Pain 83 (1999) 389400

Review Article

www.elsevier.nl/locate/pain

Efcacy of pharmacological treatments of neuropathic pain: an update and

effect related to mechanism of drug action
Sren H. Sindrup a,*, Troels S. Jensen b
a

Department of Neurology, Odense University Hospital, DK-5000 Odense C, Denmark

b
Department of Neurology, Aarhus University Hospital, Aarhus, Denmark
Received 19 May 1999; accepted 9 June 1999

Abstract
Tricyclic antidepressants and carbamazepine have become the mainstay in the treatment of neuropathic pain. Within the last decade,
controlled trials have shown that numerous other drugs relieve such pain. We identied all placebo-controlled trials and calculated numbers
needed to treat (NNT) to obtain one patient with more than 50% pain relief in order to compare the efcacy with the current treatments, and to
search for relations between mechanism of pain and drug action. In diabetic neuropathy, NNT was 1.4 in a study with optimal doses of the
tricyclic antidepressant imipramine as compared to 2.4 in other studies on tricyclics. The NNT was 6.7 for selective serotonin reuptake
inhibitors, 3.3 for carbamazepine, 10.0 for mexiletine, 3.7 for gabapentin, 1.9 for dextromethorphan, 3.4 for tramadol and levodopa and 5.9
for capsaicin. In postherpetic neuralgia, the NNT was 2.3 for tricyclics, 3.2 for gabapentin, 2.5 for oxycodone and 5.3 for capsaicin, whereas
dextromethorphan was inactive. In peripheral nerve injury, NNT was 2.5 for tricyclics and 3.5 for capsaicin. In central pain, NNT was 2.5 for
tricyclics and 3.4 for carbamazepine, whereas selective serotonin reuptake inhibitors, mexiletine and dextromethorphan were inactive. There
were no clear relations between mechanism of action of the drugs and the effect in distinct pain conditions or for single drug classes and
different pain conditions. It is concluded that tricyclic antidepressants in optimal doses appear to be the most efcient treatment of
neuropathic pain, but some of the other treatments may be important due to their better tolerability. Relations between drug and pain
mechanisms may be elucidated by studies focusing on specic neuropathic pain phenomena such as pain paroxysms and touch-evoked pain.
q 1999 International Association for the Study of Pain. Published by Elsevier Science B.V.
Keywords: Pharmacological treatments; Neuropathic pain; Placebo-controlled trials

1. Introduction
Neuropathic pain represents heterogenous conditions,
which neither can be explained by one single aetiology
nor by a particular anatomical lesion. This diversity in
cause and site is reected in entities such as peripheral
nerve injuries due to trauma and poststroke pain due to
ischaemic vascular lesions of the brain. Despite the different
aetiology and the multiple lesions giving rise to neuropathic
types of pain, many of these conditions share common clinical phenomena like: no visible injury, paradox combination
of sensory loss and hyperalgesia in the painful area, paroxysms and a gradual increase of pain following repetitive
stimulation (Fields, 1990; Bennett, 1994; Jensen, 1996).
These observations have led to the proposal that neuropathic
pain may be explained by the same or similar mechanisms.
Indeed, it has been suggested that hyperexcitability peripherally and at more central sites could be a mechanism by
* Corresponding author. Tel.: 145-6541-2474; fax: 145-6541-3389.

which these pains are explained. Consequently, recent treatments have focused on drugs that reduce neuronal hyperexcitability either peripherally or centrally. For example, in
peripheral nerve injuries and in postherpetic neuralgia, Cnociceptor sensitization is invoked and drugs like sodium
channel blockers, which silence spontaneous and evoked
activity (Devor et al., 1992) may be used. In other cases,
the hyperexcitability is more amenable to blockade by drugs
that reduce activity at NMDA receptors (Dray, 1997).
Despite this understanding of mechanisms of hyperexcitability, treatments are not always successful. Our current
knowledge has shown that hyperexcitability per se does
not represent a single mechanism, but rather a combination
of factors that add together and determine the degree and
type of hyperexcitability in the individual patient and in a
particular neuropathic condition.
One possibility to look further into mechanism is to
systematically analyse drug activity in neuropathic disorders. For example by looking at efcacy of specic drugs
across different disorders, it may be possible to assess

0304-3959/99/$20.00 q 1999 International Association for the Study of Pain. Published by Elsevier Science B.V.
PII: S 0304-395 9(99)00154-2

390

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

whether different mechanisms are at play in different pain

conditions (Woolf et al., 1998) and by looking at specic
disorders across different drugs it is possible to determine
whether different mechanisms are involved in specic pain
conditions.
Numbers needed to treat (NNT), i.e. the number of
patients needed to treat with a certain drug to obtain one
with a dened degree of pain relief, has been introduced as a
useful method to examine analgesic efcacy (for review see:
McQuay and Moore, 1998). This method permits a clinically relevant comparison between different drugs and
disorder. The purpose of this review was to evaluate efcacy
by the NNT method for pharmacological treatments of
neuropathic pain and compare efcacy across disorders
and drugs, and if possible also across mechanisms involved.
2. Methods
The NNT (Cook and Sackett, 1995) approach has been
used in recent reviews on antidepressants and anticonvulsants in neuropathic pain (McQuay et al., 1995, 1996) to
give an estimate of the analgesic efcacy of these drug
classes. The NNT for more than 50% of pain relieved
outcome seems clinically relevant and is easily understood,
and it is referred to simply as NNT in this text. It is the
number of patients we need to treat with a certain drug to
obtain one patient with at least 50% pain relief. More than
50% of pain relieved cannot always be evaluated directly
from the data given in publications, but this goal is considered to be achieved for patients with a response characterised as `excellent/good/moderate' (patient's global
evaluation or pain relief), `no pain/slight pain' (pain intensity) or more than 50% reduction in score (pain intensity or
neuropathy scale) (McQuay et al., 1996). It is calculated as
NNT l/((goal achievedactive/totalactive) 2 (goal achievedplacebo/totalplacebo)) and the 95% condence interval (CI) of NNT
can be obtained by taking the reciprocal value of the 95% CI
for the absolute risk reduction (Cook and Sackett, 1995).
Data on pharmacological treatments of neuropathic pain
were only considered to be valid for efcacy evaluation if
derived from studies performed with a randomised, placebocontrolled and double-blind design. It was decided only to
include efcacy data on current and new treatments that had
been tested in chronic dose settings in order to obtain NNTs
that could be safely compared between drugs and conditions. Treatment given as e.g. a single intravenous infusion
rarely has a place in clinical practice.
3. Results
The studies on pharmacological treatments of neuropathic pain included in this analysis are listed in Table 1.
The corresponding NNTs with 95% condence intervals
and NNTs for the current treatments as reviewed previously
(McQuay et al., 1995, 1996) or recalculated if necessary for

