NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Selph S, Carson S, Fu R, et al. Drug Class Review: Neuropathic Pain: Final Update 1 Report [Internet]. Portland (OR): Oregon Health & Science University; 2011 Jun.

Cover of Drug Class Review: Neuropathic Pain

Drug Class Review: Neuropathic Pain: Final Update 1 Report [Internet].

Show details

Results

Overview

Overall, 128 studies were included in this report (55 were identified in searches conducted for Update 1). Figure 1 shows the flow of study selection. We received dossiers from 5 pharmaceutical manufacturers: Eli Lilly, Endo, OMJUS, Ortho McNeil, and UCB. Twenty studies that were included in the original report were excluded in Update 1 either because they were outdated (8 systematic reviews) or because the inclusion criteria had changed. See Appendix D for a list of excluded studies and reasons for exclusion at full-text screening. Of the included studies, 14 were direct comparisons of drugs in this review. The remainder was placebo-controlled, observational, or systematic reviews.

Figure 1. Results of literature search.

Figure 1

Results of literature search.

Results of Search: Randomized Trials

We identified 14 head-to-head trials, 7 of which compared amitriptyline to carbamazepine, gabapentin, lamotrigine, or pregabalin.37–43 Two trials compared gabapentin to nortriptyline44, 45 and 1 to imipramine.46 There were 2 trials comparing venlafaxine to imipramine or carbamazepine.47, 48 There was 1 trial each of pregabalin compared with lidocaine and pregabalin compared with duloxetine.49, 50 Seven (50%) out of 14 the trials were parallel 40, 42, 45–47, 49, 50 and of patients with diabetic neuropathy.37, 38, 40, 42, 43, 47, 50 Fourteen percent of trials were mixed, comprising diabetic neuropathy, post herpetic neuralgia, and polyneuropathy.44, 49 There was 1 trial each of patients with central post stroke pain,39 spinal cord injury,41 polyneuropathy,48 post herpetic neuralgia,45 and cancer.46 Sample sizes of the head -to-head trials ranged from 15 to 407 patients. Most trials were short -term with duration of therapy ranging between 1 and 18 weeks. Two trials were rated poor quality 41, 50 while the remainder was rated fair. Included placebo controlled trials are summarized in Table 3 below. We did not find any includable randomized trials of milnacipran, protriptyline, or phenytoin for neuropathic pain.

Table 3. Overview of included placebo-controlled trials.

Table 3

Overview of included placebo-controlled trials.

Effectiveness compared with Efficacy

We considered all of the trials included in this report efficacy studies, as none met all criteria for effectiveness studies.51 The trials generally applied numerous inclusion criteria, were conducted in specialty settings, used rigid dosing regimens, and evaluated relatively short-term and poorly standardized outcomes.

Key Question 1. What is the comparative effectiveness of anticonvulsants, tricyclic antidepressants, serotonin–norepinephrine reuptake inhibitors (SNRIs), and the lidocaine patch for neuropathic pain?

Summary of Findings

Diabetic neuropathy and postherpetic neuralgia

  • Based on very small studies, moderate-strength direct evidence did not support a statistically significant difference between gabapentin, pregabalin, and lamotrigine compared with tricyclic antidepressants in the rate of response, defined as a 50% or more reduction in baseline pain analyzed individually or when pooled (relative risk, 1.0;95% CI, 0.84 to 1.18)
  • Low-strength evidence indicated that lidocaine 5% medicated patch was not statistically different to oral pregabalin in 50% pain reduction in the short term (relative risk, 1.21; 95% CI, 0.88 to 1.67)
  • Using only adjusted indirect comparisons:
  • Three drugs (divalproex, oxcarbazepine, and topiramate) had no direct comparative evidence and 1 drug (divalproex) had inadequate data to conduct an indirect analysis; all of these drugs were found superior to placebo in short -term trials.

Other types of neuropathic pain

Direct evidence
  • In patients with cancer-related neuropathic pain, no difference in pain relief with low-dose gabapentin (400 mg or 800 mg) plus opioids compared to low-dose imipramine (10 mg) plus opioids; combination with gabapentin plus imipramine plus opioids was more effective than therapy with either gabapentin plus opioids or imipramine plus opioids
  • In patients with spinal cord injury, amitriptyline was more effective for pain relief than gabapentin; when data were analyzed in subgroups based on patient’s depression scores, the difference was significant only in the subgroup of patients with the highest levels of depression
  • In patients with central poststroke pain, there was no difference between amitriptyline and carbamazepine
  • There was no direct evidence in patients with HIV-associated neuropathic pain, multiple sclerosis, complex regional pain syndrome, postmastectomy pain syndrome, phantom limb pain, or traumatic nerve injury pain.
Indirect evidence
  • Because of differences among studies in populations, study designs, and outcomes, it was not possible to conduct indirect analyses in patients with other types of neuropathic pain.
Evidence from fair-quality placebo-controlled trials
  • Chemotherapy-induced pain (prophylaxis)
    • Amitriptyline: no difference
    • Carbamazepine: no difference
    • Oxcarbazepine: among patients with advanced colon cancer who completed treatment (32/40), treatment reduced the occurrence of neuropathic pain (31.2% compared with 75%; P=0.03)
  • Chemotherapy-induced pain (treatment)
  • HIV-associated neuropathic pain
    • Amitriptyline: no difference (2 trials)
    • Lamotrigine: effective only in the subgroup of patients exposed to neurotoxic antiretrovirals (2 trials)
    • Gabapentin: decrease from baseline in mean pain score (−44.1%; P<0.05); analysis by ANCOVA did not show a significant difference between treatment and placebo groups
    • Pregabalin: no difference in pain score or response
  • Spinal cord injury
  • Central poststroke pain
    • Lamotrigine: pain scores decreased, but there was no difference in pain affecting daily activities or use of rescue medication
    • Pregabalin: no difference in pain score or response
  • Multiple sclerosis
  • Central pain (mixed)
    • Pregabalin: decrease in mean pain score and improved quality of life; no difference between subgroups with pain due to brain injury compared with spinal cord injury
  • Complex regional pain syndrome
    • Gabapentin: in 1 crossover trial, reduction in pain in the first treatment period; no difference in function or quality of life
  • Phantom limb pain
    • Amitriptyline: not effective
    • Gabapentin: decrease in pain; no difference in use of rescue medication or activities of daily living
  • Postmastectomy pain syndrome
    • Amitriptyline: not effective
    • Levetiracetam: no difference in pain score, response, or use of rescue medication
    • Venlafaxine: no difference in average pain intensity; pain relief significantly better with treatment; 73% had at least 50% pain relief
  • Traumatic nerve injury pain
  • Trigeminal neuralgia
  • Chronic lumbar radiculopathy
  • Polyneuropathy or mixed populations
    • Levetiracetam: not effective for pain or physical function
    • Valproic acid: not effective for pain relief
    • Amitriptyline: reduced pain in mixed group of patients with diabetic and nondiabetic polyneuropathy
    • Gabapentin: reduced pain and improved some quality of life measures

