Motor cortex stimulation for pain
control induces changes in the
J. Maarrawi, MD,
PhD
R. Peyron, MD, PhD
P. Mertens, MD, PhD
N. Costes, PhD
M. Magnin, PhD
M. Sindou, MD, DSc
B. Laurent, MD, PhD
L. Garcia-Larrea, MD,
PhD
Address correspondence and
reprint requests to Dr. Joseph
Maarrawi, INSERM U879
(Central Integration of Pain), 59
Bd Pinel, 69394, Lyon, France
joseph.maarrawi@chu-lyon.fr
or
jomaarrawi@hotmail.com
endogenous opioid system
ABSTRACT
Background: Motor cortex stimulation (MCS) for neuropathic pain control induces focal cerebral
blood flow changes involving regions with high density of opioid receptors. We studied the possi-
ble contribution of the endogenous opioid system to MCS-related pain relief.
Methods: Changes in opioid receptor availability induced by MCS were studied with PET scan and
[11C]diprenorphine in eight patients with refractory neuropathic pain. Each patient underwent
two preoperative (test–retest) PET scans and one postoperative PET scan acquired after 7
months of chronic MCS.
Results: The two preoperative scans, performed at 2 weeks interval, did not show significant
differences. Conversely, postoperative compared with preoperative PET scans revealed signifi-
cant decreases of [11C]diprenorphine binding in the anterior middle cingulate cortex (aMCC), peri-
aqueductal gray (PAG), prefrontal cortex, and cerebellum. Binding changes in aMCC and PAG
were significantly correlated with pain relief.
Conclusion: The decrease in binding of the exogenous ligand was most likely explained by recep-
tor occupancy due to enhanced secretion of endogenous opioids. Motor cortex stimulation (MCS)
may thus induce release of endogenous opioids in brain structures involved in the processing of
acute and chronic pain. Correlation of this effect with pain relief in at least two of these structures
supports the role of the endogenous opioid system in pain control induced by MCS.
Neurology® 2007;69:827–834
Chronic neuropathic pain is one of the most challenging conditions for physicians. Phar-
macotherapy often fails to control adequately this symptom,1 prompting the use of surgi-
cal procedures including neurostimulation. Stimulation of pyramidal tracts was shown to
inhibit afferent transmission in the dorsal horn of animals2 and to produce analgesic
effects on neuropathic pain after cortical injury in man.3 Despite these early reports, the
use of motor cortex stimulation (MCS) for pain control was only documented in 1991.4
Now, this technique is replacing deep brain stimulation in the surgical management of
intractable neuropathic pain, with progressive widespread use in several pain centers.5-7
The mechanisms underlying neuropathic pain relief from MCS remain poorly under-
stood. Electrophysiologic studies demonstrated inhibitory effects induced by MCS at
thalamic and spinal levels.8-11 However, most of the known physiologic effects of MCS in
humans are derived from PET studies of cerebral blood flow. This technique demon-
strated MCS-related increases in cerebral blood flow (CBF) within the ventrolateral thal-
amus (a region directly connected with the stimulated motor cortex), medial thalamus,
insula, anterior cingulate gyrus, and brainstem.9,12
Pain relief after MCS is generally delayed with respect to the periods of actual cortical
stimulation. In accordance with this, a protracted activation was recently found13 in
From INSERM, U879, Bron, France; Université de Lyon 1, Lyon, France; Université de Saint-Etienne, Saint-Etienne, France (J.M., R.P., P.M.,
M.M., B.L., L.G.-L.); Functional Neurosurgery Department, Hôpital Neurologique, HCL, Lyon, France (J.M., P.M., M.S.); Department of
Neurology and Pain Center, Hôpital Bellevue, St-Etienne, France (R.P., B.L.); CERMEP, Imagerie du vivant (PET Scan Center), Lyon, France
(N.C.); and Department of Anatomy, Claude Bernard University, Lyon, France (P.M.).
Supported by the BENOIT Foundation (Dr. Maarrawi) and the “Projet hospitalier de recherche clinique, appel d’offre 2000, CHU de St
Etienne.”
Disclosure: The authors report no conflicts of interest.
Copyright © 2007 by AAN Enterprises, Inc. 827
 Table 1 Clinical characteristics of patients

Pain rating (VAS)*

Pain
Patient Age, Type of Lesion Duration Sensory Localization response
no. y Sex pain location Etiology of pain, mo deficit of pain Pre Post to opioids Treatment