specic purposes are shown for each pain condition in Table

2.
Tricyclic antidepressants and sodium channel blockers
are currently considered to be the drug treatments of choice
for neuropathic pain. There is a large number of randomised, controlled clinical trials with evidence of a benecial
effect of antidepressants in neuropathic pain, while the
evidence of the sodium channel blockers is more scarce.
The majority of studies have been performed in patients
with postherpetic neuralgia and painful diabetic neuropathy.
3.1. Tricyclic antidepressants
From three studies in postherpetic neuralgia, the
combined NNT is 2.3 (CI 1.73.3), from 13 diabetic neuropathy studies, NNT is 3 (2.44) and in one central post
stroke pain study, NNT is 1.7 (13) (McQuay et al., l996)
for antidepressants. The studies on diabetic neuropathy
include trials on both tricyclics and selective serotonin reuptake inhibitors (SSRI). If studies on SSRIs are excluded
together with a tricyclic tested in combination with a
phenothiazine (Gomez-Perez et al., 1985), the NNT in
diabetic neuropathy is 2.4 (2.03.0). The large number of
studies in diabetic and other painful neuropathies allow an
analysis of efcacy of subtypes of tricyclics. Imipramine,
amitriptyline and clomipramine causes a balanced reuptake
inhibition of both serotonin and noradrenaline, while desipramine and maprotiline are relative selective noradrenaline
gren, 1981). In painful polyreuptake inhibitors (Hall and O
neuropathy, the combined NNT for drugs with balanced
reuptake inhibition is 2.0 (1.72.5), while it is 3.4 (2.3
6.6) for the noradrenergic compounds. In postherpetic neuralgia, there are far fewer studies and the relationship is
reversed (Table 2).
In many of the studies, the antidepressants were dosed
according to effect and side effects. This may result in an
underestimation of the potential effect of the tricyclics, since
with these drugs, side effects are often bothersome and dose
-effect or plasma drug concentrationeffect relations have
been found in several studies (Kvinesdal et al., 1984; Max et
al., 1987; Leijon and Boivie, 1989; Sindrup et al., 1990b;
McQuay et al., 1993; Kalso et al., 1995). In two studies on
imipramine in diabetic neuropathy (Sindrup et al., 1990a,
1992a), the dose was adjusted to obtain the optimal plasma
concentration of imipramine plus its active metabolite desipramine around 400 nM (Sindrup et al., 1990c). The target
concentration was obtained in 16 of 19 patients and 5 of 18
patients, respectively. From the original data of the rst
study, a NNT of 1.4 (1.11.9) is calculated, i.e. a value
below the lower 95% condence limit for entire group of
studies. This strongly indicates that the effect can be
increased by this dosage policy.
The data from several of the controlled studies indicate
that tricyclics are effective for both steady and lancinating
or brief pains (Max et al., 1987, 1991, 1992; Sindrup et al.,
1990a,b), whereas it is more difcult to judge if these

GABA-B agonist

NMDA antagonist

Ion channel blocker

SSRI

Noradrenergic

Antidepressant
Tricyclic
Serot./noradr.

Drug class

Trigeminal neurlagia

Post-stroke pain
Diabetic neuropathy
Posterpetic neuralgia
Posterpetic neuralgia
Fromm et al., 1984

McQuay et al., 1994

Max et al., 1997
Max et al., 1997
Eisenberg et al., 1998

Dejgard et al., 1988

Stracke et al., 1992
Wright et al., 1997
Oskarsson et al., 1997
Chiou-Tan et al., 1996
Cambell et al., 1966
Killian et al., 1968
Rull et al., 1968
Leijon et al., 1989
Chadda et al., 1978
Saudek et al., 1977
Zakrzewska et al., 1997
Backonja et al., 1998
Rowbotham et al., 1998

Sindrup et al., 1990a

Max et al., 1992
Sindrup et al., 1992b
Vestergaard et al., 1996

Diabetic Neuropathy
Diabetic Neuropathy
Diabetic Neuropathy
Post-stroke pain

Diabetic neuropathy
Diabetic neuropathy
Diabetic neuropathy
Diabetic neuropathy
Spinal cord injury
Trigeminal neuralgia
Trigeminal neuralgia
Diabetic neuralgia
Post-stroke pain
Diabetic neuropathy
Diabetic neuropathy
Trigeminal neurlagia
Diabetic neuropathy
Posterpetic neuralgia

Watson et al., 1982

Kvinesdal et al., 1984
Max et al., 1987
Max et al., 1988
Sindrup, 1990a
Sindrup, 1990b
Sindrup, 1992a
Vrethem et al., 1997
Kalso et al., 1995
Kishore-Kumar et al., 1990
Max et al., 1991
Vrethem et al., 1997

Trial

Postherpetic neuralgia
Diabetic neuropathy
Diabetic neuropathy
Postherpetic neuralgia
Diabetic neuropathy
Diabetic neuropathy
Diabetic neuropathy
Polyneuropathy
Nerve injury pain
Postherpetic neuralgia
Diabetic Neuropathy
Polyneuropathy

Pain condition

Baclofen

Dextrometh.
Dextrometh.
Dextrometh.
Memantine

Mexiletine
Mexiletine
Mexiletine
Mexiletine
Mexiletine
Carabamazep.
Carabamazep.
Carabamazep.
Carabamazep.
Phenytoin
Phenytoin
Lamotrigine c
Gabapentin
Gabapentin

Paroxetine
Fluoxetine
Citalopram
Citalopram

Amitriptyline
Imipramine
Amitriptyline
Amitriptyline
Imipramine
Clomipramine
Imipramine
Amitriptyline
Amitriptyline
Desipramine
Desipramine
Maprotiline

Active drug

Table 1
Randomised, double-blind, placebo-controlled trials of different drugs in neuropathic pain

6080

81
average 381
average 439
20

10 (mg/kg)
225/450/675
600
225/450/675
450
400800
4001000
200600
800
300
300
400
3600
12003600

40
40
40
1040

average 73
100
average 90
average 65
average 200
75
150
75
100
average 167
average 201
75

Daily dose (mg)

10

9
13
13
24

16
94
29
126
11
77
30
30
14
40
12
14
165
225

20
48
15
9/4

24
12
29
34
20
19
18
33
15
19
20
33

Cross-over

Cross-over
Cross-over
Cross-over
Parallel

Cross-over
Parallel
Parallel
Parallel
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Parallel
Parallel

Cross-over
Cross-over
Cross-over
Parallel

Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over
Cross-over

Design

bacl . pla

dext pla
dext . pla
dext pla
mem pla

mex . pla
mex pla
mex pla
b
mex . pla
mex pla
carb . pla
carb . pla
carb . pla
carb pla
phen . pla
phen pla
lamo . pla
gaba . pla
gaba . pla

par . pla
u pla
cit . pla
cit pla

ami . pla
imi . pla
ami . pla
ami . pla
imi . pla
clo . pla
imi . pla
ami . pla
ami . pla
desi . pla
desi . pla
mapr . pla

Outcome

1/24
0/12
1/29
5/25
3/20
1/19
2/18
8/33
2/15
2/19
2/20
8/33

Improved on placebo

8/10

0/9
7/13
5/13
No dichotomous data

No dichotomous data
No dichotomous data
No dichotomous data
23/31
No dichotomous data
144/268
19/27
28/30
5/14
28/38
No dichotomous data
7/13
47/79
47/109

1/10

0/9
0/13
3/13

1/14
25/76
14/116

35/190
0/27
19/30
1/15
10/38

20/31

10/20
3/20
22/46
19/46
3/15
1/15
No dichotomous data

16/24
7/12
15/29
15/34
17/19
10/19
8/18
22/33
a
8/15
12/19
11/20
14/33

Improved on active

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

391

Diabetic neuropathy

Postherpetic neuralgia
Diabetic neuropathy
Polyneuropathy

Diabetic neuropathy
Diabetic neuropathy
Diabetic neuropathy
Diabetic neuropathy
Diabetic neuropathy
Postherpetic neuralgia
Nerve injury pain

Pain condition

Ertas et al., 1998

Watson et al., 1998

Harati et al., 1998
Sindrup et al., 1999

Chad et al., 1990

Schefer et al., 1991
Caps. Study Group, 1991
Tandan et al., 1992
Low et al., 1995
Bernstein et al., 1989
Watson et al., 1992

Trial

Additional data provided by author or from poster presentation.