Detailed Assessment

Diabetic neuropathy and postherpetic neuralgia

In 10 publications and 1 unpublished study40 we identified 4 trials comparing gabapentin with amitriptyline or nortriptyline, 2 trials comparing pregabalin with amitriptyline, 1 trial comparing venlafaxine with carbamazepine, 1 trial comparing 5% lidocaine patch with oral pregabalin, and 1 trial comparing duloxetine with pregabalin.37, 38, 40, 42–45, 47, 49, 50, 52 Nine of these studies were rated fair quality with 1 rated poor quality, partly due to high overall attrition and differential lost to follow-up. 50 See Table 4 for a summary of the 10 head-to-head studies. Most of the trials were conducted outside the United States37, 38, 40, 42, 44, 45, 47, 52 and used adult diabetic patients as subjects,37, 38, 40, 42, 43, 47, 50 although 2 trials were of postherpetic neuralgia patients45, 52 and 1 study had a mixed diabetic/postherpetic neuralgia sample.44 Four studies employed a crossover, rather than a parallel design,37, 38, 43, 44 and 4 were open label trials.37, 42, 50, 52

Table 4. Summary of head-to-head trials with the outcome of ≥50% reduction in pain.

Table 4

Summary of head-to-head trials with the outcome of ≥50% reduction in pain.

No study of gabapentin, pregabalin, or lamotrigine demonstrated individual superiority over the tricyclic antidepressants amitriptyline and nortriptyline. When the 7 studies were pooled in a meta-analysis, there remained no statistically significant difference in 50% or more improvement in pain between amitriptyline/nortriptyline and pregabalin, gabapentin, and lamotrigine (relative risk, 1.00; 95% CI, 0.84 to 1.18; I2=13.3%).37, 38, 40, 42–45 It is possible that nonsignificant difference was influenced by the high attrition in 2 trials,38, 40 baseline differences in gender distribution and baseline pain scores in 1 trial,40 and lack of blinding in 2 trials.37, 42 Other head-to-head comparisons involved single studies only. Venlafaxine was found to be superior to carbamazepine in reducing pain intensity on an 11-point Likert scale in a per protocol analysis in a 2-week study of people with painful diabetic neuropathy (P=0.001). While doses of both drugs were lower than used clinically for other indications, the 25 mg twice daily dose of venlafaxine was relatively closer to the dosage approved by the US Food and Drug Administration than the 100 mg daily dose of carbamazepine (37.5–75 mg venlafaxine 2 to 3 times daily for major depressive disorder compared with 200–400 mg carbamazepine twice daily for trigeminal neuralgia).47 It is possible that this difference may have influenced the findings.

Lidocaine 5% medicated patch was not better than oral pregabalin in achieving at least a 50% reduction in pain from baseline in an open-label study (relative risk, 1.21; 95% CI, 0.88 to 1.67).52,49 While the benefits of the lidocaine patch were greater in patients who had postherpetic neuralgia, rather than diabetic neuropathy, neither group reached statistically significant improvement.

Additionally, there was 1 fair-quality systematic review of the 5% lidocaine medicated patch in diabetic peripheral neuropathy in which a network analysis was conducted to compare the lidocaine patch with amitriptyline, gabapentin, and pregabalin.53 No significant differences were found in pain change from baseline in the lidocaine patch and pregabalin (effect size, 1.43; 95% CI, −2.96 to 5.83), gabapentin (effect size, −0.31; 95% CI, −7.05 to 6.43), and amitriptyline (effect size, 3.48; 95% CI, −0.77 to 7.74). However, there are concerns about the network analysis’ use of a continuous, rather than a dichotomous outcome, where actual counts are known.

Duloxetine and pregabalin were found to have similar impact on pain reduction in a 12-week study of patients who had inadequate pain relief on 900 mg of gabapentin daily using an 11-point Likert Scale. However, we rated this study as poor quality, in part because 43% of the duloxetine patients withdrew from the study compared with 28% of the patients on pregabalin and 22% of the patients taking the combination of duloxetine and gabapentin.50

Adjusted indirect analysis of placebo-controlled trials of tricyclic antidepressants compared with gabapentin, pregabalin, and lamotrigine

An indirect comparison meta-analysis found gabapentin, pregabalin, and lamotrigine to be inferior to the tricyclic antidepressants (relative risk, 0.54; 95% CI, 0.30 to 0.80; relative risk, 0.39; 95% CI, 0.18 to 0.88; and relative risk, 0.30; 95% CI, 0.10 to 0.88, respectively). However, an expressed concern within the neuropathic pain literature is the variability of the placebo response.54–58 Since a drug’s effectiveness against placebo is used to calculate its effectiveness against another active drug during adjusted indirect meta-analysis, the differences in placebo response rate becomes quite important. Factors, which have been suggested to influence the magnitude of the placebo response in neuropathic pain trials, include: unadjusted year of publication,57 duration of the clinical trial, 57 baseline pain scores, 55 and rate of patient recruitment to the study.55 In our pool of studies, the rate of patient recruitment was rarely reported, but we examined the other factors here to determine if these factors are relevant in our data set. This may indicate that indirect comparisons may not be accurate.

