© 2003 U.S. Cancer Pain Relief Committee 0885-3924/03/$–see front matterPublished by Elsevier. All rights reserved. doi:10.1016/S0885-3924(02)00689-9

Vol. 25 No. 2 February 2003 Journal of Pain and Symptom Management 179

Clinical Note

Is There Still a Role for Open Cordotomyin Cancer Pain Management?

Bethan Jones, MRCP, Ilora Finlay, FRCP, FRCGP, Amit Ray, FRCS, and Brian Simpson, MD, FRCS

Departments of Palliative Medicine (B.J., I.F.) and Neurosurgery (A.R., B.S.), University Hospital of Wales, Cardiff, Wales, United Kingdom

Abstract

For a small number of cancer patients, good pain control remains difficult to achieve despite adequate assessment and medical management. In nine cases, effective control of intractable pain from malignant pelvic disease was achieved by open thoracic cordotomy. The technique was well tolerated, with no major complications. Eight of the nine patients decreased their median daily oral morphine requirement from 560 mg (range 360–2600 mg) to 160 mg (range 40–1000 mg). Maximal survival time post-cordotomy was 830 days, with a median of 107 days. No patient experienced recurrent pain in the initially painful site. For patients with intractable pain associated with advanced pelvic malignancy, the use of an open cordotomy should be considered when satisfactory pain control is not achieved by medical or minimally invasive methods.

J Pain Symptom Manage 2003;25:179–184

© 2003 U.S. Cancer Pain Relief Committee. Published by Elsevier. All rights reserved.

Key Words

Open cordotomy, cancer pain

Introduction

Pain is experienced by up to 75% of patientswith advanced cancer, with 25–30% reportingthis pain as very severe or excruciating.

1

The ex-tensive pharmacopoeia of analgesics now avail-able means that the majority of both neuro-pathic and nociceptive malignant pains arecontrollable.

2

For a small number of cancer pa-tients, however, referral for anesthetic or neuro-surgical procedures for pain management is stillindicated. Excellent local liaison in Cardiff hasfacilitated the use of open cordotomy for pa-

tients with lower quadrant cancer pain. In thisreport, we review our last 10 years’ experience.

Procedures for pain control can be classifiedinto three main categories:

• Neuroablative techniques, historically thefirst surgical attempts at pain control, inter-rupt pain transmission fibers and includesuch techniques as cordotomy, rhizotomy,myelotomy, and dorsal root entry zone lesion(Figure 1).

• Neurostimulatory procedures stimulate se-lected regions of the peripheral or centralnervous system via implanted electrodes.

• Neuropharmacological delivery systemsdeliver analgesic drugs (usually opioids)directly to the central nervous system, usu-ally via the epidural, intrathecal or intra-ventricular spaces.

3

Address reprint requests to:

Bethan Jones, MD, Depart-ment of Palliative Medicine, Velindre Hospital,Whitchurch, Cardiff CF14 2TL, United Kingdom.

Accepted for publication: March 20, 2002.

180 Jones et al. Vol. 25 No. 2 February 2003

Cordotomy involves dividing the anterolateralquadrant of the spinal cord, which contains thespinothalamic tract. This results in a selective lossof temperature and pain perception on the con-tralateral side, up to a level several segments be-low the level of the lesion. The first open cordot-omy was performed by Spiller and Martin in1912.

5

The initial results were disappointing. By1920, Frazier

4

had performed the operation suc-cessfully on six patients and outlined the openthoracic surgical technique. The procedure hascontinued to undergo a number of modifica-tions, including the development of a percutane-ous technique by Mullan et al.

6

in 1965, which in-volves the lesioning by direct current of thespinothalamic tract in the high cervical region.

Cordotomy should be considered when allmedical and minimally invasive methods, such asepidural or spinal analgesia, have failed to ade-quately control pain, the severity of the pain war-rants surgery, and there are no medical contrain-dications to the procedure. It is useful againstpain that is transmitted via spinal cord pathways,is unilateral, and is relatively well localized. Uni-lateral lower extremity pain that is due to directplexus invasion by rectal or gynecological cancersis often relieved by cordotomy.