Only the highest dose level signicantly better than placebo.
c
Lamotrogine used as add on treatment to carbamazepine.
d
Placebo on one leg and capsaicin on the other.

Levodopa

Opioid

Capsaicin

Drug class

Table 1 (continued)

Madopar

Oxycodone
Tramadol
Tramadol

Cream
Cream
Cream
Cream
Cream
Cream
Cream

Active drug

300

2060
100400
200400

0.075% qid
0.075% qid
0.075% qid
0.075% qid
0.075% qid
0.075% tid/qid
0.075% qid

Daily dose (mg)

25

38
131
34

46
41
202
10
39
32

Parallel

Cross-over
Parallel
Cross-over

Parallel
Parallel
Parallel
Parallel
Parallel d
Parallel

Design

mad . pla

oxy . pla
tram . pla
tram . pla

caps pla
caps . pla
caps . pla
caps . pla
caps pla
caps . pla
caps . pla

Outcome

8/14

22/38
43/63
11/34
a

17/24
17/19
65/91
6/10
23/39
4/16
5/13

Improved on active

3/11

7/38
23/64
3/33

11/22
11/22
57/111
2/10
26/39
1/16
1/10

Improved on placebo

392
S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

393

Table 2
Numbers needed to treat to obtain one patient with more than 50% pain relief. In case of more than one study on a drug in the pertient pain type NNT is
calculated for combined data. NA Not active, ND Not done, TCA tricyclic antidepressants, SSRI selective serotonin reuptake inhibitors
Painful neuropathy
Antidepressants
Antidepressants all types
TCA all types
TCA serot./noradr.
TCA noradrenergic
TCA serot./noradr.
Optimal dose
SSRI

Posthepertic neuralgia

Peripheral nerve injury

Central pain

Trigeminal neuralgia

3.0 (2.44.0)
2.4 (2.03.0)
2.0 (1.72.5)
3.4 (2.36.6)

2.3
2.3
2.4
1.9

2.5 (1.410.6)
2.5 (1.410.6)
2.5 (1.410.6)
ND

1.7 (1.13.0)
1.7 (1.13.0)
1.7 (1.13.0)
ND

ND
ND
ND
ND

1.4 (1.11.9)
6.7 (3.4435)

ND
ND

ND
ND

ND
NA

ND
ND

(1.73.3)
(1.73.3)
(1.83.9)
(1.33.7)

Ion channel blockers

Mexiletine
Phenytoin
Carbamazepine
Lamotrigin
Gabapentin

10.0 (31)
2.1 (1.53.6)
3.3 (29.4)
ND
3.7 (2.48.3)

ND
ND
ND
ND
3.2 (2.4 5.0)

ND
ND
ND
ND
ND

NA
ND
3.4 (1.7105)
ND
ND

ND
ND
2.6 (2.23.3)
a
2.1 (1.36.1)
ND

NMDA antagonists
Dextromethorphan
Memantine

1.9 (1.13.7)
ND

NA
NA

ND
ND

b
NA
ND

ND
ND

GABAB agonist
Baclofen

ND

ND

ND

ND

Opioids
Oxycodone
Tramadol

ND
3.4 (2.36.4)

d
2.5 (1.65.1)
ND

ND
ND

ND
ND

ND
ND

Various
Levodopa
Capsaicin

3.4 (1.51)
5.9 (3.813)

ND
5.3 (2.31)

ND
3.5 (1.61)

ND
ND

ND
ND

1.4 (1.02.6)

Add on therapy to carbamazepine.

Low dose of dextromethorphan.
c
Add on therapy to carbamazepine or phenytoin in 4 of 10 patients.
d
For 30% of patients add on therapy to tricyclic antidepressants.
b

drugs also relieve touch-evoked pain. It is an inherited

problem with these studies that none of them addressed
the issue of an effect on different pain types, but only
showed that patients with the different types of pain were
relieved of pain in general.
More recent data show that amitriptyline also relieves
nerve injury pain quite effectively, with NNT 2.5 (1.4
2.6) (Kalso et al., 1995).
A recently reported interesting aspect of tricyclic antidepressants in pain treatment is the apparent potential to
reduce the risk of developing postherpetic neuralgia by
low dose amitriptyline during the acute phase of herpes
zoster (Bowsher, 1997).
3.2. Selective antidepressants
The selective serotonin reuptake inhibitors (SSRI) is a
new class of antidepressants. They differ from classical
tricyclic antidepressants in their specic inhibition of presynaptic reuptake of serotonin but not of noradrenaline and

their lack of the postsynaptic receptor blocking effects and

quinidine-like membrane stabilization seen with the tricyclics. In two out of three studies on SSRIs in painful diabetic
neuropathy, there was a signicantly better effect of the
SSRI than of placebo (Sindrup et al., 1990a, 1992b; Max
et al., 1992) (Table 1). The individual NNT from the studies
showing a signicant effect is 2.9 (paroxetine) and 7.7 (citalopram) and the combined NNT for all three studies is 6.7
(3.4435). Paroxetine seems to relieve both steady and
lancinating pain (Sindrup et al., 1990a). Central pain is
not relieved by citalopram (Vestergaard et al., 1996).
The results from these and other studies in diabetic neuropathy indicated that drugs with a balanced inhibition of
serotonin and noradrenaline but without the postsynaptic
and quinidine-like effects of the TCAs could be as effective
as the tricyclics (Sindrup, 1994) and at the same time be
better tolerated. One such drug, venlafaxine, has been
marketed for the treatment of depression. Preliminary data
from a randomised, double-blind, placebo-controlled study
on venlafaxine in neuropathic pain following treatment of