In this review, placebo-controlled trials including the outcome ≥50 % pain reduction from baseline, which were published in the 1980s, had a placebo response rate of 6% compared with those published in the 1990s (23%) or published in the years 2000–2005 (20%) and years 2006 to the present (27%). Duration of the trial also seemed to have affected the placebo response in these studies. Trials of gabapentin, pregabalin, lamotrigine, and tricyclic antidepressants that were 6 weeks or less in duration had a placebo response rate of 15% compared with trials between 7 and 10 weeks duration where the placebo response rate was 21% and trials greater than 10 weeks duration, where the placebo response rate was 27%. Perhaps a better way to examine the effect of trial duration on the placebo response is to compare trials of the same drug using different durations: four 8-week trials of pregabalin with placebo response rates of 19% compared with 5 trials of pregabalin of 12–14 weeks duration with placebo response rates of 25%, which is consistent with Quessy’s finding of a weak tendency for the placebo response to increase as the duration of the trial increases.57 While the placebo response increased with duration, the treatment effect of pregabalin in these trials decreased from 56% in trials ≤6 weeks duration, 43% in trials 7–10 weeks duration, and 38% in trials ≥11 weeks. Both patterns may simply represent a regression to the mean as trial duration lengthens. With regard to the possible relationship between baseline pain levels and the placebo response in neuropathic pain trials, the baseline pain levels in trials of gabapentin, pregabalin, and lamotrigine were between 6 and 7 on an 11-point Likert Scale with the exception of 1 trial with a baseline pain score of 5.7.59 However, with the exception of the most recent study with a baseline score of 6.3,40 none of the tricyclic antidepressants trials reported scores on an 11-point Likert Scale. Although the specific year of publication of the study, the duration of the study, and the baseline pain levels appear to predict the placebo response, in studies of gabapentin, pregabalin, lamotrigine, and the tricyclic antidepressants, factors all correlate highly with each other. An important additional factor, not previously mentioned, is study design. All placebo-controlled trials of tricyclics for painful diabetic neuropathy or postherpetic neuralgia in this review are crossover studies. The majority of studies of the other classes of drugs for neuropathic pain are not crossover, making the tricyclic studies different – not only in their placebo response rate, but also in core ways the studies were conducted.

The differences in placebo response rate that we observed, based on year and duration of study, imply that the indirect meta-analysis may not be valid. Our own assessment of the response rates (both placebo and drug) in the older, tricyclic antidepressant studies compared to any newer study also indicated differences that were concerning. Therefore, although the indirect comparisons of gabapentin, pregabalin, and lamotrigine with tricyclic antidepressants significantly favor the tricyclics for providing ≥50% pain relief, we do not feel that these indirect comparisons are valid.

Indirect analysis of other comparisons

Based on 6 placebo-controlled trials of gabapentin,59–65 15 trials of pregabalin,40, 66–80 3 trials of duloxetine,81–83 1 trial of venlafaxine,84 4 trials of lacosamide,85–88 2 trials of oxcarbazepine,89, 90 and 2 studies representing 4 trials of topiramate,91,92 we conducted adjusted indirect comparisons. The primary outcome for comparison was ≥ 50% reduction in pain from baseline pain scores. See Table 5 below.

Table 5. Indirect comparison of pain measured as ≥50% pain reduction.

Table 5

Indirect comparison of pain measured as ≥50% pain reduction.

Duloxetine, pregabalin, and gabapentin were superior to lamotrigine and lacosamide when measuring ≥50% pain relief.

Other indirect comparisons were possible using other, frequently-used neuropathic pain outcomes such as pain rated on an 11-point Likert scale, on a 0–100 visual analogue scale, and on the 0–45 short form of the McGill Pain Questionnaire (SF-MPQ). See table 6 below. (The mean difference indicates that a drug reduces pain scores from baseline more than placebo reduces pain scores. The difference of the difference indicates that a drug reduces pain scores from baseline, better than another drug, after the reductions due to placebo have been taken into account.)

Table 6. Significant indirect comparisons of pain reduction on 3 different scales.

Table 6

Significant indirect comparisons of pain reduction on 3 different scales.

In indirect analysis, duloxetine was again superior to lamotrigine and lacosamide using the 11-point scale. Pregabalin was, again, superior to lamotrigine using the McGill Pain Questionnaire. Pregabalin was also superior to lacosamide on the 11-point scale and the 0–100 visual analogue scale. Gabapentin was, again, superior to lamotrigine on both the 11-point scale and the McGill Pain Questionnaire. In addition, gabapentin was superior to lacosamide on the 11-point Likert scale. No other comparisons were significant using these scales.

Placebo-controlled trials

For drugs with no head-to-head or indirect comparative evidence regarding efficacy or effectiveness available (divalproex and venlafaxine) and no US Food and Drug Administration approval for treatment of neuropathic pain, placebo-controlled trials were reviewed to determine evidence of basic efficacy. Additionally, there were several diabetic neuropathy and postherpetic trials not incorporated into this review due to poor quality,93–100 no longer of an included drug,101 did not report result statistics,102 substituted drugs based on tolerability,103 or based drug dosages on sparteine phenotype.104

As a group divalproex, lacosamide, lamotrigine, oxcarbazepine, and topiramate were superior to placebo in achieving response, defined as at least a 50% reduction in pain (relative risk, 1.34; 95% CI, 1.11 to 1.62). See Table 7 for a summary of these anticonvulsant trials. The individual anticonvulsant drugs with significant results included divalproex105 and topiramate.91

Table 7. Anticonvulsant trials measuring 50% response rate in pain reduction.