4

Neuropathic pain is probably caused by arange of neurophysiological disturbances. Taskeret al.

7

proposed one model in which neuropathicpain had at least three components; a steady,often dysesthetic causalgia; a neuralgia with anintermittent shooting pain; and an evoked pain,including allodynia and hyperpathia. Cordotomyshould relieve neuralgic or evoked pain in thedistribution of damaged sensory fibers, but it may

not relieve steady burning dysesthetic pain thatmay be centrally generated.

The site of pain is important in selecting pa-tients for cordotomy. There may be a progres-sive loss of analgesia for up to six segments be-low the level of the sectioning of the cord, sothat it is not uncommon to find a sensory levelat T10 or T8 after a technically adequate opencordotomy at T4–5. The highest level of persis-tently achievable analgesia with a percutaneouscordotomy at C1–2 level is C5.

8

Cordotomy should be limited to patientswith a life expectancy of less than one to twoyears. Failure after this time is increasinglycommon and may be further complicated bythe development of neuropathic “post-cordot-omy pain” and dysesthesia.

4

Technical Considerations

Cordotomy may be percutaneous or open.Although the percutaneous method has theadvantage of being minimally invasive, it canbe done only in the cervical region, generally,but not only at C1/C2. It carries the risks asso-ciated with unwanted cervical cord damage, in-cluding hemiplegia.

The major concern during percutaneous cor-dotomy is severe respiratory dysfunction. Auto-matic respiratory fibers may intermingle with thespinothalamic fibers and can be damaged duringsurgery. Bilateral C1-2 lesions therefore producea high incidence of sleep apnea.

8,9

Unilateral le-sions are likely to cause respiratory embarrass-ment only if there is a contralateral pulmonaryproblem; in a patient with lung cancer and bra-chial plexopathy requiring pain control, thepulmonary lesion may be a contraindication.

8

Similar considerations apply to hemi-diaphragmparalysis due to phrenic nerve infiltration.

Open cordotomy is a more invasive tech-nique and carries a higher operative morbidity.However, the lesion may be tailored to suit thepatient’s needs. For example, a patient with legand hip girdle pain may have a lesion at T4/5that would adequately control the pain withoutthe risk of respiratory paralysis and without in-volving the upper limb.

Open cordotomy is performed with the pa-tient prone under general anesthesia. A laminec-tomy is performed approximately at T3-5, thedura is then opened, and the general anestheticis reversed. An anterolateral cordotomy is per-formed with a blade under direct vision, aiming

Fig. 1. Cross-section of spinal cord showing sites ofneuroblative procedures. (Reprinted from Sundare-san et al.4 with permission of Cancer.)

Vol. 25 No. 2 February 2003 Open Cordotomy and Cancer Pain Management 181

to divide the anterior half of the cord withoutdamaging the adjacent corticospinal tract, whichwould risk weakness to the ipsilateral limb.

10

Theaffected limb is tested for reduction of pinprickand pain sensation to assess completeness of thelesion and the ipsilateral limb is examined toensure no weakness is being caused. General an-esthesia is reinstated for wound closure. Thewake-up technique is usually well tolerated bythe patients and this method, we believe, opti-mizes the outcome.

Methods

To evaluate the role of cordotomy in thetreatment of cancer pain, we performed a ret-rospective case note review. Patients under thecare of the palliative care team who had under-gone cordotomy during the last 10 years wereidentified from the neurosurgical theater data-base and the records of the two local inpatientpalliative care units in South East Wales.

Patient demographics, including age andsex, cancer diagnosis, and surgical and onco-logical interventions were recorded. Data werealso collected regarding time to recurrence,site of recurrent disease and the symptomscaused, therapeutic interventions tried prior tocordotomy, opioid requirements immediatelypre-cordotomy and two weeks post-cordotomy,and survival postoperatively.

The primary outcome measure was a changein opioid requirement and coanalgesic use attwo weeks post-cordotomy. The secondary out-come measure was documented qualitativecomments, which reflected changes in activi-ties of daily living and symptoms.