394

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

breast cancer indicate that venlafaxine has some effect but

data for calculation of NNT are not available (Tasmuth and
Kalso, 1998).
3.3. Ion channel blockers
Lidocaine is the prototype of an unspecic sodium channel blocker. More than 10 years ago, it was reported that
infusions of lidocaine relieved painful diabetic neuropathy
(Kastrup et al., 1987) and the data from that study show a
favourable NNT of 3 (510). Lidocaine is not convenient for
chronic treatment, since it cannot be dosed orally. Studies
on its oral analogue mexiletine have therefore been carried
out. Two of the studies found a better effect of mexiletine
than of placebo (Dejgard et al., 1988; Oskarsson et al., 1997)
while the two other studies either reported no effect (Wright
et al., 1997) or only effect in sub-analyses and not on
primary effect variables (Stracke et al., 1992) (Table 1).
Dichotomous data are reported in one of the studies favouring mexiletine (Oskarsson et al., 1997), but it is important to
mention that only the highest of three dose levels tested in
that study was better than placebo. The NNT for this dose
level (675 mg/day) is 10 (3.01). Data on the effect on
different pain categories are not available.
One trial studied the effect of mexiletine in chronic
dysaesthethic pain after spinal cord injury but failed to
nd an effect (Chiou-Tan et al., 1996).
Phenytoin and carbamazepine are widely used as anticonvulsants. These drugs exert their membrane-stabilizing
properties by blocking sodium channels unspecically
(Dray, 1997) and therefore reduce neuronal excitability in
sensitized C-nociceptors. In painful diabetic neuropathy,
carbamazepine has an NNT of 3.3 (29.4) and phenytoin
an NNT of 2.1 (1.53.6) (McQuay et al., 1995). There is
only one trial on each drug (Rull et al., 1969;Chadda and
Mathur, 1978) and it is important that a second placebocontrolled study with phenytoin failed to demonstrate a
signicant effect of this drug (Saudek et al., 1977). The
carbamazepine dose used (titration from 200 to 600 mg/
day) is rather low and it is possible that the efcacy could
be increased by using higher doses. Although carbamazepine is frequently used in post-stroke pain, it is not signicantly better than placebo in the small study (n 14) that
showed a clear effect of the tricyclic antidepressant amitriptyline in this condition (Leijon and Boivie, 1989). However,
the corresponding NNT of 3.4 (1.7105) is quite acceptable
and the lack of statistical signicance may simply be related
to the small sample size.
Carbamazepine is the treatment of choice for trigeminal
neuralgia and data from three controlled trials have been
published (Campbell et al., 1966; Killian and Fromm,
1968; Nicol, 1969). It is difcult to interpret the data from
the latter study, which also failed to nd an effect. For the
two other studies, there is a combined NNT of 2.6 (2.23.3)
(McQuay et al., 1995). It is reported that carbamazepine
both causes a reduction in pain intensity, a reduction in

pain paroxysms and in triggers (Campbell et al., 1966).

There are only anecdotal data on oxcarbamazepine (Zakrzewska and Patsalos, 1989), a new carbamazepine-like
anticonvulsant acting by sodium channel blockade and characterized by a less bothersome side effect prole than carbamazepine.
Lamotrigine is a new anticonvulsant, which acts by stabilising the slow inactivated conformation of a subtype of
sodium channels and probably by this mechanism suppress
the neuronal release of glutamate. Anecdotal observations in
patients with central pain (Canavero and Bonicalzi, 1996)
and an uncontrolled study in painful diabetic neuropathy
(Eisenberg et al., 1998) indicate that lamotrigine will in
fact relieve neuropathic pain. Lamotrigine has been tried
as add on treatment to carbamazepine in trigeminal neuralgia (Zakrzewska et al., 1997), and in this setting it has a
favourable NNT of 2.1 (1.36.1).
In the studies on diabetic neuropathy and central pain,
there are no data on the types of pain that respond to the
sodium channel blockers, but the efcacy of several of
these drugs in trigeminal neuralgia (Campbell et al.,
1966; Killian and Fromm, 1968; Zakrzewska et al., 1997)
strongly indicate an effect on lancinating pains and pain
paroxysms.
Gabapentin is a novel anticonvulsant with an unknown
mechanism of action. It does not interact with GABA receptors or GABA metabolism and it has no effect on sodium
channels. Recent data suggest that it blocks a subtype of
calcium channels on neurones. After several uncontrolled
studies (e.g. Rosenberg et al., 1997) with gabapentin in
neuropathiuc pain, the rst two controlled studies have
recently been presented (Table 1). In an adequately
designed study with 165 patients with painful diabetic
neuropathy, gabapentin in a dose of 3600 mg/day was
compared to placebo and the NNT was 3.7 (2.48.3) (Backonja et al., 1998). There are no data on the effect on individual pain types in that study. Data from a placebo-controlled
study in postherpetic neuralgia show a NNT value of 3.2
(2.45.0) (Rowbotham et al., 1998).
3.4. GABA-B receptor agonist
Baclofen is a GABA-B receptor agonist, which has been
examined in trigeminal neuralgia (Fromm et al., 1984;
Fromm and Terrence, 1987). These authors demonstrated a
signicant pain-relieving effect of baclofen and the NNT for
a signicant decrease in painful paroxysms was 1.4 (1.02.6)
and later showed that the L-baclofen is more effective than
racemic baclofen.
3.5. NMDA antagonists
The evidence of involvement of excitatory amino acids in
neuropathic pain has also prompted studies on drugs with an
NMDA-antagonistic effect. One such drug is ketamine and
intravenous infusion studies with a double-blind, placebocontrolled design have shown an immediate effect of keta-

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

395

Fig. 1. Numbers needed to treat (NNT) to obtain one patient with more than 50% pain relief for different drugs in different neuropathic pain conditions. NNT is
shown in two directions to enable comparison between different drugs in the same condition and across different conditions for the same drug or drug classes.
Symbol placed in the upper right hand corner of a box indicates that the drug is inactive in the pertinent pain condition or has a NNT 10 (confer Table 2).
TCA tricyclic antidepressants, SSRI selective serotonin reuptake inhibitors, PHEN phenytoin, CARB carbamazepine, GABA gabapentin,
LAMO lamotrigine, MEX mexiletine, DEXTRO dextromethorphan, MEMA memantine, OXY oxycodone, TRAM tramadol, L-DOPA
levodopa with benzerazide, CAPS capsaicin, BACLO baclofen.

mine in patients with chronic neuropathic pain (Max et al.,

1995; Felsby et al., 1996). Ketamine orally for a few months
has been tried in single patients with neuropathic pain and
the results are rewarding (Broadley et al., 1996; Nikolajsen
et al., 1997), but controlled trials on oral ketamine in neuropathic pain have not been published.
The low afnity NMDA channel blocker dextromethorphan relieved diabetic neuropathy with NNT 1.9 (1.13.7),
whereas it had no effect in postherpetic neuralgia (Nelson et
al., 1997). It was reported that in diabetic patients, there was

no difference between good and poor responders with

respect to pain quality.
A preceding study on dextromethorphan included both
patients with central post-stroke pain and different types
of neuropathic pain of peripheral origin (McQuay et al.,
1994). This study used a much lower dose than the above
study and no effect of dextromethorphan was found. A similar lack of effect was also seen in 9 patients with post-stroke
pain, which was the largest patient subgroup.
In a recent study on the non-competitive NMDA receptor