1 60 F CPSP Capsulolenticular Stroke 34 R (⫺F) RLL 8 (9) 5 (5) — G/P

2 32 F CPSP Juxtathalamic Stroke 51 R R 8 (9) 2 (3) — Clm/G

3 60 F CPSP Thalamic Stroke 53 R R (LL ⬍ UL ⫹F) 7 (10) 4 (8) Morphine (Pt) G/O

4 54 M CPSP Capsulothalamic Stroke 18 R R 8 (9) 7 (7) — Clm/P

5 60 M CPSP Capsulolenticular Stroke 48 L LUL 7 (10) 2 (3) Dextropropoxyphene (Pt) Clz/G

6 62 M CPSP Brainstem Stroke 30 L LLL ⬎ LUL; 8 (9); 3 5 (7); 2 — G

RLL ⬎ RUL

7 47 M CPSP Wallenberg Stroke 19 RUL ⫹RLL ⫹LF LF (V.2, V.3) 4 (10) 2 (6) — Cb/Clm/P

8 64 M P Trigeminal V.2 Trauma 192 LF (V.2) LF (V.2) 10 (10) 7 (7) — Clm/G

* Continuous pain rating on visual analog scale (VAS) ranging from 0 (no pain) to 10 (worst imaginable pain); in parentheses, paroxysmal pain on the same
VAS scores.
CPSP ⫽ central poststroke pain; P ⫽ peripheral neuropathic pain; R ⫽ right; L ⫽ left; ⫹F ⫽ involving face; ⫺F ⫽ not involving face; UL ⫽ upper limb; LL ⫽ lower
limb; Pre ⫽ preoperative; Post ⫽ postoperative; Pt ⫽ partial pain relief; Cb ⫽ carbamazepine (1,000 mg/day); Clm ⫽ clomipramine (125 to 150 mg/day); Clz ⫽
clonazepam (2.5 to 3 mg/day); G ⫽ gabapentin (2,400 to 3,200 mg/day); O ⫽ oxcarbazepine (1,200 mg/day); P ⫽ paracetamol (1,500 mg/day).

anumber of brain regions related to pain
processing, including the anterior cingulate
gyrus (ACG) and the periaqueductal gray
matter (PAG), which persisted more than
60 minutes after MCS discontinuation.
This long-lasting activation of ACG and
PAG is in accord with the persistence of
analgesic effects for a long time after MCS
discontinuation, and suggests that the
MCS clinical effects might be mediated by
long-lasting functional changes in these
structures. ACG and PAG contain a high
density of opioid receptors14 and pertain to
the cortical–subcortical network activated
during opioid analgesia in humans15; there-
fore, we considered the hypothesis that
long-lasting MCS effects could be, at least
partly, mediated through endogenous
opioids.
To test this hypothesis, we used the non-
selective opioid antagonist [11C]diprenor-
phine and PET to investigate changes in
opioid receptor availability before and af-
ter chronic MCS in a sample of patients
with chronic intractable neuropathic pain.
METHODS Patients. Between 2002 and 2005, eight pa-
tients (five men) aged 32 to 64 years (mean 54.9 years), all
right-handed, were included in this study. Seven of them had
strictly unilateral (n ⫽ 7) or strongly asymmetrical (n ⫽ 1)
chronic neuropathic pain, which was of central supraspinal
origin (central poststroke pain [CPSP]) in all but one with
trigeminal nerve injury.
Patients were in good general health, excepting motor
and sensory symptoms related to their neurologic disease.
Clinical data are summarized in table 1. Mean visual analog
scale (VAS) ratings under medication were 7.5 ⫾ 1.7 for con-
tinuous pain and 9.5 ⫾ 0.5 for paroxysmal pain. Pain was
considered refractory to medical therapy, because subjective
VAS ratings had remained superior to 5/10 for at least 2
years despite pharmacologic polytherapy. Accordingly, pa-
tients were candidates for MCS for pain control, and they all
underwent surgery. All patients were off opioid medication
for at least 2 months before entering the protocol, and re-
mained so during the whole inclusion period of the study. All
gave their written informed consent to the study, which was
approved by the local medical ethics committee (University
Hospital St. Etienne, France).
Surgical procedure. Primary motor cortex localization
was determined using somatosensory evoked potential re-
cordings16 and magnetic resonance “neuronavigation’”
(Stealth Station, Medtronic Sofamor-Danek). One or two
4-electrode strips (Resume Medtronic) were implanted epi-
durally over the motor representation of the painful area
contralateral to clinical pain, and then connected to a subcu-
taneously implanted stimulator: radiofrequency Xtrel or
Synergy, in a one-stage surgery. Stimulation intensity was
adapted gradually during the postoperative stage so as to
optimize the analgesic effect. After adaptation, the stimulus
(0.5 to 5 V, pulse duration 180 ␮s, 35 Hz) was maintained
under motor threshold. A cyclic mode of stimulation was
used in all patients, with “on” periods of 30 or 60 minutes
and “off” periods of 2 hours.
PET acquisitions and radioligand. PET acquisitions
were performed using a Siemens ECAT HR⫹ scanner in
three-dimensional mode (63 slices, 2.425 mm thick each,
transaxial matrix of 128 ⫻ 128 pixels), with a maximum
center field resolution of 4.4 ⫻ 4.41 ⫻ 4.41 mm. The nonse-
lective opioid receptor antagonist [11C]diprenorphine was
synthesized in situ at the PET center of Lyon using the meth-
ylation method.17 Radiochemical purity was ⬎95%, and spe-