Table 7

Anticonvulsant trials measuring 50% response rate in pain reduction.

Although there was only 1 trial of divalproex which included 50% pain reduction as an outcome, there were a total of 3 trials of divalproex.105, 107, 108 All 3 trials of divalproex were small (N≤60).105, 107, 108 Two studies focused on painful diabetic neuropathy,107, 108 while the third trial was of postherpetic neuralgia patients.105 All 3 demonstrated significant pain reduction on 1200 mg daily (P<0.05)107 and 1000 mg daily (P<0.001)108 in diabetic patients and 1000 mg daily (P<0.001) in postherpetic neuralgia patients.105

Of 3 trials of oxcarbazepine, 1 was rated poor quality, in part due to 41% attrition in the oxcarbazepine group compared with 24% in the placebo group and lack of clarity regarding which subjects were analyzed.94 The remaining 2 were fair-quality, 16-week, parallel group trials.89, 90 In 1 of the studies, patients experienced a larger decrease in pain as recorded on a visual analogue scale with oxcarbazepine compared to placebo (P=0.0108) with a mean of 1445 mg daily.90 In the second study, there was no difference using the visual analogue scale between oxcarbazepine at doses of 600 mg daily, 1200 mg daily, and 1800 mg daily compared with placebo, although there was a trend toward significance with the latter 2 doses (P=0.101, P=0.096, respectively).89 One noted difference between the eligibility criteria of the 2 studies was that the first study required an average pain score of 50 on the visual analogue scale over 4 of the last 7 days prior to randomization and the second study required an average visual analogue scale pain rating of 40 during the prerandomization phase. This difference of baseline pain scores may have contributed to the different findings in the 2 fair-quality trials.

The results from 2 publications representing 4 trials of topiramate were mixed.91, 92 In a 12-week trial demonstrating significant pain reduction,91 the mean baseline pain score on a 0 -100 visual analogue scale was 68.4, whereas the mean baseline pain score on the same scale was 57.9 in the 18–22 week trials demonstrating no statistically significant effect.92 Additionally, the differences in trial duration may have contributed to the mixed results.

A single, fair-quality trial of venlafaxine was a parallel study of 6 weeks duration where 150–225 mg daily of extended release venlafaxine showed benefit on a 0–100 visual analogue scale compared to placebo (P<0.001), but 75 mg daily did not.52, 84

Results of pooled analysis of placebo-controlled trials are mentioned here. Studies of tricyclic antidepressants demonstrated superiority in a pooled analysis of trials reporting ≥ 50% pain relief from baseline (relative risk, 4.85; 95% CI, 1.86 to 12.67).40, 109–113 Likewise, pooled placebo-controlled trials of pregabalin (relative risk, 1.87; 95% CI, 1.50 to 2.32), 40, 66–71, 73–80 gabapentin (relative risk, 2.19; 95% CI, 1.79 to 2.68), and duloxetine (relative risk, 1.72; 95% CI, 1.42 to 2.07)81–83 reporting ≥ 50% pain relief from baseline showed effectiveness over placebo.

The lidocaine patch also demonstrated superiority over placebo as a therapy for postherpetic neuralgia. On a 0–100 visual analogue scale, the lidocaine patch averaged a 10.2 mm reduction in pain, which was superior to both observation only (P<0.001 to 0.038, depending on time point) and the placebo patch at 2 hours (P=0.016) and 6 hours (P=0.41). The lidocaine patch was not significantly superior to the placebo patch at other time periods measured.114 Two studies with a primary outcome of “time to exit the study” found that subjects left the study sooner if they had received the placebo patch rather than the lidocaine patch—by 10.2 days in 1 study115 and by 4.5 days in the other.116

Chemotherapy-induced or cancer-related neuropathic pain

Direct evidence

We identified 1 fair-quality head-to-head trial in 52 patients with cancer-related neuropathic pain.46 Patients with pain not controlled by opioids and NSAIDs were randomized to low-dose gabapentin (400 or 800 mg), low-dose imipramine (10 mg), or a combination of the 2. All patients continued their opioids. Gabapentin-imipramine combination treatment significantly reduced total pain score, daily paroxysmal pain episodes, and opioid rescue dose. Monotherapy with low-dose gabapentin or low-dose imipramine did not control pain sufficiently.

Indirect evidence

Eight fair-quality randomized controlled trials compared a drug for neuropathic pain to placebo for prevention or treatment of chemotherapy-induced or cancer-related neuropathic pain.117–124 Three trials, 1 each of amitriptyline, carbamazepine, and oxcarbazepine, were designed to assess the effectiveness of treatment to prevent pain in patients undergoing chemotherapy (Table 8).117, 119, 124 Only 1 of the 3 trials found a significant reduction in neuropathic pain with treatment. Two trials found no difference in the development of neuropathic pain with either amitriptyline or carbamazepine. An open-label trial of oxcarbazepine compared with usual care in patients with advanced colorectal cancer found a reduction in the development of neuropathic pain in patients given oxcarbazepine (31.2% compared with 75%; P=0.03).117 These percentages are for patients who completed treatment (32 of 40, 80%); intent-to-treat results also showed efficacy of oxcarbazepine (P=0.05; data not reported). Severity of pain was also reduced in the oxcarbazepine group (per-protocol results).

Table 8. Placebo-controlled trials of drugs to prevent chemotherapy-induced neuropathic pain.

Table 8

Placebo-controlled trials of drugs to prevent chemotherapy-induced neuropathic pain.

Four fair-quality placebo-controlled trials were conducted in patients with chemotherapy-induced neuropathic pain.118, 120, 122, 123 They included 1 trial each of gabapentin, lamotrigine, amitriptyline, and nortriptyline. None of these found a difference between treatment and placebo in mean pain score, response, or quality-of-life measures.

A fifth trial found gabapentin plus an opioid reduced burning or shooting pain more than an opioid alone.121 The results of this trial may not be valid, however. It was rated poor quality due to lack of blinding of outcome assessment, baseline differences between groups, and no intent-to-treat analysis combined with a 16% withdrawal rate(Table 9 ).