396

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

antagonist memantine it was reported that it did not relieve

postherpetic neuralgia ( Eisenberg et al., 1998).
3.6. Opioids
Following the dispute 10 years ago on opioid sensitivity
of neuropathic pain (Arner and Meyerson, 1988; Portenoy et
al., 1990; McQuay et al., 1992), there is now less controversy whether opioid analgesics are of any benet in neuropathic pain. Evidence is now compiling that opioids are
effective in certain types of neuropathic pain. Intravenous
infusion of morphine relieves postherpetic neuralgia
(Rowbotham et al., 1991), and it was recently shown that
infusion of fentanyl, which is also a m-opioid receptor
agonist, relieves different types of neuropathic pain states
(Dellemijn and Vanneste, 1997). The effect of the m-opioid
receptor against oxycodone has been tested in postherpetic
neuralgia in a long-term trial with oral dosing (Watson and
Babul, 1998) (Table 1). Oxycodone was superior to placebo,
but in 30% of the patients oxycodone was used as an add on
drug to treatment with antidepressants. The calculated NNT
of 2.5 (1.65.1) for oxycodone has therefore to be judged
with caution. It was clearly shown that oxycodone relieves
both steady pain, brief pain and allodynia and 67% of the
patients preferred oxycodone as compared to 11% preferring placebo.
Tramadol is an analgesic drug, which probably acts
through both monoaminergic and opioid mechanisms. The
monoaminergic effect is shared with the tricyclic antidepressants. Further, development of tolerance and dependence during long-term tramadol treatment appears to be
uncommon and tramadol seems to have a low abuse liability. Therefore, tramadol may be an alternative to strong
opioids, and two recent studies tested the drug in painful
diabetic polyneuropathy (Harati et al., 1998) and painful
polyneuropathy of different etiologies (Sindrup et al.,
1998). Tramadol was superior to placebo in both studies
and came up with NNTs of 3.1 and 4.3, and the combined
NNT was 3.4 (2.36.4). In the latter study, nearly 80%
preferred tramadol for placebo and it was shown that tramadol parallel with its relief of on-going pain reduced touchevoked pain and experimentally induced mechanical allodynia.
3.7. Levodopa
Dopamine agonists inhibit noxious input to the spinal
cord (Jensen and Yaksh, 1984). Levodopa, being a dopamine precursor, may have a similar effect or may act as a
precusor to noradrenaline and modulate pain via noradrenergic mechanisms. Nearly 20 years ago, it was found in an
adequately designed study that levodopa reduced pain in
acute herpes zoster (Kembaum and Hauchecome, 1981)
and recent data show that this is also the case in diabetic
neuropathy (Ertas et al., 1998). In diabetic neuropathy, NNT
was 3.4 (1.51) and the study did not differentiate the type
of pain that responded to the treatment.

3.8. Capsaicin
Capsaicin, an alkaloid derived from chillies, depletes the
neurotransmitter substance P from sensory nerves.
Topically applied capsaicin cream showed a signicant
effect in 3 of 5 studies in diabetic neuropathy (Chad et al.,
1990; Schefer et al., 1991; Capsaicin Study Group, 1991;
Tandan et al., 1992; Low et al., 1995) with NNTs in the
positive studies from 2.5 to 4.9 and a combined NNT for all
studies of 5.9 (3.813).
Capsaicin has been tried with a positive outcome in postherpetic neuralgia (Bernstein et al., 1989) and pain after
nerve injury (Watson and Evans, 1992) and the NNTs are
5.3 and 3.5, respectively.
Although this treatment may have few side effects besides
the burning pain on application at treatment start, it may be
less convenient in many patients, since it has to be applied 4
times daily on the entire painful area.
4. Discussion
The present review indicates that NNT to determine treatment efcacy in neuropathic pain provides consistent values
for different drug classes. In painful diabetic neuropathy, the
most extensively studied pathological condition, NNT
values for tricyclic antidepressants and sodium channel
blockers are around 23. This means that in neuropathic
pain, 23 patients have to be treated before one patient
with $ 50% pain relief is obtained. The benets and limitations of this simple approach will be discussed briey in the
following (see Fig. 1).
4.1. Outcome measures
A major problem in many clinical drug trials is the failure
to predict outcome because individual trials often include a
small sample size. In a recent study (Moore et al., 1998), it
was shown that it is only possible to determine efcacy of a
treatment from large trials. Thus, a meta-analysis of several
trials may be necessary to generate a sufcient number of
patients to determine the usefulness of a specic drug. With
the use of a common outcome measure such as NNT-values,
it is possible to compare and combine several studies. For
pain, an essential parameter is obviously the degree of pain
relief associated with a particular drug. From available
studies, it is not always possible to determine a 50% pain
relief and accordingly this gives rise to some uncertainty.
However, it has been argued that since NNT describes the
difference between treatment and control, a changing
threshold for calculating NNT does not affect the result
much (McQuay and Moore, 1998). An almost equally
important aspect in determining drug efcacy is the potential harmfulness associated with a particular drug. In a nal
analysis of a drug, it may therefore be essential also to
determine the numbers needed to harm (NNH), i.e. the
numbers needed to treat before there is one patient with

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

pronounced or intolerable side effects. Future studies in

which quantitation of pain relief and harmful side effects
are decided a priori may be important in determining the
degree of uncertainty associated with such retrospective
outcome measures.
4.2. New versus established treatments
The calculated NNT for tricyclic antidepressants in painful polyneuropathy are based on a sufcient total number of
patients to make an accurate estimate. On the other hand, the
NNT for tricyclics in other neuropathic pain conditions, the
NNTs for carbamazepine and phenytoin as well as for the
more recent treatments are all calculated from a relatively
small number of patients. For these treatments, the NNT
may be considered indicative of their potential effect in
the pertinent pain condition. In painful polyneuropathy,
selective serotonin reuptake inhibitors, mexiletine and
capsaicin, are clearly less effective than the established
treatment, whereas gabapentin, tramadol, dextromethorphan and l-dopa appear to be equally effective with the
older treatments. The new treatments are generally better
tolerated and therefore the equally effective of them may be
considered superior to the established treatments and even
the less effective may be useful in situations where other
treatments cannot be used. Among the new drug types,
dextromethorphan has the most favourable NNT, but it
still is not superior to optimally dosed tricyclics.
In the other pain conditions, tricyclic antidepressants also
seem to be superior with respect to effect, although oxycodone is equally effective in postherpetic neuralgia and in
trigeminal neuralgia carbamazepine and lamotrigine are
the only treatments that have been tested in adequately
designed trials.
4.3. NNT values for specic pain conditions
In the present review, it is seen that for diabetic neuropathy, an NNT value between 24 is obtained for the tricyclic antidepressants, the sodium channel blockers phenytoin
and carbamazepine, the dopamine precusor L-dopa, the
weak opioid tramadol and for gabapentin, whereas selective
serotonin reuptake inhibitors and mexiletine are either ineffective or have substantially higher NNT values. Tricyclics
are known to exert a multitude of actions such as a monoamine reuptake inhibition (Hall and Ogren, 1981), sodium
and calcium channel blockade (Lavoie et al., 1990; Pancrazio et al., 1998), an antihistaminergic action (Taylor and
Richelson, 1980) and a weak NMDA receptor-blocking
action (Reynolds and Miller, 1988). Carbamazepine, on
the other hand, exerts probably solely a sodium channelblocking effect, yet both drugs have almost similar NNT
value. The reason for this similarity in NNT between
drugs despite differences in mechanism of action is not
clear. It is possible that the NNT value is in fact a sufcient
measure to predict outcome regardless of the differential
mechanisms of action of these drugs. Alternatively, the

397

NNT value may be an inaccurate measure for separating

different drugs on a specic pathological pain condition.
One way to optimise the sensitivity of NNT to detect differences between drugs must be to calculate it separately for
different aspects of pain (steady pain, pain paroxysms,
touch-evoked pain etc.), which is not possible from the
available studies. The observation that many different
drug classes have similar NNT in a pain condition supports
the notion that other factors than underlying pathological
mechanism dictate treatment response, since these drugs
all have different mechanisms of action.
The present data analysis shows that treatment efcacy
with tricyclic antidepressants depends on the dosage policy.
Thus, by optimising the dose according to plasma drug
concentrations, the NNT in painful polyneuropathy could
be reduced signicantly. Some of the NNT values obtained
in studies with dosing according to effect and side effects
may primarily reect the tolerability of the drugs instead of
their potential efcacy. This may explain why studies with
dosing according to effect and side effects fail to nd a
difference in efcacy between tricyclics with both serotonin
and noradrenaline reuptake inhibition and tricyclics with
relatively selective inhibition of noradrenaline reuptake
(Max et al., 1992), while this is found when xed doses
are used (Sindrup et al., 1990b; Vrethem et al., 1997). It
could also be the reason for the nding that in postherpetic
neuralgia there is a reversed relation between the efcacy of
these subgroups of tricyclics (Table 2).
For many drugs, the optimal dose interval is not known
and a comparison of the NNT values is only meaningful if it
can be related to the corresponding NNH (H.J. McQuay,
personal communication). In this respect, it is of interest
to note that the sodium channel blocker mexiletine has a
higher NNT in painful diabetic neuropathy than the other
sodium channel blockers carbamazepine and phenytoin.
This may reect a less effective drug or a side effect prole
that does not allow optimal dosing with respect to efcacy.
A more appropriate measure for clinical purposes in the
future may be one that uses the ratio of NNT and NNH.
Drugs with a low NNT/NNH ratio will generally be superior
to drugs with a high NNT/NNH ratio.