828 Neurology 69 August 28, 2007

 Table 2 Individual Mean BP data, in the two preoperative and the postoperative PET scan sessions, within anatomic regions where significant
BP changes were observed

PAG aMCC Cerebellum Prefrontal cortex

Patient PET Pre Post Chg, % Pre Post Chg, % Pre Post Chg, % Pre Post Chg, %

1 1 0.64 (0.11) 0.44 (0.09) 30 0.55 (0.1) 0.42 (0.09) 22.2 0.46 (0.09) 0.38 (0.05) 15.5 0.70 (0.2) 0.49 (0.13) 31.5

2 0.62 (0.14) 0.53 (0.08) 0.44 (0.06) 0.73 (0.18)

2 1 0.61 (0.15) 0.38 (0.08) 37 0.53 (0.12) 0.38 (0.1) 30.3 0.45 (0.1) 0.35 (0.09) 22.2 0.71 (0.17) 0.54(0.16) 25.5

2 0.6 (0.11) 0.56 (0.1) 0.45 (0.07) 0.74 (0.18)

3 1 0.59 (0.12) 0.51 (0.11) 16.3 0.52 (0.09) 0.41 (0.09) 23.4 0.47 (0.06) 0.39 (0.07) 14.3 0.75 (0.16) 0.58 (0.09) 21.6

2 0.62 (0.11) 0.55 (0.13) 0.44 (0.16) 0.73 (0.15)

4 1 0.64 (0.14) 0.53 (0.1) 16.3 0.54 (0.07) 0.48 (0.1) 10.3 0.43 (0.11) 0.4 (0.1) 10.11 0.71 (0.2) 0.6 (0.08) 17.2

2 0.62 (0.14) 0.53 (0.11) 0.46 (0.08) 0.74 (0.21)

5 1 0.65 (0.12) 0.35 (0.11) 45.7 0.55 (0.1) 0.4 (0.08) 28 0.45 (0.09) 0.36 (0.05) 20.9 0.76 (0.16) 0.48 (0.12) 35.5

2 0.64 (0.16) 0.56 (0.08) 0.46 (0.1) 0.73 (0.17)

6 1 0.63 (0.16) 0.5 (0.12) 21.8 0.54 (0.09) 0.43 (0.09) 19.6 0.44 (0.07) 0.34 (0.09) 25.3 0.72 (0.22) 0.53 (0.11) 25.9

2 0.65 (0.12) 0.53 (0.08) 0.47 (0.08) 0.71 (0.17)

7 1 0.61 (0.14) 0.51 (0.08) 17 0.51 (0.14) 0.43 (0.12) 18.9 0.48 (0.09) 0.38 (0.05) 19.1 0.7 (0.15) 0.55 (0.14) 22

2 0.62 (0.12) 0.55 (0.11) 0.46 (0.09) 0.71 (0.18)

8 1 0.63 (0.16) 0.49 (0.12) 21 0.56 (0.09) 0.45 (0.08) 17.2 0.47 (0.11) 0.37 (0.11) 17.7 0.7 (0.13) 0.57 (0.12) 20.3

2 0.61 (0.11) 0.53 (0.12) 0.43 (0.07) 0.73 (0.18)

Changes (Chg) in binding potential (BP) were calculated using this formula: [100 * (mean preoperative value ⫺ postoperative value)/mean preoperative value].
Numbers in parentheses are standard deviations.
PAG ⫽ periaqueductal gray; aMCC ⫽ anterior middle cingulate cortex; Pre ⫽ preoperative; Post ⫽ postoperative.