Table 9. Randomized controlled trials of drugs for treatment of chemotherapy-induced and cancer-related neuropathic pain.

Table 9

Randomized controlled trials of drugs for treatment of chemotherapy-induced and cancer-related neuropathic pain.

HIV-associated neuropathic pain

Direct evidence

We identified no head-to-head trials in patients with HIV-associated neuropathic pain.

Indirect evidence

We identified 6 fair-quality placebo-controlled trials of drugs to treat HIV-associated neuropathic pain (Table 10).125–130 Two trials included amitriptyline, 2 included lamotrigine, 1 included gabapentin, and 1 included pregabalin.

Table 10. Placebo-controlled trials of drugs for HIV-associated neuropathic pain.

Table 10

Placebo-controlled trials of drugs for HIV-associated neuropathic pain.

In both amitriptyline trials, there was no difference between treatment and placebo in pain score or response.126, 127 In the 2 lamotrigine trials, treatment was more effective than placebo only in the subgroup of patients who were on neurotoxic antiretroviral treatment.128, 129 No other trials reported data by exposure to neurotoxic antiretrovirals. In the trial of gabapentin, both groups significantly improved from baseline but the difference between groups was not significant.125 Pregabalin was no more effective than placebo in 1 trial.130

Spinal cord injury-related neuropathic pain

Direct evidence

A fair quality, head-to-head crossover trial compared gabapentin (maximum dose 3600 mg) to amitriptyline (maximum dose 150 mg) in 38 patients with spinal cord injury.41 Diphenhydramine (maximum dose 75 mg) was also included as an active placebo so that subjects would think they were getting gabapentin or amitriptyline due to the side effects of diphenhydramine. Twenty-two patients (58%) completed all 3 phases of the trial. Analysis of the 22 completers found average visual analogue scale pain intensity score at week 8 was significantly lower with amitriptyline than with gabapentin (P=0.03) or diphenhydramine (P=0.01). There was no significant difference between gabapentin and diphenhydramine. An analysis by patients’ level of depression found that among those with the lowest levels of depression, there was no difference in pain scores between the 3 groups, however. Among those with the highest levels of depression according to The Center for Epidemiologic Studies Depression Scale-Short Form (CESD-SF), amitriptyline reduced pain scores more than gabapentin or diphenhydramine. Among patients in the low CESD-SF group (score <10), response rates (defined as 30% or more decrease in pain score) were 50% with amitriptyline, 42.9% with gabapentin, and 35.7% with diphenhydramine. Among those in the high CESD-SF group(sc ore ≥10), response rates were 62.5%, 12.5%, and 25%, respectively.

Indirect evidence

A recent, good-quality systematic review summarized the evidence for effectiveness of pharmacologic treatments for pain after spinal cord injury, including anticonvulsants and antidepressants.131 Searches were conducted from 1980 to June 2009; 9 trials were included.41, 132–139 We did not identify any more recent studies that were not included in this review. Methodological quality of studies was rated based on internal validity of studies, with trials assigned a rating of excellent, good, fair, or poor. A level of evidence hierarchy was used to determine the strength of evidence for each intervention. No quantitative meta-analysis was conducted.

Table 11Treatment effectiveness summary for drugs to treat neuropathic pain following spinal cord injury (from Teasell 2010)

Drug (references)Effectiveness
Gabapentin41, 135, 137Effective
Pregabalin136, 138Effective
Lamotrigine134Effective only in subgroup of persons with incomplete spinal cord injury
Valproic acid133Not effective
Amitriptyline41, 132Effective only in depressed persons
Levetiracetam139Not effective

The review concluded that there is Level 1 evidence (based on good-quality randomized controlled trials) of effectiveness of gabapentin and pregabalin in pain after spinal cord injury. Level 1 evidence also showed effectiveness of lamotrigine, but only in persons with incomplete spinal cord injury. There was Level 1 evidence from 1 small trial that valproic acid was not effective, but there was a trend toward improvement in the treatment group. Amitriptyline was effective only in persons with comorbid depression. Levetiracetam was not effective compared with placebo.

Central pain due to stroke or multiple sclerosis

Direct evidence

We identified 1 fair-quality, head-to-head crossover trial of amitriptyline compared with carbamazepine in 15 patients with central poststroke pain.39 After 4 weeks, mean pain intensity scores did not differ between treatment groups (4.2 for both). On the global assessment of change in pain, more patients reported improvement with amitriptyline than carbamazepine (67% vs. 36%), but the difference was not statistically significant.

Indirect evidence

Five fair -quality placebo controlled trials were conducted in patients with central pain due to stroke or multiple sclerosis (Table 12).138, 140–144 One of these is unpublished; its results were provided by the study sponsor.142 One trial of nortriptyline in patients with various types of central pain was rated poor quality.

Table 12. Placebo-controlled trials in patients with central neuropathic pain.

Table 12

Placebo-controlled trials in patients with central neuropathic pain.

Two trials of lamotrigine had mixed results, with 1 finding improvement in pain score in patients with central poststroke pain144 and the other showing no difference from placebo in patients with multiple sclerosis.140 Pregabalin showed no effect on pain score or response in patients with central poststroke pain in 1 trial.142 In a second trial of pregabalin in patients with various kinds of central neuropathic pain, pregabalin was significantly better than placebo for the overall group. There was no difference between the subgroups of patients with pain due to brain injury compared with those with pain due to spinal cord injury.138 Levetiracetam significantly reduced pain score in patients with multiple sclerosis in 1 trial141

Complex regional pain syndrome

We identified 1 placebo controlled crossover trial of gabapentin in 58 patients with Complex Regional Pain Syndrome type I.145 Patients were treated for 3 weeks with a 2 -week washout period in between treatments. All had chronic pain for several years that was refractory to various treatments. The target dose of gabapentin was 1800 mg per day. There was significantly greater pain relief in the first treatment period for gabapentin users. Therapy effect was reduced in the second period and there was no significant effect when the results of both periods were combined. There was no difference between gabapentin and placebo in measures of function or quality of life.