4.4. NNT values for specic drugs

As also seen from the present survey, when looking at
individual drug groups, the NNT value for TCAs is almost
identical across different painful conditions such as postherpetic neuralgia, painful polyneuropathy and peripheral
nerve injury pain. Again, from available data, it is not possible to determine whether this apparent failure to distinguish
between pain conditions on the basis of NNT reects an
insufciency in the NNT measure to detect differences
between conditions. In the listed conditions, a combination
of phenomena including allodynia, paroxysms and sympathetic dysfunction may be present and probably in different

398

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

degrees. While the pathology from one condition to another

may differ, the combination of phenomena seen in one
group of patients may not necessarily differ from that seen
in another group. So, in these cases, identical NNT may be
seen even if different pathologies are involved. It is possible
that introduction of more sensitive scales to monitor effect
can `dissect' the various phenomena and clarify the contribution of different phenomena to a global effect.
However, some differences within drug class between
different pain conditions are observed. Dextromethorphan
relieves painful diabetic neuropathy, but not postherpetic
neuralgia and central post-stroke pain, and SSRIs relieve
diabetic neuropathy slightly but have no effect on poststroke pain. This may reect differences in pain mechanism,
but also differences in sensitivity of the trial and maybe the
severity of pain in the pertinent pain condition.
4.5. Current classication of neuropathic pain
Neuropathic types of pain are heterogenous in terms of
underlying pathology and anatomical site in the nervous
system. They do have a common core of essential phenomena, which are met by many neuropathic patients and seen in
various combinations depending on the particular type of
neuropathic pain. So far, neuropathic pains have been classied according to the type of underlying pathology, e.g.
diabetes, herpes zoster or stroke, or into peripheral versus
central lesions. However, this distinction does not take into
account the possible mechanisms that may be present, e.g.
touch-evoked allodynia due to A-beta ber recruited central
sensitisation, C-ber mechanosensitivity and sympathetic
hyperactivity. Experimental and recent small clinical trials
have shown that touch-evoked pain or pain responses can be
reduced by NMDA receptor antagonists with known action
on central hyperexcitability (Dray et al., 1994; Eide et al.,
1994; Felsby et al., 1996). Hyperalgesic phenomena can
also be reduced by sodium channel blockers like lidocaine.
At present, there are no data available to show whether such
an approach can be applied to patients with different neuropathic pain conditions. Animal studies on experimental
painful neuropathy in which a single neuropathic phenomenon like touch-evoked avoidance behaviour or paw withdrawal threshold to graded von Frey hairs are recorded have
permitted separation of drugs with an action mainly on allodynia-like pain or on pin prick hyperalgesia (for review
see:Yaksh, 1997). It is to be noted, however, that the simple
and often single abnormality recorded in behavioral animal
studies (e.g. paw-withdrawal to von Frey hair stimulation)
rarely are matched by a similar simple picture in the clinic.
An extensive examination of the various phenomena
encountered in neuropathic patients with specic conditions, and how these are modied by drugs, will in the future
tell us whether a mechanism-based classication can be
applied to NNT. Such an approach may also be useful if
combination of treatments are used.

5. Conclusion and perspectives

Despite the increasing number of trials of different drugs
in different neuropathic pain conditions, it is still only the
efcacy of tricyclic antidepressants in painful polyneuropathy and postherpetic neuralgia that relies on a sufciently
large total number of patients studied. Some of the studies
on the new treatments, such as tramadol in painful polyneuropathy, and gabapentin in painful polyneuropathy and
postherpetic neuralgia may rely on an adequate number of
patients, but they are based on only one or two trials. None
of the new treatments appear to be more effective than the
tricyclics or the older anticonvulsants, and optimal dosing of
the tricyclics may improve the efcacy of these drugs.
However, the new treatments may represent a progress
due to their supposed better tolerability. The NNT measure
of efcacy did not reveal any large or consistent differences
in effect across different pain conditions for each drug class
or by different drug classes in single pain condition. Establishment of NNT and NNH values for each type of drug and
for various pain conditions will be a useful clinical guide.
Future studies on the effect of different drugs on different
neuropathic pain phenomena may unmask pain mechanisms
and guide choice of treatment for single patients. In this
respect it may be useful to develop new scales that permit
a quantication of different phenomena. Pre-emptive treatment with e.g. tricyclic antidepressants may be a new
research direction.
Acknowledgements
This study was supported by a grant from the Danish
Medical Research Council (Grant no. 42820). We would
like to thank Dr. H.J. McQuay for helpful comments on
an earlier draft version of this paper.
References
Arner S, Meyerson BA. Lack of effect of opioids on neuropathic and idiopathic forms of pain. Pain 1988;33:1123.
Backonja M, Beydoun A, Edwards KR, Schwartz SL, Fonseca V, Hes M,
LaMoreaux L, Garofalo E. for the Gabapentin Diabetic Neuropathy
Study Group. Gabapentin for the symptomatic treatment of painful
neuropathy in patients with diabetes mellitus. A randomised controlled
trial. J Am Med Assoc 1998;280:18311836.
Bennett GJ. Neuropathic pain. In: Wall PD, Melzack PD, Melzack R,
editors. Textbook of Pain, 3rd ed. Edinburgh: Churchill Livingstone,
1994. pp. 494510.
Bernstein JE, Korman NJ, Bickers DR, Dahl MV, Millikan LE. Topical
capsaicin treatment of chronic postherpetic neuralgia. J Am Acad
Dermatol 1989;21:265270.
Bowsher D. The effects of pre-emptive treatment of postherpetic neuralgia
with amitriptyline: a randomized, double-blind, placebo-controlled
trial. J Pain Symptom Manage 1997;13:327331.
Broadley KE, Kurowska A, Tookman A. Ketamine injection used orally.
Palliat Med 1996;10:247250.
Campbell FG, Graham JG, Zilkha KJ. Clinical trial of carbamazepine