Results

Ten patients were identified (4 men and 6women). One male patient’s notes had beendestroyed; the remaining 9 were analyzed. Theirmedian age was 54 years (range 30–65).

Eight of the patients had primary pelvic dis-ease; seven had a rectal or gynecological malig-nancy and one had a sarcoma of the right isch-ium (Table 1). The ninth patient had a pre-sacral recurrence of a renal cell carcinoma. Sixhad surgery at the time of diagnosis, 7 receivedchemotherapy, and 8 underwent radiotherapyas part of their initial treatment schedule. Themedian time from initial diagnosis to confir-

mation of recurrent disease, heralded by theonset of pain, was 24 months (range 0–72).

The median time from onset of pain to refer-ral for cordotomy was 18 months (range 4–27months). Five patients had unilateral lowerlimb pain of mixed nociceptive/neuropathicnature. Three had perineal and unilateral limbpain of mixed nature. The patient with the re-nal cell carcinoma had bilateral leg pain andby the time of his cordotomy was paraplegicand incontinent. Four of the non-paraplegicpatients had some weakness of the painful legand one had additional sphincter controlproblems.

Morphine dosage had been titrated in allcases until either further pain control was notgained by dose increases or intolerable side ef-fects occurred. Opioid rotation was used whenside effects were not acceptable. All patientshad tried a range of coanalgesics. Seven hadused nonsteroidal anti-inflammatory drugs(NSAIDs), 4 had received a trial of steroids,and 7 had used ketamine. Six had also receivedeither bolus injections or continuous epiduralinfusions, one patient had received pudendaland caudal nerve blocks, and one had a lum-bar sympathectomy. Transcutaneous electricalnerve stimulation had been tried by 4 patients,with a short-term benefit reported by one. Sixof the patients had received palliative chemo-therapy and 8 were given palliative radiother-apy, with pain relief as the primary intention oftreatment.

All patients underwent an open thoracic cor-dotomy performed by the same neurosurgeon(BAS). All tolerated the procedure well. Onepatient went home directly from the neurosur-gical ward and 8 were transferred back to thereferring inpatient palliative care unit prior todischarge home. All patients experienced nearcomplete pain relief in the affected lower limbas a result of their cordotomy. One of the 3 pa-tients with perineal pain continued to experi-ence significant perineal discomfort postopera-tively. Eight of the 9 patients had a decreasedmorphine requirement from a median oralmorphine daily requirement of 580 mg (range360–2600) pre-operatively to a median daily re-quirement of 160 mg (range 40–1000) twoweeks postoperatively (Table 2).

All other potent analgesics (ketamine

n

�

4,methadone

n

�

2, and fentanyl

n

�

2) werestopped postoperatively. However, one patient

182 Jones et al. Vol. 25 No. 2 February 2003

Tab

le 1

Des

crip

tion

of

Cas

es

Cas

eA

ge

(Sex

)Pr

imar

ySu

rger

y O

nco

logy

Site

of

Rec

urre

nce

Tim

e to

Rec

urre

nce

Site

of

Wor

st P

ain

Oth

er

Prob

lem

sO

pioi

ds T

ried

Co-

anal

gesi

csK

etam

ine

Epi

dura

l or

Ner

ve B

lock

Tim

e Fr

omPa

in to

C

ordo

tom

y

160

(M

)R

enal

cel

l ca

rcin

oma

No

surg

ery

Sacr

al R

TPr

e-sa

cral

18 m

onth

sL

ower

but

tock

sB

oth

legs

Para

pleg

icPo

or s

phin

cter

Mor

phin

eFe

nta

nyl

Sodi

um V

alpr

oate

NSA

IDs

No

Sin

gle

bolu

s n

o re

lief

21 m

onth

s

265

(F)

Squa

mou

s ce

ll ce

rvix

Wer

thei

m’s

h

yste

rect

omy

Ch

emo/

RT

Lef

t ilia

c w

ing

& fe

mor

al

ner

ve r

oot

60 m

onth

sL

eft f

emor

al

ner

ve

dist

ribu

tion

Mul

tipl

e pa

ins

else

wh

ere

Mor

phin

eD

iam

orph

ine

Am

itri

ptyl

ine

Dot

hie

pin

Car

bam

azep

ine

Dex

amet

has

one

Ben

zodi

azep

ine

NSA

IDs

Yes

Infu

sion

part

ial r

elie

f24

mon

ths

350

(F)