cific activity at injection time was in the range of 1,100 to
2,960 MBq/␮mol.
Before injection, a 10-minute transmission scan using an
external rotating rod source of 68Ge was performed to cor-
rect radiation attenuation of tissues of different densities.
Then, 171 to 272 MBq of high-specific-activity [11C]di-
prenorphine was injected IV, and a 70-minute continuous
acquisition was performed on dynamic mode comprising 37
consecutive time frames (15 ⫻ 20, 15 ⫻ 120, 7 ⫻ 300
seconds).
Each patient underwent two consecutive PET sessions at
2-week intervals before surgery (to ensure reproducibility)
and then a third PET study after surgery. Mean pain scores
remained unchanged between the first and the second (pre-
operative) sessions. The tracer distribution18 (table 2) was
similar in the two preoperative PET sessions. Thereafter, we
included both of them (as repetitions) in the SPM statistical
design used for comparisons. After surgery and once the
characteristics (intensity, frequency) of the stimulation were
considered “optimal” for every patient, a third postoperative
PET scan (generator switched off for 12 hours before scan)
was performed after 2 months of chronic stimulation (i.e., 7
⫾ 0.88 months after surgery). Medication was kept un-
changed across the three PET sessions.

Data processing. Data processing was performed using

Figure 1 Comparative time–activity curves (Bq/ two software programs: CAPP (Clinical Application Pro-
mL) during 70 minutes after gramming Package, CTI, Knoxville, TN) and SPM99 (Statis-
injection of [11C]diprenorphine, in tical Parametric Mapping, Wellcome Department of
thalamus (region with a high opioid
Cognitive Neurology, London, UK). The first step included
BP) and occipital lobe (region with a
orientation of the interhemispheric fissure along the y–z
low opioid BP)
plane at x ⫽ 0, and realignment of anteroposterior commis-
sural line to overlap the x–y plane at z ⫽ 0. Parametric im-
Stable values of the curve
after time frame 19 in the
ages of the binding potential (BP) index for
thalamus reflect ligand [11C]diprenorphine were obtained using the reference tissue
fixation to receptors, model19,20 with occipital cortex as reference ([pixel value ⫺
whereas the low density of occipital value]/occipital value). This BP estimation was ap-
opioid receptors in the plied to all brain voxels of the mean summated value of ra-
occipital lobe results in dioactivity image from Frame 20 to Frame 37. The occipital
rapid decline of the activity values were calculated using regions of interest with identi-
with time. The ascending
cal volume and shape for all patients. Frames 20 to 37 were
common part of the curves
considered as reflecting [11C]diprenorphine binding to recep-
(0 to 5 minutes) mainly
reflects the blood flow tors with minimal flow effect because mean radioactivity
effect after bolus injection. curve over the 70-minute acquisition in the thalamus (a re-
BP ⫽ binding potential. gion exhibiting a high receptor density) showed stable cor-

