J Neurosurg 92:150–155, 2000

Motor cortex stimulation for central and peripheral

deafferentation pain

Report of eight cases

YOUICHI SAITOH, M.D., PH.D., MASAHIKO SHIBATA, M.D., PH.D.,

SHUN-ICHIRO HIRANO, M.D., PH.D., MASAYUKI HIRATA, M.D.,
TAKASHI MASHIMO, M.D., PH.D., AND TOSHIKI YOSHIMINE, M.D., PH.D.
Departments of Neurosurgery and Anesthesiology, Osaka University Graduate School of Medicine,
Osaka, Japan

 The authors tested a modified motor cortex stimulation protocol for treatment of central and peripheral types of deaf-
ferentation pain. Four patients with thalamic pain and four with peripheral deafferentation pain were studied. Preoperative
pharmacological tests of pain relief were performed using phentolamine, lidocaine, ketamine, thiopental, and placebo. In
five patients we placed a 20- or 40-electrode grid in the subdural space to determine the best stimulation point for pain re-
lief for a few weeks before definitive placement of a four-electrode array. In three patients, the four-electrode array was
implanted in the interhemispheric fissure as a one-stage procedure to treat lower-extremity pain. In two patients with pain
extending from the extremity to the trunk or hip, dual devices were implanted to drive two electrodes.
Six of eight patients experienced pain reduction (two each with excellent, good, and fair relief) from motor cortex stimu-
lation. No correlation was apparent between pharmacological test results and the effectiveness of motor cortex stimulation.
Patients with peripheral deafferentation pain, including two with phantom-limb pain and two with brachial plexus injury,
attained pain relief from motor cortex stimulation, with excellent results in two cases. Testing performed with a subdural
multiple-electrode grid was helpful in locating the best stimulation point for pain relief. Motor cortex stimulation may be
effective for treating peripheral as well as central deafferentation pain.

KEY WORDS • thalamic pain • phantom-limb pain • motor cortex stimulation

pain, such as thalamic pain, sec- stimulated cerebral hemisphere is atrophic, which requires
D
EAFFERENTATION
ondary to central nervous system lesions is the a relatively high voltage for pain relief. We evaluated a
most difficult pain syndrome to control with med- different protocol: after induction of general anesthesia a
ication.23 According to reports, thalamic stimulation re- 20-electrode grid was placed in the subdural space over-
lieves deafferentation pain in some cases, but electrode lying the sensorimotor cortex. Various stimulation pat-
implantation is a complicated procedure and the rate of terns were tested for a few weeks, after which patients
effectiveness has been unsatisfactory.7,10,11,25 Katayama, et who experienced pain relief underwent implantation of a
al.,9 have reported that an alternative treatment, long-term smaller, four-electrode array (0.5 cm in diameter, each
motor cortex stimulation, consistently offers relief to ap- separated by 0.5 cm) at the position determined empirical-
proximately 50% of patients with thalamic pain. Mey- ly to be most effective.
erson, et al.,15 have shown that facial deafferentation pain We inferred from previously reported preoperative
also can be decreased by motor cortex stimulation. How- pharmacological tests of pain relief that thiopental- and
ever, little is known about the effectiveness of motor ketamine-responsive and morphine-resistant patients re-
cortex stimulation for treating peripheral deafferentation ceive lasting benefit from long-term motor cortex stimu-
pain.20 lation.27 We also examined correlations between pharma-
In several previous reports on this technique, the opti- cological test results and the efficacy of stimulation by
mum stimulation point was determined using test stimula- using a somewhat different protocol.
tion after induction of local anesthesia.9,17,24 However, lim-
ited operating time can interfere with the usefulness of this
method for determining the point most effective for pain Patients and Methods
reduction. In these studies, electrodes were usually placed Patient Population
in the epidural space, and this can be a drawback when the
Between 1996 and 1998, eight patients with either tha-
Abbreviations used in this paper: DREZ = dorsal root entry zone; lamic or peripheral deafferentation pain were treated with
MR = magnetic resonance; SEP = sensory evoked potential; SSS = cortical stimulation at Osaka University Medical Hospital
superior sagittal sinus; VAS = visual analog scale. (Table 1). The patients were all men whose ages ranged

150 J. Neurosurg. / Volume 92 / January, 2000

Motor cortex stimulation for deafferentation pain

TABLE 1
Clinical data in eight patients with deafferentation pain*
Age Effective Drugs Results of
Case (yrs), on Pharmaco- Test Stim- Outcome
No. Sex Underlying Disease Previous Treatment† logical Testing ulation (mos)