Ade

noc

arci

nom

aof

rec

tum

AP

rese

ctio

nA

djuv

ant

Ch

emo/

RT

S2/S

3 le

ft s

ide

49 m

onth

sSc

iati

c n

erve

dist

ribu

tion

Vas

cula

r ch

ange

sin

left

leg

Mor

phin

eFe

nta

nyl

Met

had

one

Am

itri

ptyl

ine

Dot

hie

pin

Clo

mip

ram

ine

NSA

IDs

No

Sin

gle

bolu

s an

din

fusi

on

som

e ef

fect

Lum

bar

sym

path

ecto

my

littl

e re

spon

se

6 m

onth

s

459

(M

)D

uke’

s C

ca

rcin

oma

of r

ectu

m

AP

rese

ctio

nA

djuv

ant

Ch

emo/

RT

Lef

t sem

inal

ve

sicl

e36

mon

ths

Lef

t per

ineu

mD

ecre

ased

po

wer

le

ft le

gN

o sp

hin

cter

co

ntr

ol

Mor

phin

eFe

nta

nyl

Met

had

one

Am

itri

ptyl

ine

Dot

hie

pin

Val

proa

teC

arba

maz

epin

eFl

ecai

nid

eN

SAID

s

Yes

Sin

gle

bolu

s an

d in

fusi

onlit

tle

impr

ovem

ent

24 m

onth

s

564

(F)

Ade

noi

d cy

stic

ca

rcin

oma

ofB

arth

olin

’s

glan

d

Exc

isio

nA

djuv

ant

chem

o

Lef

t lab

ia a

nd

loca

l in

filt

rati

on

24 m

onth

sR

igh

t vag

inal

w

all,

sacr

um a

nd

righ

t leg

Non

eM

orph

ine

Fen

tan

ylT

ram

adol

Am

itri

ptyl

ine

Dot

hie

pin

Clo

mip

ram

ine

Car

bam

azep

ine

Top

ical

lign

ocai

ne

Lig

noc

ain

e in

fusi

onM

exile

tin

e

Yes

Gan

glio

n n

erve

bloc

kPe

rin

eal i

nfi

ltra

tion

Pude

nda

l ner

ve b

lock

Cau

dal n

erve

bloc

k

27 m

onth

s

636

(M

)C

hon

dros

arco

ma

of r

igh

tis

chiu

m

No

surg

ery

Ch

emo

Rig

ht i

sch

ium

0 m

onth

sR

igh

t pel

vis

butt

ock

to

knee

Dec

reas

ed

pow

er in

ri

ght l

eg

Mor

phin

eM

eth

adon

eG

abap

enti

nM

idaz

olam

Dex

sam

eth

ason

eN

SAID

Yes

Non

e4

mon

ths

730

(F)

Ade

noc

arci

nom

aof

cer

vix

No

surg

ery

Ch

emo

Rad

ical

RT

Pres

acra

l an

dpe

lvic

re

curr

ence

6 m

onth

sR

igh

t leg

an

d h

ipM

any

soci

al

prob

lem

sM

orph

ine

Dia

mor

phin

eFe

nta

nyl

Met

had

one

Am

itri

ptyl

ine

Dot

hie

pin

Val

proa

teC

arba

maz

epin

eG

abap

enti

nD

exam

eth

ason

eN

SAID

s

Yes

Epi

dura

l in

fusi

onso

me

effe

ct18

mon

ths

838

(F)

Car

cin

oma

ofce

rvix

TA

H a

nd

BSO

Bow

el a

nd

blad

der

rese

ctio

n

Rig

ht p

elvi

c w

all

adja

cen

t to

scia

tic

not

ch

72 m

onth

sR

igh

t leg

Dec

reas

ed

pow

erin

rig

ht l

eg

Mor

phin

eD

iam

orph

ine

Am

itri

ptyl

ine

Paro

xeti

ne

Clo

naz

epam

Gab

apen

tin

Dex

amet

has

one

Yes

Non

e6

mon

ths

954

(F)