Neurology 69 August 28, 2007 829

 rected decay values after Frame 19 (figure 1). Hemispheres
were standardized across subjects as “contralateral” or “ipsi-
lateral” to pain in each patient; hence, parametric images
were flipped along the x-axis when necessary,20,21 so that the
side of MCS was the left side in all patients (left hemisphere
contralateral to pain in all images).
Individual parametric images were coregistered with
their magnetic resonance (MR) images (T1-weighted) in
each patient by application of the translational and rota-
tional transformation values of coregistration of the MR im-
ages with the mean summated image (from Frame 20 to
Frame 37). Each patient’s MR image was transformed with
nonlinear basis functions into the International Consortium
for Brain Mapping (ICBM) standardized space (www.loni.
ucla.edu/ICBM/, Montreal, Canada). Identical transforma-
tion characteristics as those used for MR images were then
applied to the previously coregistered parametric images of
each patient. These transformations yielded a standardized
set of images mutually comparable on a voxel-to-voxel basis.
Because the two hemispheres of the standardized ICBM
space are not strictly symmetric, patients’ MR images
flipped along the x-axis were first transformed using nonlin-
ear functions into the standardized space of ICBM, and the
transformation characteristics were then applied to the cor-
responding flipped parametric images. Using this method,
the contralateral hemisphere was “normalized” to left and
the ipsilateral hemisphere to the right hemisphere of the
standard ICBM space. Before all statistical analyses,
smoothing of images was performed using a three-
dimensional Gaussian kernel of 10 mm.
Statistical analyses. Paired preoperative and postopera-
tive PET scans were compared using a between-group one-
way analysis of variance model on a voxel-wise manner,
with SPM99 software (first group: preoperative PET scans;
second group: postoperative PET scans). Contrasts of inter-
est were 1) preoperative vs postoperative and 2) postopera-
tive vs preoperative scans. To account for the effect of
random global differences, the global binding index was in-
troduced as a covariate in the comparisons.
An inclusive mask of opioid rich structures was con-
structed using inclusive three-dimensional masks obtained
from the VOI Tools module of SPM99 software. This mask
included, in each hemisphere, regions known to have a high
density of opioid receptors, i.e., the insula, thalamus, stria-
tum, caudate nucleus, orbitofrontal, prefrontal and poste-
rior temporal cortices, mid and anterior cingulate gyrus,
posterior midbrain, and cerebellum. For all comparisons, we
used corrected probability thresholds at ␣ ⫽ 0.05 (Bonfer-
roni corrected for multiple resel comparisons) corresponding
to p ⬍ 0.0005 at cluster level and p ⬍ 0.001 at voxel level. An
exception was made in the posterior midbrain, small area
where the cluster level for significance was set at 100 contig-
uous voxels.
For each patient, mean numerical pain rating of continu-
ous and paroxysmal pain component was calculated for each
PET session.
RESULTS Changes in opioid receptor binding after
MCS. Voxel-wise comparison of preoperative and
postoperative PET scans showed significant de-
crease of opioid receptor binding in the postoper-
ative compared with the preoperative scans.
830 Neurology 69 August 28, 2007
Statistical (Z) images at the group level, superim-
posed on normalized MR images, are shown in
figures 2 and 3. BP decrease concerned the poste-
rior part of the midbrain (25.6% decrease), ante-
rior portion of the middle cingulate gyrus
(aMCC; 21.2% decrease), lateral prefrontal cor-
tex (23.3% decrease), and cerebellum (18.3% de-
crease). The posterior midbrain cluster was
consistent with the localization of the PAG in the
mesencephalon. Table 3 indicates the spatial co-
ordinates of the peak of significance within each
cluster in the normalized ICBM space, along with
the cluster size, t score, significance level, and per-
centage of binding decrease. Decrease of opioid
binding was a consistent finding across all sub-
jects, as shown by individual BP data summarized
in table 2.
All significant changes detected, without ex-
ception, corresponded to relative reductions of
opioid receptor BP in the postoperative period.
The mirror contrast (postoperative vs preopera-
tive) exploring possible relative increases of opi-
oid BP in the postoperative phase did not yield
any significant result.
Correlations between opioid receptor BP changes
and clinical pain relief. VAS scores before and af-
ter chronic MCS differed significantly. Statistical
correlation between the percentage of BP decrease
(ratio of postoperative vs preoperative BP) and
that of pain relief on VAS (ratio of postoperative
vs preoperative mean scores) was tested in each
anatomic region where significant BP decrease
was observed. Spearman nonparametric correla-
tion demonstrated high correlation coefficients in
PAG (r ⫽ 0.78, one-tailed p ⫽ 0.01) and anterior
middle cingulate cortex (r ⫽ 0.66; one-tailed p ⫽
0.04), whereas it approached significance in pre-
frontal cortex (r ⫽ 0.62; one-tailed p ⫽ 0.048)
(figure 4).
DISCUSSION The aim of this study was to com-
pare opioid receptor availability in patients with
chronic refractory neuropathic pain, before and
after chronic MCS. The main finding is a decrease
in the availability of opioid receptors to [11C]di-
prenorphine after several months of stimulation,
relative to the period before surgery. Decrease in
receptor availability to our external ligand was
significant in the aMCC, PAG, prefrontal cortex,
and cerebellum. The magnitude of binding de-
crease in the aMCC and PAG was significantly
and positively correlated with the degree of clini-
cal pain relief. Our interpretation is that regional
decrease of opioid binding was due to an increase
in endogenous opioid secretion. The correlation
Both binding potential (BP)
and magnetic resonance Figure 2 Clusters of significantly reduced [11C]diprenorphine BP in the postoperative session relative to the
(MR) images were two preoperative ones, superimposed onto T1-weighted MR images
normalized to the standard
stereotactic International
Consortium for Brain
Mapping space. The color
scale (left) corresponds to
the Z score values of
displayed clusters. PAG ⫽
periaqueductal gray;
aMCC ⫽ anterior middle
cingulate cortex. Diagrams
on the right correspond to
the two preoperative and
postoperative mean BP
values of the eight patients
in the four relevant regions.

of this metabolic effect with the clinical pain relief
suggests that the activation of the endogenous
opioid system is one of the mechanisms of analge-
sia induced by MCS.
The diprenorphine BP refers to the ratio be-
tween the number of opioid receptor sites avail-
able for the [11C]diprenorphine (Bmax) and the
affinity of such receptors for this ligand (KD).19
Although variations in BP may result from
changes in either Bmax or affinity, pain-related
changes in opioid peptide expression did
not change receptor affinity.22 Moreover, KD did
not affect the binding of opioid receptor antago-
nists,23 of which [11C]diprenorphine is an exam-

Figure 3 Clusters of significantly reduced [11C]diprenorphine BP in the postoperative group, superimposed

onto axial slices of T1-weighted MR images

Both binding potential (BP)

and magnetic resonance
(MR) images were
normalized to the standard
stereotactic International
Consortium for Brain
Mapping space.
Stereotactic z coordinates
(in mm) of each axial slice
are relative to the anterior
commissure–posterior
commissure horizontal
plane. The color scale (right)
corresponds to the Z score
values of displayed
clusters.