Postmastectomy pain syndrome or phantom limb pain

Five placebo-controlled trials evaluated efficacy of amitriptyline, venlafaxine, levetiracetam, or gabapentin in patients with postmastectomy pain syndrome or phantom limb pain (Table 13).146–150 Three studies found no significant effect of levetiracetam, venlafaxine, or amitriptyline on pain scores.148–150 In the 2 trials finding differences, the effect on pain was mild. A crossover study of amitriptyline reported 53% of patients overall had a decrease in pain intensity, but an analysis between groups is not given.147 The trial of gabapentin found a decrease in pain intensity after 6 weeks of treatment.146 The re was no effect on pain at other time points and no difference between groups in the use of rescue medication, sleep interference, or activities of daily living.

Table 13. Trials in patients with postmastectomy pain syndrome or phantom limb pain.

Table 13

Trials in patients with postmastectomy pain syndrome or phantom limb pain.

Traumatic nerve injury pain

Two placebo controlled trials evaluated efficacy of gabapentin or pregabalin in patients with neuropathic pain due to traumatic peripheral nerve injury.151, 152 Both found treatment more effective than placebo in reducing pain scores.

Trigeminal neuralgia

Six placebo-controlled trials evaluated neuropathic pain medications for trigeminal neuralgia.95, 153–157 Although all 6 trials found carbamazepine (4 trials 95, 153, 154, 156), lamotrigine (1 trial 157), or topiramate (1 trial 155) more effective than placebo, results may not be reliable because 595, 153–156 of the 6 trials were rated poor quality, with 4 of the trials (all of carbamazepine95, 153, 154, 156) published in 1966 or 1968.

Patients with polyneuropathy or mixed populations

Seven studies included patients with polyneuropathy, neuropathy of various etiologies, or did not specify the etiology of pain in the population.48, 97, 98, 158–161 Two of these (1 lamotrigine, 1 lidocaine patch) were rated poor quality97, 98 and the rest were fair.

In 1 small trial comparing venlafaxine with imipramine (N=32), about half of enrolled patients had diabetic neuropathy and half had neuropathic pain due to another etiology. Venlafaxine and imipramine were similar in efficacy on a number of pain scales, with no statistically significant difference in the likelihood of achieving pain relief (relative risk, 0.55; 95% CI, 0.27 to 1.12).48

In a 6-week crossover trial of 35 patients, levetiracetam was no more effective than placebo on measures of pain relief (P=0.979), total pain intensity (P=0.293), or any other outcome measure, including measures of physical function and health-related quality of life.161 In a trial of 37 patients with polyneuropathy, treatment with valproic acid was no more effective than placebo for reducing total pain score (P=0.24).158 More patients experienced pain relief with valproic acid (42% compared with 17%) but the difference was not statistically significant (P=0.13). In a trial including a mixed group of patients with diabetic or nondiabetic polyneuropathy, amitriptyline relieved pain scores more than placebo and was similarly effective in diabetic and nondiabetic patients.160 Gabapentin was more effective than placebo for reducing average pain score and improving some quality of life measures in 1 trial of patients with different neuropathic pain syndromes.159

Chronic lumbar radiculopathy

We identified only 1 placebo-controlled trial in patients with neuropathy associated with lumbar radicular pain.162 Nortriptyline was not effective in reducing average daily leg pain (the primary outcome) or any other leg or back pain scores.

Key Question 2. What are the comparative harms of anticonvulsants, tricyclic antidepressants, SNRIs, and the lidocaine patch for neuropathic pain?

Summary of Findings

Diabetic neuropathy and postherpetic neuralgia

  • Moderate evidence supported a lack of difference in withdrawals due to adverse events between gabapentin, pregabalin, and lamotrigine compared with amitriptyline and nortriptyline (relative risk, 0.61; 95% CI, 0.33 to 1.12)
  • Moderate evidence supported the finding of greater withdrawals due to adverse events of oral pregabalin compared with the 5% lidocaine patch (relative risk, 4.39; 95% CI, 2.25 to 8.69)
  • Moderate evidence indicated that gabapentin or pregabalin (as a group) were less likely to cause dry mouth than tricyclic antidepressants (relative risk, 0.27; 95% CI, 0.14 to 0.56)
  • Low-strength evidence indicated that gabapentin or pregabalin (as a group) were more likely to cause ataxia than tricyclic antidepressants (relative risk, 3.70; 95% CI, 1.18 to 11.65)
  • Using only adjusted indirect comparisons:
    • Low-strength evidence supported a lack of difference in withdrawals due to adverse events between duloxetine, pregabalin, lacosamide, and lamotrigine (with a range of relative risks from 0.82 [95% CI, 0.42 to 1.61] for gabapentin compared with lacosamide to 1.78 [95% CI, 0.91 to 3.48] for duloxetine compared with gabapentin)
    • Low-strength evidence indicated that gabapentin and lamotrigine caused fewer withdrawals due to adverse events than topiramate or oxcarbazepine (with a range of relative risks from 0.44 [95% CI, 0.21 to 0.90] for gabapentin compared with oxcarbazepine to 0.60 [95% CI, 0.37, 0.97] for lamotrigine compared with topiramate)

Other types of neuropathic pain

Direct evidence
  • Evidence was insufficient to evaluate comparative harms in other types of neuropathic pain
  • Among 3 head-to-head trials, 1 reported no withdrawals due to adverse events with either amitriptyline or carbamazepine and the others reported similar proportions of patients withdrawing due to adverse events for amitriptyline or imipramine compared with gabapentin

Detailed Assessment

Painful diabetic neuropathy and postherpetic neuralgia

Withdrawal from study due to an adverse event: direct comparisons

Overall, withdrawal from study due to an adverse event showed wide variability across the studies. Statistically significant differences were found in only 2 comparisons. In a study of 5% lidocaine patch compared with pregabalin, the relative risk of withdrawing due to an adverse event was 4.39 (95% CI, 2.25 to 8.69) times greater in the pregabalin group than the lidocaine group.52 In a smaller study (N=106), lamotrigine had a lower discontinuation rate than amitriptyline (relative risk, 0.42; 95% CI, 0.20 to 0.85).38 See Table 14 for a summary of withdrawals due to adverse events in head-to-head trials.