1 68, M lt thalamic hemorrhage medication thiopental poor —

2 60, M lt putaminal hemorrhage medication lidocaine, ketamine, good good (26)
fentanyl
3 52, M rt thalamic hemorrhage medication, block none poor —
4 68, M rt thalamic hemorrhage medication none fair fair (12)
5 56, M lt brachial plexus avulsion medication, DREZ le- morphine, ketamine, good fair (15)
sioning, block, DCS thiopental
6 62, M rt phantom-limb pain medication, DCS, block morphine, ketamine, excellent good (20)
thiopental
7 64, M lt brachial plexus avulsion medication, DCS, block none excellent excellent (19)
8 53, M rt phantom-limb & stump medication, DCS, block morphine fair fair (6)
pain
* DCS = dorsal column stimulation; — = no reduction on stimulation.
† Block refers to sympathetic ganglion block.

from 52 to 68 years; three presented with thalamic hem- stimulation of median and/or posterior tibial nerves at the
orrhage, one with putaminal hemorrhage, two with brachi- time of implantation of the electrode grid. Phase reversal
al plexus injuries, and two with phantom-limb pain. In the of N20 or N40 waves was used to confirm that grid elec-
patients with thalamic pain, small thalamic hemorrhages trodes were correctly placed over the sensorimotor cor-
were clearly identifiable on MR imaging or computerized tex.26 We also assessed SEPs in all patients before surgery.
tomography scanning. The interval between the causative The patient in Case 3 showed no difference between af-
neurological event and the onset of pain ranged from 1 to fected and healthy sides. The patients in Cases 5 and 7,
4 years. The patients had been treated with various med- who had brachial plexus avulsion, and those in Cases 6
ications including anticonvulsant and/or antidepressant and 8, who had phantom-limb pain, showed no SEPs on
drugs. stimulation of affected extremities.
The four patients with central pain also exhibited vary-
ing degrees of hemiparesis. These patients reported in- Pain Assessment
tense spontaneous pain with a burning or tearing quality
occurring in the extremities or trunk. Various degrees of Changes in pain level were evaluated in each patient by
analgesia to pinprick were demonstrable in patients with a physician who worked at a pain clinic unaffiliated with
thalamic pain, and two patients with brachial plexus injury our department. More than 20 different stimulatory pat-
exhibited such analgesia while experiencing burning pain terns were evaluated using the grid electrodes to deter-
in the affected upper extremities. No patients with severe- mine the best stimulation point for pain relief. Each pa-
ly depressive or neurotic responses as demonstrated on a tient was asked to describe the pain intensity according to
psychological assessment underwent therapeutic trial of a VAS and the McGill Pain Questionnaire.12 Effects of
cortical stimulation. stimulation were classified into four categories: excellent,
reduction of pain level by 80 to 100%; good, 60 to 79%
Pharmacological Tests reduction; fair, 40 to 59% reduction; and poor, less than
40% reduction. The pain level was evaluated after stimu-
To clarify pathophysiological mechanisms in these pa- lation for 30 minutes no more than three times per day
tients, we performed pharmacological tests of pain relief during the period of test stimulation.
(Fig. 1), including administration of phentolamine (0.17
mg/kg/hour for 3 hours); lidocaine (initial dose 0.5 mg/
kg; then 1 mg/kg/hour continuously administered for 3
hours); ketamine (0.17 mg/kg/hour for 3 hours); thiopen-
tal (0.5 mg/kg, repeated 30 minutes after the first admin-
istration); and placebo (saline solution, initial dose 5 ml;
then 0.1 ml/kg/hour for 3 hours). In the patient in Case 2,
morphine was contraindicated and fentanyl citrate was
substituted. We evaluated the analgesic effects of these
drugs on ongoing pain by using a VAS hourly for 6 hours.
For the thiopental test, assessment with the VAS was per-
formed before and soon after administration of the drug.
Evoked Potential Testing
A specialized instrument (Neuropack 8; Nihon Koh-
den Co., Ltd., Tokyo, Japan) was used to evaluate SEPs on FIG. 1. Graphs showing pharmacological evaluation in Case 6.

J. Neurosurg. / Volume 92 / January, 2000 151

 Y. Saitoh, et al.

FIG. 2. Case 8. Plain radiographs illustrating placement of the

four-electrode arrays driven by a wireless stimulation system. One
electrode is positioned in the interhemispheric fissure and the oth-
er beside the SSS. Both phantom-limb and stump pain were re-
duced by stimulation of both electrodes (VAS score reduction from
10 to 6).