Neurology 69 August 28, 2007 831

 Table 3 Anatomic structures with significant decreased binding potential (cluster size, T score, p value), after
MCS

Peak coordinates Cluster Percentage

Anatomic structure x/y/z within cluster size T score p Value decrease

PAG ⫺8/⫺26/⫺2 135 4.64 0.00023 25.6

aMCC ⫺3/⫺26/⫺2 42 3.9 0.00091 21.2

Prefrontal cortex ⫺36/⫺4/⫺4 203 4.31 0.00042 23.3

Cerebellum ⫺54/⫺22/⫺8 128 3.94 0.00085 18.3

Cluster size is in number of contiguous voxels. Peak coordinates (in mm) within each cluster according to the standardized
space of the International Consortium for Brain Mapping.
MCS ⫽ motor cortex stimulation; PAG ⫽ periaqueductal gray; aMCC ⫽ anterior middle cingulate cortex.

ple. Therefore, the decrease in opioid binding in
our patients can be reasonably attributed to a re-
duction in the density of available receptors. Al-
though receptor binding and density are in turn
influenced by age, sex,24 and analgesic drug in-
take, this latter was kept unchanged between the
two PET scan sessions (see Methods, PET acqui-
sitions and radioligand), and age or sex could not
be confounding factors because every patient was
compared with himself/herself, and the delay of 7
months between the preoperative and postopera-
tive PET seems too short to let age be responsible
for any change in receptor density.
Changes in opioid receptor density cannot be
considered a consequence of pain relief, because
any sustained decrease in pain intensity should
have led to a decreased release of endogenous opi-
oids,25 and as a corollary to increased diprenor-
phine BP and receptor up-regulation—i.e., the
contrary of what was observed in this study. De-
crease in receptor availability can be secondary
either to enhanced secretion of endogenous opi-
oid peptides or to loss of opioid receptors. Focal
loss of receptors has indeed been suggested to ex-
ist in patients with CPSP18,26 and thus should have
existed in our patients too, in the basal state (i.e.,

Figure 4 Statistical correlation between the percentage of BP decrease (ratio of postoperative vs preoperative
BP) and that of pain relief on visual analog scale (ratio of postoperative vs preoperative mean
scores) tested in the four regions presenting a significant variation of opioid BP after motor cortex
stimulation

Level of significance is
established in
periaqueductal gray and
anterior middle cingulate
gyrus, whereas it
approaches the significance
limit in the prefrontal
cortex. BP ⫽ binding
potential.