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

(Tegretol) in trigeminal neuralgia. J Neurol Neurosurg Psychiatry
1966;29:265267.
Canavero S, Bonicalzi V. Lamotrigine control of central pain. Pain
1996;68:179181.
Capsaicin Study Group. Treatment of painful diabetic neuropathy with
topical capsaicin. A multicenter, double-blind, vehichle-controlled
study. Arch Intern Med 1991;151:22252229.
Chad DA, Aronin N, Lundstrom R, McKeon P, Ross D, Molitch M, Schipper HM, Stall G, Dyess E, Tarsy D. Does capsaicin relieve the pain of
diabetic neuropathy. Pain 1990;42:387388.
Chadda VS, Mathur MS. Double blind study of the effects of dipenylhydantoin sodium on diabetic neuropathy. J Asso Phys Ind 1978;26:403
406.
Chiou-Tan FY, Tuel SM, Johnson JC, Priebe MM, Hirsh DD, Strayer JR.
Effect of mexiletine on spinal cord injury dysesthetic pain. Am J Phys
Med Rehab 1996;75:8487.
Cook RJ, Sackett DL. The number needed to treat: a clinically useful
measure of treatment effect. Br Med J 1995;310:452454.
Dejgard A, Petersen P, Kastrup J. Mexiletine for treatment of chronic
painful diabetic neuropathy. Lancet 1988;2:911.
Dellemijn PLI, Vanneste JAL. Randomised double-blind active-controlled
crossover trial of intravenous fentanyl in neuropathic pain. Lancet
1997;349:753758.
Devor M, Wall PD, Catalan N. Systemic lidocaine silences ectopic
neuroma and DRG discharge without blocking nerve conduction.
Pain 1992;48:261268.
Dray A. Agonists and antagonists of nociception. In: Jensen TS, Turner JA,
Wiesenfeld-Hallin Z, editors. Proceedings of the 9th World Congress
on Pain, Seattle: IASP press, 1997. pp. 279292.
Dray A, Urban L, Dickenson A. Pharmacology of chronic pain. Trends
Pharmacol Sci 1994;15:190197.
Eide PK, Jrum E, Stubhaug A, Bremnes J, Breivik H. Relief of postherpetic neuralgia with the NMDA receptor antagonist ketamine: a doubleblind cross-over comparison with morphine and placebo. Pain
1994;58:347354.
Eisenberg E, Alon N, Ishay A, Daoud D, Yarnitsky D. Lamotrigine in the
treatment of painful diabetic neuropathy. Eur J Neurol 1998a;5:167
173.
Eisenberg E, Kleiser A, Dortort A, Haim T, Yarnitsky D. The NMDA (Nmethyl-D-aspartate) receptor antagonist memantine in the treatment of
postherpetic neuralgia: a double-blind, placebo-controlled study. Eur J
Pain 1998b;2:321327.
Ertas M, Sagduyu A, Arac N, Uludag B, Ertekin C. Use of levodopa to
relieve pain from painful symmetrical diabetic polyneuropathy. Pain
1998;75:257259.
Felsby S, Nielsen J, Arendt-Nielsen L, Jensen TS. NMDA receptor blockade in chronic neuropathic pain: a comparison of ketamine and magnesium chloride. Pain 1996;64:283291.
Fields HL. Pain syndromes in neurology, London: Butterworths, 1990.
Fromm GH, Terrence CF. Comparison of L-baclofen and racemic baclofen
in trigeminal neuralgia. Neurology 1987;37:17251728.
Fromm GH, Terrence CF, Chattha AS. Baclofen in the treatment of trigeminal neuralgia: double-blind study and long term follow up. Ann Neurol
1984;15:240244.
Gomez-Perez FJ, Rull JA, Dies H, Rodriguez-Rivera G, Gonzlez-Barranco
J, Lozano-Castaeda O. Nortriptyline and uphenazine in the symptomatic treatment of diabetic neuropathy. A double-blind cross-over
study. Pain 1985;23:395400.
gren SO. Effects of antidepressant drugs on different receptors in
Hall H, O
brain. Eur J Pharmacol 1981;70:393407.
Harati Y, Gooch C, Swenson M, Edelman S, Greene D, Raskin P, Donofrio
P, Cornblath D, Sachdeo R, Siu CO, Kamin M. Double-blind randomized trial of tramadol for the treatment of pain of diabetic neuropathy.
Neurology 1998;50:18421846.
Jensen TS. Mechanism of neuropathic pain. In: Campbell JN, editor. Pain,
1996, an updated review, Seattle: IASP Press, 1996. pp. 7786.
Jensen TS, Yaksh TL. Effects of an intrathecal dopamine agonist, apomor-

399

phine, on thermal and chemical evoked noxious responses in rats, Brain

Res 1984;296:285293.
Kalso E, Tasmuth T, Neuvonen PJ. Amitriptyline effectively relieves
neuropathic pain following treatment of breast cancer. Pain
1995;64:293302.
Kastrup J, Petersen P, Dejgard A, Angelo H, Hilsted J. Intravenous lidocaine infusiona new treatment of chronic painful diabetic neuropathy?
Pain 1987;28:6975.
Kembaum S, Hauchecome J. Administration of levodopa for relief of
herpes zoster pain. J Am Med Assoc 1981;246:132134.
Killian JM, Fromm GH. Carbamazepine in the treatment of neuralgia. Use
and side effects, Arch Neurology 1968;19:129136.
Kvinesdal B, Molin J, Frland A, Gram LF. Imipramine treatment of painful diabetic neuropathy. J Am Med Assoc 1984;251:17271730.
Lavoie PA, Beauchamp G, Elie R. Tricyclic antidepresants inhibit voltagedependent calcium channel and Na 1-Ca 21 exchange in rat brain
synaptosomes. Can J Physiol Pharmacol 1990;68:14141418.
Leijon G, Boivie J. Central post-stroke paina controlled trial of amitriptyline and carbamazepine. Pain 1989;36:2736.
Low PA, Opfer-Gehrking TL, Dyck PJ, Litchy WJ, O'Brien PC. Doubleblind, placebo-controlled study of the application of capsaicin cream in
chronic distal painful polyneuropathy. Pain 1995;62:163168.
Max MB, Culnane M, Schafer SC, Gracely RH, Walther DJ, Smoller B,
Dubner R. Amitriptyline relieves diabetic neuropathy pain in patients
with normal or depressed mood. Neurology 1987;37:589596.
Max MB, Kishore-Kumar R, Schafer SC, Meister B, Gracely R, Smoller B,
Dubner R. Efcacy of desipramine in painful diabetic neuropathy: a
placebo-controlled trial. Pain 1991;45:39.
Max MB, Lynch SA, Muir J, Shoaf SE, Smoller B, Dubner R. Effects of
desipramine, amitriptyline, and uoxetine on pain in diabetic neuropathy. N Eng J Med 1992;326:12501256.
Max MB, Byas-Smith MG, Gracely RH, Bennett GJ. Intravenous infusion
of the NMDA antagonist, ketamine, in chronic post-traumatic pain and
allodynia: a double-blind comparison to alfentanil and placebo. Clin
Neuropharmacol 1995;18:360368.
McQuay HJ, Moore RA. An evidence based resource for pain relief,
Oxford: Oxford University Press, 1998.
McQuay HJ, Carroll D, Glynn CJ. Dose-response for analgesic effect of
amitriptyline in chronic pain. Anaesthesia 1993;48:281285.
McQuay HJ, Jadad AR, Carroll D, Faura C, Glynn CJ, Moore RA, Liu Y.
Opioid sensitivity of chronic pain: a patient-controlled analgesia
method. Anaesthesia 1992;47:757767.
McQuay HJ, Carroll D, Jadad AR, Glynn CJ, Jack T, Moore RA, Wiffen PJ.
Dextromethorphan for the treatment of neuropathic pain: a double-blind
randomised controlled crossover trial with integral n-of-1 design. Pain
1994;59:127133.
McQuay H, Carroll D, Jadad AR, Wiffen P, Moore A. Anticonvulsant drugs
for management of pain: a systematic review. Br Med J
1995;311:10471052.
McQuay HJ, Tramer M, Nye BA, Carroll D, Wiffen P, Moore A. A
systematic review of antidepressants in neuropathic pain. Pain
1996;68:217227.
Moore RA, Gavaghan D, Tramer MR, Collins SL, McQuay HJ. Size is
everything - large amounts of information are needed to overcome
random effects in estimating direction and magnitude of treatment
effects. Pain 1998;78:209216.
Nelson KA, Park KM, Robinovitz E, Constantine T, Max MB. High-dose
oral dextromethorphan versus placebo in painful diabetic neuropathy
and potherpetic neuralgia. Neurology 1997;48:12121218.
Nicol CF. A four year double blind study of tegretol in facial pain. Headache 1969;9:5457.
Nikolajsen L, Hansen PO, Jensen TS. Oral ketamine therapy in the treatment of post-amputation stump pain. Acta Anaesthesiol Scand
1997;41:329331.
Oskarsson P, Lins PE, Ljunggren JG, Mexiletine Study Group. Efcacy and
safety of mexiletine in the treatment of painful diabetic neuropathy.
Diabetes Care 1997;20:15941597.