Surgical Procedures FIG. 3. Case 4. Diagram showing stimulation points on a 40-

The location of the central sulcus was approximated electrode grid and a four-electrode array. The central sulcus was
according to preoperative MR imaging studies. After in- localized by N20 phase reversal (broken line). Broad arrows in-
duction of general anesthesia, a skin incision was made in dicate the most effective position for pain relief (VAS score reduc-
a location corresponding to the site of pain. Craniotomy tion from 10 to 5); narrow arrows indicate slightly effective sites
(reduction from 10 to 7 or 8); and broken arrows indicate minimal
was performed over a 5  6–cm area overlying the senso- effectiveness.
rimotor cortex, except that in three patients (Cases 2, 6,
and 8) suffering from leg pain, craniotomy was performed
overlying the SSS. In four patients a 20-electrode grid was control in test stimulations (Table 1), but in the other
placed subdurally (four-by-five arrangement; 0.5-cm elec- two patients, more than 30 different patterns of stimula-
trode diameter; 0.5-cm separation); in an additional pa- tion on the 20-electrode grid were tried without successful
tient with thalamic pain (Case 4) a 40-electrode grid was reduction of pain. When electrode placement was in the
placed. The three patients with leg pain underwent place- interhemispheric fissure, possible electrode positions were
ment of a four-electrode array in the interhemispheric limited. More than 20 different stimulatory patterns were
fissure. Electrodes were placed in the interhemispheric evaluated in three patients (Cases 4, 5, and 7) by using the
fissure from the parietal or prefrontal directions, accord- electrode grid to determine the best point for pain relief
ing to the development of cortical veins. One of the pa- (Figs. 3–5). In each patient who had a good response, sev-
tients (Case 8) received two four-electrode arrays (one in eral stimulation points were effective, delivering pain re-
the interhemispheric fissure and the other beside the SSS; lief even with stimulation of the sensory cortex. However,
Fig. 2). For the 20-electrode grid, locations of precentral the position judged best for pain relief in each of these pa-
and postcentral gyri were confirmed from phase reversal tients consisted of motor cortex alone, with electrical po-
of the N20 component with median nerve stimulation or larity almost perpendicular to the central sulcus and the
the N40 component with tibial nerve stimulation. current directed from parietal to frontal areas. In Cases 2,
After implantation of the electrode grid, electrical stim- 6, and 8, four-electrode arrays were positioned in proxim-
uli were delivered to various areas no more than three ity to the motor cortex that corresponded to the affected
times per day. Stimuli consisted of monophasic square- lower extremities.
wave pulses lasting 0.2 msec. The frequency (25–50 Hz) In Case 5 an epidural electrode was placed, which
and voltage (1.5–5 V) that produced the best pain relief provided a decreasing level of relief. After approximately
were determined. Stimulation usually was applied contin- 6 months, the electrode was repositioned in the subdural
uously for 30 minutes on each occasion. When some pa- space just underlying the previous location, and its origi-
tients developed seizures on stimulation, the stimulatory nal effectiveness returned at a lower voltage. It was found
voltage was decreased to a level lower than the threshold that granulation tissue had proliferated under the epidural
for muscle contraction. electrode. The patient in Case 4 suffered a subdural effu-
Long-term stimulation was performed after definitive sion after implantation of a four-electrode array.
electrode implantation by using a wireless stimulation The patients in Cases 3 and 4 (thalamic pain) and 7
system (X-trel model 3425, Mattrix model 3210; Med- (brachial plexus avulsion) showed no response to pharma-
tronic, Inc., Minneapolis, MN). cological tests, but in Cases 4 and 7 the patients attained
pain reduction with stimulation, with the latter in particu-
lar experiencing excellent pain relief. On the other hand,
Results the patient in Case 1 experienced pain relief after thiopen-
Six of eight patients showed various degrees of pain tal injection, but he failed to respond to stimulation. The

152 J. Neurosurg. / Volume 92 / January, 2000

Motor cortex stimulation for deafferentation pain

FIG. 4. Case 5. Diagram showing stimulation points on a 20-

electrode grid. The central sulcus was localized by N20 phase re- FIG. 5. Case 7. Diagram showing stimulation points on a 20-
versal (broken line). Broad arrow indicates the most effective posi- electrode grid. The central sulcus was localized by N20 phase re-
tion for pain relief (VAS score reduction from 10 to 2); narrow versal (broken line). Broad arrow indicates the most effective po-
arrows indicate slightly effective sites (reduction from 10 to 5 or sition for pain relief (VAS score reduction from 8 to 1); and narrow
6); and broken arrows indicate minimal effectiveness. arrows indicate slightly effective sites (reduction from 8 to 3).