832 Neurology 69 August 28, 2007

before MCS operation). A further loss of recep-
tors due to chronic MCS, however, seems un-
likely, especially because it would be difficult to
explain why such further receptor loss was associ-
ated with pain relief. Conversely, an increase in
endogenous opioid secretion due to chronic MCS
would be consistent with our results, because it
would decrease receptor availability to diprenor-
phine by occupying binding sites in membrane re-
ceptors, rendering them unavailable to the
exogenous ligand.27 An increase in endogenous
opioids should in turn lead to reactive down-
regulation and internalization of receptors, fur-
ther decreasing receptor availability and BP.28
Thus, in this context, an increase in endogenous
opioid secretion after MCS may have led to a de-
crease in opioid receptor availability by two
mechanisms, namely 1) an increased occupancy
of available opioid receptors by endogenous opi-
oids and 2) a (possible) down-regulation of re-
maining ones secondary to enhanced opioid
secretion.
One of the confounding factors that may have
influenced our results is the pooling of data from
patients with right- and left-sided clinical pain,
using the flip method along the x-axis (see Meth-
ods). However, no interhemispheric differences in
the distribution of the BP of diprenorphine have
been documented to date. Indeed, we verified ab-
sence of interhemispheric differences in 18 control
subjects (unpublished data) who underwent PET
-scans with [11C]diprenorphine. Although we can
reasonably rule out this confounding factor, it re-
mains that having pain in the left or right side of
the body may have different metabolic conse-
quences that we could not measure separately in
our study.
Endogenous opioid sampling in the CSF had
been used in the literature in several painful con-
ditions29 and after analgesic interventions such as
acupuncture30 or transcutaneous electrical nerve
stimulation.31 However, the accuracy CSF sam-
pling to reflect exactly what happens inside the
synapses is still debated.32,33 Thus, in our study,
CSF sampling for endogenous opioids might have
been more useful to validate the ability of CSF
sampling in detecting endogenous opioid changes
rather than to confirm our results.
Our findings suggest that MCS triggers endog-
enous opioid secretion in part of the remaining
structures of the medial pain system, not affected
by opioid receptor loss in CPSP. This compensa-
tion by inducing endogenous secretion in the rela-
tively spared part of the opioid system may
explain one of the mechanisms of action of MCS.
Efferent projections from precentral and pre-
motor cortices toward the midbrain have been
documented in nonhuman primates.34 Specifi-
cally, the PAG receives projections from the arm
and leg areas of the primary motor cortex as well
as from the dorsal premotor area,34 and is exten-
sively interconnected with the middle and ante-
rior cingulate gyrus.35 Although these connections
may represent the anatomic substrate of the acti-
vation of PAG after MCS, the question of
whether they are essential for MCS clinical effi-
cacy remains open. Our group has recently shown
that relatively long-lasting blood flow changes in
anterior cingulate and PAG are mutually corre-
lated after MCS and may predict the clinical effi-
cacy of the procedure.13 It is therefore conceivable
that the local metabolic increase indexed by re-
gional blood flood changes may underlie the
modifications of opioid secretion in aMCC and
PAG that have been shown in the present study.
Received December 15, 2006. Accepted in final form March
23, 2007.
REFERENCES
1. Finnerup NB, Otto M, McQuay HJ, Jensen TS, Sin-
drup SH. Algorithm for neuropathic pain treatment: an
evidence based proposal. Pain 2005;118:289–305.
2. Lindblom UF, Ottosson JO. Influence of pyramidal
stimulation upon the relay of coarse cutaneous affer-
ents in the dorsal horn. Acta Physiol Scand 1957;38:
309–318.
3. Fields HL, Adams JE. Pain after cortical injury by elec-
trical stimulation of the internal capsule. Brain 1974;
97:169–178.
4. Tsubokawa T, Katayama Y, Yamamoto T, Hirayama
T, Koyama S. Chronic motor cortex stimulation for
the treatment of central pain. Acta Neurochir Suppl
(Wien) 1991;52:13–139.
5. Mertens P, Nuti C, Sindou M, et al. Precentral cortex
stimulation for the treatment of central neuropathic
pain: results of a prospective study in a 20-patient se-
ries. Stereotact Funct Neurosurg 1999;73:122–125.
6. Saitoh Y, Shibata Y, Mashimo T. Motor cortex stimu-
lation for phantom limb pain. Lancet 1999;353:212.
7. Nuti C, Peyron R, Garcia-Larrea L, et al. Motor cortex
stimulation for refractory neuropathic pain: four year
outcome and predictors of efficacy. Pain 2005;118:43–
52.
8. Tsubokawa T, Katayama Y, Yamamoto T, Hirayama
T, Koyama S. Treatment of thalamic pain by chronic
motor cortex stimulation. Pacing Clin Electrophysiol
1991;14:131–134.
9. Garcia-Larrea L, Peyron R, Mertens P, et al. Electrical
stimulation of motor cortex for pain control: a com-
