with neuropathic pain. Neu-
rology 2004;63(10):1838—46.
[41] Picarelli H, Teixeira MJ, de Andrade DC, Myczkowski ML, Luvi-
sotto TB, Yeng LT, et al. Repetitive transcranial magnetic
stimulation is efficacious as an add-on to pharmacological ther-
apy in complex regional pain syndrome (CRPS) type I. J Pain
2010;11(11):1203—10.
[42] Rocha Rde O, Teixeira MJ, Yeng LT, Cantara MG, Faria VG, Lig-
gieri V, et al. Thoracic sympathetic block for the treatment
of complex regional pain syndrome type I: a double-blind ran-
domized controlled study. Pain 2014;155(11):2274—81.
[43] Silva GD, Lopes PS, Fonoff ET, Pagano RL. The spinal anti-
inflammatory mechanism of motor cortex stimulation: cause
of success and refractoriness in neuropathic pain? J Neuroin-
flammation 2015;12:10.
[44] Smith BH, Torrance N, Ferguson JA, Bennett MI, Serpell MG,
Dunn KM. Towards a definition of refractory neuropathic pain
for epidemiological research. An international Delphi survey of
experts. BMC Neurol 2012;12:29.
[45] Zunhammer M, Schweizer LM, Witte V, Harris RE, Bingel U,
Schmidt-Wilcke T. Combined glutamate and glutamine levels
in pain-processing brain regions are associated with individual
pain sensitivity. Pain 2016;157(10):2248—56.