patient in Case 6 (phantom-limb pain) evidenced remark- stage procedure including a protocol in which a 20-elec-
able analgesic effects with ketamine, thiopental, and mor- trode grid was typically placed over the sensorimotor cor-
phine; in Case 8 (phantom-limb and stump pain) the tex to test various geometric patterns of stimulation in the
patient experienced excellent pain relief after administra- course of a few weeks to determine the optimum pattern
tion of morphine; in Case 2 pain relief was attained with for pain relief. The second stage was implantation of a
lidocaine, ketamine, fentanyl citrate, and motor cortex four-electrode array for long-term stimulation based on
stimulation; and in Case 5 pain improvement was seen the results of the first stage.
with administration of morphine, ketamine, thiopental, A number of questions remain concerning mechanisms,
and motor cortex stimulation. In these patients, no correla- indications, implantation strategy, and further technical
tion was apparent between pharmacological test responses development of motor cortex stimulation. In practice, the
and pain reduction with stimulation (Table 1). location of the central sulcus is difficult to correlate with
The patient in Case 4 reported pain extending from the landmarks on the outer surface of the skull, particularly
upper extremity to the hip, and in Case 8 there was phan- because the motor cortex has a variable extent and con-
tom-limb and stump pain, a distribution difficult to treat figuration. Preoperative scalp or epidural recording of
completely with a single electrode. Mattrix dual-electrode evoked responses to identify the site of N20 wave rever-
systems (model 3210; Medtronic, Inc.) were placed after sal is often inaccurate in localizing the motor cortex.14 We
confirmation of the most effective stimulation points, with therefore estimated the location of the central sulcus based
reduction of pain (Fig. 2). on preoperative MR imaging and intraoperative SEP as-
The patients in Cases 6 and 7 experienced pain relief for sessments. Some differences were noted between the mor-
24 hours after 30 minutes’ stimulation. On the other hand,
in Cases 5 and 8 the pain returned within 1 hour after the
electrical stimulation, and in Cases 2 and 4 pain relief was
obtained for 3 to 5 hours. The results of VAS score chang-
es are summarized in Fig. 6. There was no correlation be-
tween the duration of pain relief and motor cortex stimu-
lation and pharmacological tests. The participants were
followed on an outpatient basis for 6 to 26 months. The
patient in Case 6 reported reduced effect of the stimulation
after 12 months (Table 1).

Discussion
In previous reports implantation of a four-electrode
array over the precentral motor cortex has been de-
scribed.1,2,4,6,9,16,18,24 Such an approach might not provide
optimum pain relief, because both the methods and the ar-
ea of test stimulation were restricted by a brief operating
time and use of a local anesthetic. Even sensory cortex
stimulation has been described as effective in some cases FIG. 6. Graph showing summary of the changes of VAS score
under such circumstances.24 We therefore chose a two- with motor cortex stimulation.