bined PET-scan and electrophysiological study. Pain
1999;83:259–273.
10. Garcia-Larrea L, Peyron R, Mertens P, Laurent B,
Mauguiere F, Sindou M. Functional imaging and neu-
rophysiological assessment of spinal and brain thera-
Neurology 69 August 28, 2007 833
 peutic modulation in humans. Arch Med Res 2000;31:
248–257.
11. Senapati AK, Huntington PJ, Peng YB. Spinal dorsal
horn neuron response to mechanical stimuli is de-
creased by electrical stimulation of the primary motor
cortex. Brain Res 2005;1036:173–179.
12. Peyron R, Garcia-Larrea L, Deiber MP, et al. Electrical
stimulation of precentral cortical area in the treatment
of central pain: electrophysiological and PET study.
Pain 1995;62:275–286.
13. Peyron R, Faillenot I, Mertens P, Laurent B, Garcia-
Larrea L. Motor cortex stimulation in neuropathic
pain: correlations between analgesic effect and hemo-
dynamic changes in the brain. A PET study. Neuroim-
age 2007;34:310–321.
14. Jones AKP, Liyi QI, Fujiwara T, et al. In vivo distribu-
tion of opioid receptors in man in relation to the corti-
cal projections of the medial and lateral pain systems
measured with positron emission tomography. Neuro-
sci Lett 1991;126:25–28.
15. Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo
and opioid analgesia: imaging a shared neuronal net-
work. Science 2002;295:1737–1740.
16. Wood CC, Spencer DD, Allison T, McCarthy G, Wil-
liamson PD, Goff WR. Localization of human sensori-
motor cortex during surgery by cortical surface
recording of somatosensory evoked potentials. J Neu-
rosurg 1988;68:99–111.
17. Luthra SK, Brady F, Turton DR, et al. Automated ra-
diosyntheses of [6-0-methyl-11C] diprenorphine and
[6-0-methyl-11C] buprenorphine from 3-06trityl pro-
tected precursors. Appl Radiat Isot 1994;45:857–873.
18. Maarrawi J, Peyron R, Mertens P, et al. Differential
brain opioid receptor availability in central and periph-
eral neuropathic pain. Pain 2007;127:183–194.
19. Frost JJ, Douglass KH, Mayberg HS, et al. Multicom-
partmental analysis of [11C]carfentanil binding to opi-
ate receptors in humans measured by positron emission
tomography. J Cereb Blood Flow Metab 1989;9:398–
409.
20. Willoch F, Schindler F, Wester HJ, et al. Central post-
stroke pain and reduced opioid receptor binding within
pain processing circuitries: a [11C]diprenorphine PET
study. Pain 2004;108:213–220.
21. Peyron R, Garcia-Larrea L, Gregoire MC, et al. Allo-
dynia after lateral-medullary (Wallenberg) infarct: a
PET study. Brain 1998;121 (pt 2):345–356.
22. Millan MJ, Millan MH, Czlonkowski A, et al. A model
of chronic pain in the rat: response of multiple opioid
834 Neurology 69 August 28, 2007
systems to adjuvant-induced arthritis. J Neurosci 1986;
6:899–906.
23. Yonehara N, Kudo T, Iwatsubo K, Maeda S, Saito k,
Inoki R. A possible involvement of the central endor-
phin system in the autoanalgesia induced by chronic
administration of Freund’s adjuvant solution in rats.
Pain 1983;17:91–98.
24. Zubieta JK, Dannals RF, Frost JJ. Gender and age in-
fluences on human brain mu-opioid receptor binding
measured by PET. Am J Psychiatry 1999;156:842–848.
25. Fields H. State-dependent opioid control of pain. Nat
Rev Neurosci 2004;5:565–575.
26. Jones AKP, Watabe H, Cunningham VJ, Jones T. Ce-
rebral decreases in opioid receptor binding in patients
with central neuropathic pain measured by [11C]di-
prenorphine binding and PET. Eur J Pain 2004;8:479–
485.
27. Sprenger T, Berthele A, Platzer S, Boecker H, Tolle
TR. What to learn from in vivo opioidergic brain imag-
ing? Eur J Pain 2005;9:117–121.
28. Horner KA, Zadina JE. Internalization and down-
regulation of mu opioid receptors by endomorphins
and morphine in SH-SY5Y human neuroblastoma
cells. Brain Res 2004;1028:121–132.
29. Almay BGL, Johansen F, Knorring L, Terenius L,
Wahlstrom A. Endorphins in chronic pain: I. Differ-
ences in CSF endorphin levels between organic and
psychogenic pain syndromes. Pain 1978;5:153–162.
30. Clement-Jones V, Tomlin S, Rees LH, McLoughlin L,
Besser GM, Wen HL. Increased ␤-endorphin but not
met-enkephalin levels in human cerebrospinal fluid af-
ter acupuncture for recurrent pain. Lancet 1980;ii:946–
948.
31. Han JS, Chen XH, Sun SL, et al. Effect of low- and
high-frequency TENS on Met-enkephalin-Arg-Phe and
dynorphin A immunoreactivity in human lumbar CSF.
Pain 1991;47:295–298.
32. Chan AW, McPharlain IA, Masson EA, Bonsher D,
Morley JS, Venn RF, Bowsher D. CSF enkephalins in
diabetic neuropathy. Neuropeptides 1992;22:125–128.
33. Dehen H, Amsallem B, Colas-Linhart N. Cerebrospi-
nal fluid beta-endorphin in chronic pain in man [in
French]. Rev Neurol (Paris) 1990;146:155–157.
34. Hartmann-von Monakow K, Akert K, Künzle H. Pro-
jections of precentral and premotor cortex to the red
nucleus and other midbrain areas in Macaca fascicu-
laris. Exp Brain Res 1979;34:91–105.
35. Carmichael ST, Price JL. Limbic connections of the or-
bital and medial prefrontal cortex in macaque mon-
keys. J Comp Neurol 1995;363:615–641.