Squa

mou

sca

rcin

oma

of a

nus

AP

rese

ctio

nad

juva

nt

Ch

emo/

RT

Rig

ht p

elvi

s an

dva

gin

al w

all

24 m

onth

sPe

rin

eum

an

d ri

ght

leg

Dec

reas

edpo

wer

righ

t leg

wit

h r

igh

t fo

ot d

rop

Mor

phin

eM

eth

adon

eO

xyco

don

e

Am

itri

ptyl

ine

Val

proa

teC

lon

azep

amG

abap

enti

nD

iaze

pam

, NSA

ID

Yes

2 ep

idur

al in

fusi

ons

part

ial e

ffec

t8

mon

ths

Vol. 25 No. 2 February 2003 Open Cordotomy and Cancer Pain Management 183

required the reintroduction of ketamine five dayspostoperatively for contralateral neuropathic legpain, with good effect. One patient started metha-done in the postoperative period for neuropathicpain unmasked in the contralateral leg, againwith good effect. Six patients reduced their co-analgesic intake postoperatively. The mediannumber of different types of co-analgesics fellfrom four types preoperatively (range 2–4) to twotypes postoperatively (range 0–3).

No major complications were identified. Inparticular, there were no respiratory compli-cations or transitory or persistent paresis inthe ipsilateral limb. The two patients with poorsphincter control prior to cordotomy had con-tinued difficulties. Of the two patients men-tioned above, one patient had transient hesitancyof micturition postoperatively and another hadslight fecal leakage, which spontaneously re-solved.

All the patients have died from disease pro-gression. The median survival post-cordotomywas 107 days (range 28–830 days). Six patientsdeveloped pain in the contralateral leg or atanother site prior to death, but symptom con-trol was achieved by non-invasive methods.

Discussion

The cases reviewed in this report represent avery small proportion of the total number of pa-tients with difficult pain problems managed bythe palliative care team in South-East Wales. Withhindsight, some patients, particularly Case 1,should have been referred earlier for cordotomy;subsequently routes of referral became morestreamlined, as illustrated by Case 6. Although ar-guments exist about open or closed cordotomy,

our results demonstrate a fall in analgesic require-ments post-cordotomy in all but Case 3. This pa-tient experienced very severe postoperative painin the wound that responded poorly to the esca-lating doses of opioid.

The chief complications of unilateral cordot-omy

11

are given in Table 3. In this carefully se-lected patient group, there were no significantpostoperative complications.

Six patients subsequently developed pain onthe opposite side, controlled by non-invasivemethods. It was felt that these pains developedeither as a result of the unmasking of pain asopioids were weaned off in the postoperativeperiod, or that they represented ongoing dis-ease progression. These were not felt to be ex-amples of mirror pain.

Four patients (Cases 3, 5, 7, and 8) were ableto have holidays away from home.

To date, only one of the SE Wales patientslived longer than two years following cordot-omy. She developed severe pain in the con-tralateral leg and perineum but did not de-velop dysesthetic pain in the initially painfullimb. No patient experienced recurrent painin the leg for which the cordotomy was under-taken. Interestingly, one patient died with agangrenous leg secondary to arterial thrombosison his initially painful side, without experienc-ing any pain.

There are some limitations to the data pre-sented here. This retrospective review covered10 years, during which time no one pain assess-ment tool had been consistently used. There-fore, changes in opioid and co-analgesics havebeen used as markers for symptom control im-provement, along with anecdotal comments inthe notes. The most marked increase in the ac-

Table 2

Opioid Analgesic Requirements Per Day Pre- and Post-Cordotomy

Oral Morphine(mg or equiv.)