J. Neurosurg. / Volume 92 / January, 2000 153

 Y. Saitoh, et al.

phologically identified central sulcus on MR images and Our results indicated that motor cortex stimulation can
the site of N20 wave reversal. These discrepancies under- be effective for peripheral as well as for central deafferen-
score the need for test stimulation by means of an elec- tation pain. Although motor cortex stimulation has been
trode grid before long-term electrode placement. reported to be effective for facial deafferentation pain,15,17
The optimum placement and orientation of a four-elec- this treatment has not been studied for phantom-limb pain
trode array in relation to the motor cortex has not yet been or brachial plexus avulsion.20 Phantom-limb pain is a very
established.17 In our test results, stimulation of the pre- rare symptom for which several medical3,8,21,22 and surgi-
frontal cortex with the grid electrode failed to reduce pain cal19 approaches have been pursued. Postcentral stimula-
in all five cases, whereas stimulation of the sensory cortex tion of appropriate areas can actually exacerbate phantom-
seemed to reduce pain somewhat. In all six patients in limb pain in some patients. Even with excision of cortical
whom treatment by stimulation was successful, the best or thalamic areas, phantom-limb pain tends to recur with
position for pain relief was limited to the motor cortex, time.5 Dorsal root entry zone lesioning is reportedly effec-
and currents were directed perpendicular to the central tive in 60 to 70% of brachial plexus avulsions.4 However,
sulcus in a parietal-to-frontal direction with electrode motor cortex stimulation is less invasive than DREZ le-
grids. The polarity difference we found is not in agree- sioning. Note that the patient in Case 5 did not respond to
ment with Tsubokawa, et al.,24 who reported no polarity- DREZ lesioning but responded well to stimulation.
related differences in pain relief for most patients. In our Motor cortex stimulation was effective in approximate-
series, the duration of pain relief after 30 minutes’ stimu- ly half of patients with central deafferentation pain in the
lation ranged from 1 hour to 24 hours (Fig. 6). It was dif- largest published series9 and in many other reports.1,2,16,18
ficult to explain the variability of the effective duration Katayama, et al.,9 have speculated that the pain control af-
in relation to several factors, including the results of forded by motor cortex stimulation requires neuronal cir-
pharmacological tests, underlying diseases, and stimula- cuits maintained by the presence of intact corticospinal
tion point. neurons originating from the motor cortex. In our patients
The results of the VAS were used in our report to eval- with central deafferentation pain (four cases), Grade 3 or
uate the effect of stimulation. In the test stimulation in 4 weakness was present in affected extremities, and two
which the grid electrodes were used, patients were repeat- of these patients failed to respond to stimulation, most no-
edly and randomly asked to assess the VAS at each stim- tably Case 1.
ulation of an electrode, and the results were mostly consis- In our study, we could not clarify the effective mecha-
tent. Our patients also gauged whether the pain was better nism of motor cortex stimulation for central and peripher-
or worse than before stimulation and described the inten- al deafferentation pain. It was thought that activation of
sity and quality of the pain according to the McGill Pain hypothetical sensory neurons by means of motor cortex
Questionnaire12 before and after the stimulation. It ap- stimulation might inhibit deafferentation nociceptive neu-
peared that the VAS score changes for each patient were rons within the cortex in patients with central deafferenta-
reliable (Fig. 6). tion pain.24 The mechanism of peripheral deafferentation
In several reports epidural placement of electrodes for pain such as phantom-limb pain is unknown; however,
motor cortex stimulation has been favored to minimize both hyperactivity of peripheral nerves and sensitization
invasiveness.6,9,17,24 However, in the patient in Case 5, his of spinal neurons may play a part.23 On the other hand,
epidural electrode had become diminished in effective- Melzack13 proposed that the anatomical structure of the
ness for almost 6 months, and when the electrode was body is represented in a neuromatrix extending through-
moved to the subdural space in a second operation, epidu- out the brain, which is genetically determined and later
ral granulation tissue was found beneath the device. After modified by sensory inputs. According to this author,
nerve deafferentation can eventually result in neuron dam-
the electrode was moved, pain reduction was accom- age and pain generated in the brain. Therefore, in patients
plished at a lower voltage. One patient’s surgery was com- with peripheral deafferentation pain, motor cortex stimu-
plicated by subdural effusion, but no other complications lation may also inhibit deafferentation nociceptive neu-
were noted. rons within the cortex.
Yamamoto, et al.,27 have reported that thiopental- and
ketamine-responsive and morphine-resistant patients dis-
played long-lasting pain reduction with long-term use of Conclusions
motor cortex stimulation. Unfortunately, we failed to find
clear correlation between pharmacological test results and For patients with central or peripheral deafferentation
effectiveness of motor cortex stimulation. There were rel- pain, we currently recommend subdural placement of an
atively few patients in our study, but motor cortex stimu- electrode grid overlying the sensorimotor cortex to deter-
lation resulted in effective pain relief in Cases 4 and 7 mine the best site for pain relief by a period of test stimu-
even though no response to pharmacological agents was lation. We have found pharmacological tests to be unre-
seen. On the other hand, the patient in Case 1 obtained liable.
poor pain reduction with stimulation even though thiopen-
tal injection had provided relief. Based on these results, References
predicting the effectiveness of motor cortex stimulation by 1. Canavero S, Bonicalzi V: Cortical stimulation for central pain.
using pharmacological tests may be difficult. We also sub- J Neurosurg 83:1117–1118, 1995 (Letter)
stantially modified the pharmacological tests from those 2. Ebel H, Rust D, Tronnier V, et al: Chronic precentral stimu-
used in the earlier protocol and believe that more uniform lation in trigeminal neuropathic pain. Acta Neurochir 138:
testing will be needed in the future. 1300–1306, 1996