Table 14. Adverse event withdrawals in head-to-head trials for painful diabetic neuropathy/postherpetic neuralgia.

Table 14

Adverse event withdrawals in head-to-head trials for painful diabetic neuropathy/postherpetic neuralgia.

Adjusted indirect comparison

Withdrawals due to adverse events were higher in the treatment arm, regardless of the drug, than in the placebo arm. See Table 15 for a summary of indirect comparisons of withdrawals due to adverse events. There were no significant differences in the withdrawal rate due to adverse events among duloxetine, pregabalin, gabapentin, and lacosamide trials. There were fewer withdrawals due to adverse events among patients receiving gabapentin or lamotrigine when compared to topiramate or oxcarbazepine.

Table 15. Indirect comparisons of withdrawals due to adverse events.

Table 15

Indirect comparisons of withdrawals due to adverse events.

Most common adverse events

Gabapentin and pregabalin combined were significantly less likely to cause dry mouth than amitriptyline/nortriptyline (relative risk, 0.27; 95% CI, 0.14 to 0.56).37, 40, 43–45 However, gabapentin and pregabalin combined were significantly more likely to cause ataxia than the tricyclic antidepressants (relative risk, 3.70; 95% CI, 1.18 to 11.65).40, 43, 44

There were no deaths or suicide attempts reported in any of the 7 head-to-head studies which included a tricyclic antidepressant arm. Blurred vision was reported in 2 studies of gabapentin (relative risk, 1.56; 95% CI, 0.12 to 20.97).43, 44 There was also 1 instance of pneumonia and 1 instance of cholecystitis in the pregabalin arm of 1 study40 and 4 instances of an elevated creatinine by 25% in the lamotrigine arm of 1 study.38

In the comparison of venlafaxine and carbamazepine, there were a total of 46 adverse events during the trial, 29 adverse events in the venlafaxine group (43.9%) and 17 (25.8%) in the carbamazepine group.47 This difference in total numbers in each group was not significant (P=0.06). In the venlafaxine group the most frequent adverse events were gastrointestinal discomfort (18%), dizziness (14%), and somnolence (12%). In the carbamazepine group, the most frequent adverse events were dizziness (11%) and somnolence (14%). There was also 1 patient in the venlafaxine group who withdrew due to palpitations and 1 patient in the carbamazepine group whose alanine aminotransferase increased from 15 to 121 IU.

Five percent lidocaine medicated patch resulted in significantly fewer adverse events in the lidocaine group (48/155 patients) compared with oral pregabalin (194/153 patients; P<0.0001).52 The most common adverse events occurred in the pregabalin arm: dizziness (12%), fatigue (9%), and vertigo (8%). Five percent of the pregabalin group developed headache compared to 1% in the lidocaine group. One percent of the lidocaine group developed application site reaction.

Adverse events in placebo-controlled trials

The most common adverse events in duloxetine trials were nausea, dizziness, somnolence, constipation, and increased sweating.81, 83 The relative risk for the most common adverse event, nausea, was 1.45 (95% CI, 0.87 to 2.42) at the dose of 20 mg daily;81 2.70 (95% CI, 1.10 to 6.64) at 60 mg daily;81, 83 3.62 (95% CI, 2.12 to 6.16) at 120 mg daily;81, 83 and 2.97 (95% CI, 1.31 to 6.77) regardless of dose.81, 83 Vomiting was reported by 3.4% of duloxetine patients treated with 60 mg twice daily.82 Long-term trials indicated that duloxetine may slightly increase fasting glucose163 or hemoglobin A1C,164 although 1 long-term trial165 showed no effect.

The most common adverse events in the trials of lacosamide, lamotrigine, oxcarbazepine, topiramate, and divalproex were dizziness, nausea, headache, and somnolence.85–92, 108, 166–169 The most common adverse event was dizziness and was reported by 402/3624 (11.1%) study participants (relative risk, 2.46; 95% CI, 1.52 to 3.98).85–91, 100, 166–169

Additionally, 3 drugs were studied in long-term, open label trials or extension studies —duloxetine, lacosamide, and the lidocaine patch. The long-term effects of 60 mg duloxetine twice daily compared to usual care were explored in 3 publications of 2 trials. 163–165 There was no difference in withdrawals due to adverse events between groups in either study (relative risk, 1.49, 95% CI 0.82, 2.71). Two studies examined the long-term effects of lacosamide titrated to a maximum dose of 400 mg daily170 or with a modal dose of 400 mg daily.171 All patients were treated with lacosamide without a placebo or usual care group. In the larger study 18.3% of patients withdrew due to treatment emergent adverse events,171 whereas 23.2% withdrew due to total adverse events in the smaller study.170 The most common adverse events were nausea (13.2%), dizziness (20.2%),171 and headache (16%).170 A 12-month study of the lidocaine patch in postherpetic neuralgia patients found that 12.4% experienced a drug-related adverse event, the most common of which was pruritus at the administration site (2.8%).172 None of the long-term studies noted significant safety concerns with extended use.

Other types of neuropathic pain

There was very little direct evidence available to evaluate comparative harms in patients with other types of neuropathic pain. Among 3 head-to-head trials, 1 reported no withdrawals due to adverse events with either amitriptyline or carbamazepine39 and the others reported similar proportions of patients withdrawing due to adverse events for amitriptyline or imipramine compared to gabapentin.41, 46 Dizziness was more frequent with gabapentin, whereas dry mouth and constipation were more frequently reported with amitriptyline (Table 16).