Ketamine(mg)

Methadone(mg)

Fentanyl (mg)

Pre op Post op Pre op Post op Pre op Post op Pre op Post op

Case 1 900 160 — — — — 2.5 0Case 2 580 120 — — — — — —Case 3 600 1000 — — 25 0 2.4 0Case 4 400 180 350 0 — — — —Case 5 360 160 — — — — — —Case 6 480 400 700 200 200 0 — —Case 7 2600 300 100

epidural0 0 30 — —

Case 8 930 140 500 0 — — — —Case 9 480 40 — — — — — —

184 Jones et al. Vol. 25 No. 2 February 2003

tivities of daily living was the ability to go onholiday, as prior to cordotomy, patients wereseverely restricted.

Conclusions

Cordotomy may be an underused method ofpain control in patients with unilateral lowerquadrant (including perineal) pain that doesnot respond to oral analgesics. Unlike an in-dwelling epidural or intrathecal cannula, thereis no ongoing risk of infection at the cordot-omy site and patients can return home withoutneeding special equipment or training of com-munity staff in the administration of epidural/spinal medication. The fall in medication re-quirements may represent some cost saving, al-though a true cost-benefit analysis is not possi-ble on this cohort in the absence of matchedcontrols. We recommend that every specialistpalliative care center should know the referralpathways for cordotomy in their area, if indi-cated. Open cordotomy, by a specialist neuro-surgeon, still has a role in palliative care.

References

1. Bonica JJ. Cancer pain: current status and fu-ture needs. In: Bonica JJ, ed. The management ofpain, 2nd ed. Philadelphia: Lea and Febinger, 1990:400–445.

2. Makin MK. Strong opioids for Cancer Pain. JRoyal Soc Med 2001;94(1):17–21.

3. Cramond T, Stuart G. Intraventricular mor-phine for intractable pain of advanced cancer. JPain and Symptom Manage 1993;8(7):465–473.

4. Sundaresan N, Digiacinto G, Hughes J. Neuro-surgery in the treatment of cancer pain. Cancer1989; 63 (11 Suppl):2365–2377.

5. Spiller WG, Martin E. The treatment of persis-tent pain of organic origin in the lower part of thebody by division of the anterolateral column of thespinal cord. JAMA 1912;58:1489–1490.

6. Mullan S, Hekmatpanah J, Dobben G, et al. Per-cutaneous intramedullary cordotomy utilizing theunique unipolar anodal electrolytic system. J Neuro-surgery 1965;22:548.

7. Tasker RR, DeCarvalho GT, Dolan EJ. Intracta-ble pain of spinal cord origin: clinical features andimplications for surgery. J Neurosurgery 1992;77(3):373–378.

8. Rosen S. Procedural control of cancer pain.Semin Oncol 1994:21(6):740–747.

9. Krieger AJ, Rosomoff HL. Sleep-induced apnea.A respiratory and autonomic dysfunction syndromefollowing bilateral percutaneous cervical cordot-omy. J Neurosurgery 1974:40(2); 68–80.

10. Nathan PW. Effects on movement of surgical in-cisions into the human spinal cord. Brain 1994;117:337–346.

11. Tasker RR. Central neurosurgery procedures onthe spinal cord. In Wall PD, Melzack R, eds. Text-book of pain, 4th ed. Edinburgh: Churchill Living-stone, 1999:1309–1315.

12. Nagaro N, Amakawa K, Kimura S, et al. Refer-ence of pain following percutaneous cervical cordot-omy. Pain 1993;53:205–211.

Table 3

Chief Complications of Unilateral Cordotomy (Including High Cervical Cordotomy)

Complication Comment

High Cervical Cordotomy onlyDeath (0–5%) as a result of respiratory complications Patient selectionReversible respiratory complications (up to 10%)

High Cervical and Thoracic CordotomyPersistant paresis or ataxia (up to 10%) Damage to adjacent corticospinal tract

10

Worsening of control of micturition (up to 15%) Due to bilateral innervation usually only a exaggeratesan existing problem

Postcordotomy dysesthesia (less than 15%) An example of idiosyncratic central pain caused by an iatrogenic cord lesion

Mirror pain (9–63%) Pain in the same location but on the opposite side to originalpain, occurring immediately postoperatively and abolished by epidural blockade

12