154 J. Neurosurg. / Volume 92 / January, 2000

Motor cortex stimulation for deafferentation pain

3. Elliott F, Little A, Milbrandt W: Carbamazepine for phantom- 18. Peyron R, Garcia-Larrea L, Deiber MP, et al: Electrical stimu-
limb phenomena. N Engl J Med 295:678, 1976 (Letter) lation of precentral cortical area in the treatment of central pain:
4. Gybels JM: Indications for neurosurgical treatment of chronic electrophysiological and PET study. Pain 62:275–286, 1995
pain. Acta Neurochir 116:171–175, 1992 19. Raj PP (ed): Practical Management of Pain, ed 2. St. Louis:
5. Horrax G: Experiences with cortical excisions for the relief of Year Book Medical, 1992
intractable pain in the extremities. Surgery 20:593–602, 1946 20. Saitoh Y, Shibata M, Sanada H, et al: Motor cortex stimulation
6. Hosobuchi Y: Motor cortical stimulation for control of central for phantom limb pain. Lancet 353:212, 1999
deafferentation pain. Adv Neurol 63:215–217, 1993 21. Smith PP, Turkington D: Fluoxetine in phantom limb pain. Br
7. Hosobuchi Y: Subcortical electrical stimulation for control of J Psychiatry 163:105–106, 1993
intractable pain in humans. Report of 122 cases (1970–1984). J 22. Stannard CF, Porter GE: Ketamine hydrochloride in the treat-
Neurosurg 64:543–553, 1986 ment of phantom limb pain. Pain 54:227–230, 1993
8. Jaeger H, Maier C: Calcitonin in phantom limb pain: a double- 23. Tasker RR: Deafferentation, in Wall PD, Melzack R (eds):
blind study. Pain 48:21–27, 1992 Textbook of Pain. Edinburgh: Churchill Livingstone, 1984, pp
9. Katayama Y, Fukaya C, Yamamoto T: Poststroke pain control 119–132
by chronic motor cortex stimulation: neurological characteris- 24. Tsubokawa T, Katayama Y, Yamamoto T, et al: Chronic motor
tics predicting a favorable response. J Neurosurg 89:585–591, cortex stimulation in patients with thalamic pain. J Neurosurg
1998 78:393–401, 1993
10. Kumar K, Toth C, Nath RK: Deep brain stimulation for intrac- 25. Tsubokawa T, Katayama Y, Yamamoto T, et al: Deafferen-
table pain: a 15-year experience. Neurosurgery 40:736–747, tation pain and stimulation of the thalamic sensory relay nu-
1997 cleus: clinical and experimental study. Appl Neurophysiol 48:
11. Mazars GJ: Intermittent stimulation of nucleus ventralis pos- 166–171, 1985
terolateralis for intractable pain. Surg Neurol 4:93–95, 1975 26. Wood CC, Spencer DD, Allison T, et al: Localization of hu-
12. Melzack R: The McGill Pain Questionnaire: major properties man sensorimotor cortex during surgery by cortical surface re-
and scoring methods. Pain 1:277–299, 1975 cording of somatosensory evoked potentials. J Neurosurg 68:
13. Melzack R: Phantom limbs and the concept of a neuromatrix. 99–111, 1988
Trends Neurosci 13:88–92, 1990 27. Yamamoto T, Katayama Y, Hirayama T, et al: Pharmacological
14. Meyerson BA: Pharmacological tests in pain analysis and in classification of central post-stroke pain: comparison with the
prediction of treatment outcome. Pain 72:1–3, 1997 results of chronic motor cortex stimulation therapy. Pain 72:
15. Meyerson BA, Lindblom U, Linderoth B, et al: Motor cortex 5–12, 1997
stimulation as treatment of trigeminal neuropathic pain. Acta
Neurochir Suppl 58:150–153, 1993
16. Migita K, Uozumi T, Arita K, et al: Transcranial magnetic coil Manuscript received February 26, 1999.
stimulation of motor cortex in patients with central pain. Neu- Accepted in final form August 25, 1999.
rosurgery 36:1037–1040, 1995 Address reprint requests to: Youichi Saitoh, M.D., Ph.D., De-
17. Nguyen JP, Keravel Y, Feve A, et al: Treatment of deafferenta- partment of Neurosurgery, Osaka University Graduate School of
tion pain by chronic stimulation of the motor cortex: report of a Medicine, 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan. email:
series of 20 cases. Acta Neurochir 68:54–60, 1997 saitoh@silver.ocn.ne.jp.

J. Neurosurg. / Volume 92 / January, 2000 155