400

S.H. Sindrup, T.S. Jensen / Pain 83 (1999) 389400

Pancrazio JJ, Kamatchi GL, Roscoe AK, Lynch C. Inhibition of neuronal

Na 1 channels by antidepressant drugs. J Pharmacol Exp Ther
1998;284:208214.
Portenoy RK, Foley KM, Inturrisi CE. The nature of opioid responsiveness
and its implications for neuropathic pain: new hypotheses derived from
studies of opioid infusions. Pain 1990;43:273286.
Reynolds IJ, Miller RJ. Tricyclic antidepressants block N-methyl-D-aspartate receptors: similarities to the action of zinc. Br J Pharmacol
1988;95:95102.
Rosenberg JM, Harrell C, Ristic H, Werner RA, de Rosayro AM. The effect
of gabapentin in neuropathic pain. Clin J Pain 1997;13:251255.
Rowbotham M, Harden N, Stacey B, Bernstein P, Magnus-Miller L. for the
Gabapentin study group. Gabapentin for the treatment of postherpetic
neuralgia: a randomized controlled trial. J Am Med Assoc
1998;280:18371842.
Rowbotham MC, Reisner-Keller LA, Fields HL. Both intravenous lidocaine and morphine reduce pain of postherpetic neuralgia. Neurology
1991;41:10241028.
Rull JA, Quibrera R, Gonzalez-Millan H, Castaneda OL. Symptomatic
treatment of peripheral diabetic neuropathy with Carbamazepine
(Tegretol w): double blind crossover trial. Diabetologia 1969;5:215
218.
Saudek CD, Werns S, Reidenberg M. Phenytoin in the treatment of diabetic
symmetrical polyneuropathy. Clin Pharmacol Ther 1977;22:196199.
Schefer NM, Sheitel PL, Lipton MN. Treatment of painful diabetic neuropathy with capsaicin. J Am Pediatr Med Assoc 1991;31:288293.
Sindrup SH. Antidepressants in the treatment of diabetic neuropathy symptoms. Pharmacodynamic, -kinetic, and -genetic aspects, Dan Med Bull
1994;41:6678.
Sindrup SH, Gram LF, Brsen K, Eshj O, Mogensen EF. The selective
serotonin reuptake inhibitor paroxetine is effective in the treatment of
diabetic neuropathy symptoms. Pain 1990a;42:135144.
Sindrup SH, Gram LF, Skjold T, Grodum E, Brsen K, Beck-Nielsen H.
Clomipramine vs. desipramine vs. placebo in the treatment of diabetic
neuropathy symptoms. A double-blind cross-over study. Br J Clin Pharmacol 1990b;30:683691.
Sindrup SH, Gram LF, Skjold T, Frland A, Beck-Nielsen H. Concentration-response relationship in imipramine treatment of diabetic neuropathy symptoms. Clin Pharmacol Ther 1990c;47:509515.
Sindrup SH, Tuxen C, Gram LF, Grodum F, Skjold T, Brsen K, BeckNielsen H. Lack of effect of mianserin on the symptoms of diabetic
neuropathy. Eur J Clin Pharmacol 1992a;43:251255.

Sindrup SH, Bjerre U, Dejgaard A, Brssen K, Aaes-Jrgensen T, Gram LF.

The selective serotonin reuptake inhibitor citalopram relieves the symptoms of diabetic neuropathy. Clin Pharmacol Ther 1992b;52:547552.
Sindrup SH, Andersen G, Madsen C, Brsen K, Smith T, Jensen TS.
Tramadol relieve pain and allodynia in polyneuropathy. A randomised,
double-blind, placebo-controlled trial. Pain 1999;83:8590.
Stracke H, Meyer UF, Schumacher HF, Federlin K. Mexiletine in the treatment of diabetic neuropathy. Diabetes Care 1992;15:15501555.
Tandan R, Lewis GA, Krusinski PB, Badger GB, Fries TJ. Topical capsaicin in painful diabetic neuropathy - controlled study with long-term
follow-up. Diabetes Care 1992;15:814.
Tasmuth, Kalso. 21st Annual meeting of the Scandinavian Association for
the study of Pain, 1998.
Taylor JF, Richelson F. High afnity binding of tricyclic antidepressants to
histamine H1-receptors: fact and artifact. Eur J Pharmacol 1980;67:41
46.
Vestergaard K, Andersen G, Jensen TS. Treatment of central post-stroke
pain with a selective serotonin reuptake inhibitor. Eur J Neurol
1996;3(suppl. 5):169.
Vrethem M, Boivie J, Arnqvist H, Holmgren H, Lindstrom T, Thorell LH.
A comparison of amitriptyline and maprotiline in the treatment of painful polyneuropathy in diabetics and non-diabetics. Clin J Pain
1997;13:313323.
Watson CPN, Babul N. Efcacy of oxycodone in neuropathic pain: a randomized trial in postherpetic neuralgia. Neurology 1998;50:18371841.
Watson CPN, Evans RJ. The post-mastectomy pain syndrome and topical
capsaicin: a randomised trial. Pain 1992;51:375379.
Woolf CJ, et al. Towards a mechanism-based classication of pain. Pain
1998;77:227229.
Wright JM, Oki JC, Graves L. Mexiletine in the symptomatic treatment of
diabetic peripheral neuropathy. Ann Pharmacother 1997;31:2934.
Yaksh TL. Preclinical models of nociception. In: Yaksh TL, Lynch C,
Zapol WM, Biebuyck JF, Saidman LJ, editors. Anesthesia: biological
foundations, Philadelphia: Lippencott-Raven Publishers, 1997. pp.
685718.
Zakrzewska JM, Patsalos PN. Oxcarbazepine: a new drug in the management of intractable trigeminal neuralgia. J Neurol Neurosurg Psychiatry
1989;52:472476.
Zakrzewska JM, Chaudhry Z, Nurmikko TJ, Patton DW, Mullens EL.
Lamotrigine (Lamictal) in refractory trigeminal neuralgia: results
from a double-blind placebo controlled trial. Pain 1997;73:223230.