Table 16. Withdrawals due to adverse events in head-to-head trials of drugs for other types of neuropathic pain.

Table 16

Withdrawals due to adverse events in head-to-head trials of drugs for other types of neuropathic pain.

Key Question 3. Are there differences in effectiveness or harms of anticonvulsants, tricyclic antidepressants, SNRIs, and the lidocaine patch based on demographics, socioeconomic status, comorbidities, or drug-drug interactions, when used to treat neuropathic pain?

Summary of Evidence

  • There was no evidence assessing differences in effectiveness or harms based on demographics, socioeconomic status, comorbidities, or cointerventions
  • Post hoc analyses have not found older age to have an impact on response or treatment emergent adverse events with duloxetine, but older patients withdrew from studies more often than younger patients due to adverse events, regardless of assigned treatment (duloxetine or placebo)
  • Only insufficient to low-strength evidence suggests that combinations of duloxetine and pregabalin; lidocaine patch and pregabalin; or gabapentin with imipramine, nortriptyline, or venlafaxine had a potential benefit compared to monotherapy, but that there was a risk of increased adverse events – although if lower doses of the combined drugs are used, benefits may be seen in both efficacy and harms.

Detailed Assessment

The strength of evidence to answer Key Question 3 was insufficient. We identified no studies addressing differences in effectiveness or harms based on demographics, socioeconomic status, or comorbidities.

Age

In a post hoc analysis of three 12-week placebo-controlled trials of duloxetine (N=701 < 65 years and N=323 ≥65 years), the incidence of treatment emergent adverse events, serious adverse events, and specific adverse events did not differ between patients < 65 years and those ≥ 65 years, regardless of assigned treatment group (placebo or duloxetine 60 mg or 120 mg).173 In all 3 treatment groups, more patients ≥ 65 years withdrew from the studies due to adverse events compared with the younger groups (P<0.001). The rate of withdrawal was highest in the 120 mg daily group (24%). Rates of response, based on 24-hour pain assessments, were similar between age groups.

Cointerventions

We could not address the impact of cointerventions with other drugs on effectiveness or harms of the drugs included in this review because no study analyzed results based on specific cointerventions taken by participants during the study period.

Combination therapy

While 5 studies44, 46, 49, 50, 65 examined combination therapy of drugs included in this review compared with monotherapy, we found the evidence to be insufficient to low strength to answer the question because of the dearth of evidence for a given combination, the small sizes, and only fair- or poor-quality studies. The drug combinations studied were duloxetine and pregabalin; lidocaine patch and pregabalin; and gabapentin with imipramine, nortriptyline, or venlafaxine.

A fair-quality randomized, double-dummy crossover trial (N = 56) found that combination therapy with gabapentin and nortriptyline was superior to monotherapy with either drug in pain control.44 The daily pain intensity was 2.3 (out of 10) on combination therapy compared with 3.2 on gabapentin (P=0.01) and 4.1 with nortriptyline (P=0.02). The mean percentage pain reduction was also greater (52.8% compared with 38.8% and 31.1%, respectively). The total drug exposure period for each assigned treatment was 35 days, but only 5 days were at the maximum tolerated doses. A greater percentage withdrew during a gabapentin monotherapy phase (14%) than either of the other drug assignments (nortriptyline 3.6%, combination 1.8%). Dry mouth was the most common adverse event, significantly more common in the combination and nortriptyline monotherapy groups compared to the gabapentin group (P<0/0001).

In an attempt to determine if lower doses of gabapentin taken with imipramine would result in better pain control and fewer adverse events, a fair-quality pilot study randomized 52 patients with neuropathic pain due to cancer and who were having inadequate response to opioids to low-dose gabapentin (800 mg daily) plus imipramine (20 mg daily), gabapentin 800 mg daily, gabapentin 400 mg daily, or imipramine 20 mg daily for 7 days.46

Although only 7 days long, the results indicated that the combination was superior to the other treatments. In total pain score, the combination therapy resulted in a lower final score (2) than the other groups (4.5 to 5); the difference was statistically significant for the comparison with 400 mg gabapentin and imipramine (P<0.05 for both), but not compared to 800 mg daily of gabapentin. Nausea and drowsiness were seen in all groups, with drowsiness being the most common adverse event across all groups. The 800 mg daily gabapentin group had statistically significantly higher rates of mild and severe dizziness compared with the other groups (P=0.014 and 0.015, respectively).

Following completion of a fair-quality randomized trial of gabapentin or placebo, 11 patients on gabapentin who had not achieved adequate response after 8 weeks were then randomized to treatment with gabapentin (at their maximum tolerated dose) plus venlafaxine (titrated to 150 mg per day) or continuing gabapentin with placebo added.65 In addition to being very small, the study selection criteria were biased against gabapentin monotherapy, which they have already shown not to respond to. After 5 weeks of the maximum tolerated doses, patients who had failed gabapentin monotherapy had more improvement in pain scores with combination therapy (−2 out of 11 points) than those continuing monotherapy (−0.5). Similarly, 75% of patients in the combination group reported much or moderately improved symptoms compared with only 33.3% in the monotherapy group. One of 6 patients in the combination group stopped the study due to adverse events while none in the monotherapy group did.

In a small nonrandomized study (N = 250), patients who had reached the end of a 4-week randomized trial of pregabalin and lidocaine 5% patch continued their originally assigned drug (pregabalin or lidocaine 5%) if they were having clinical response (numerical rating scale over past 3 days of ≤ 4 on a 0 to 10 scale) for 8 more weeks. If their pain was > 4 on the rating scale they were given a combination of the currently assigned drug and the other drug.49 The number of patients who withdrew from the study due to drug-related adverse events was greater in the combination group (5.5%) than in either monotherapy group (lidocaine 1.3%, pregabalin 1.6%). Similarly, the number of patients reporting drug-related adverse events was also greater in the combination drug group (18% compared with 5.1% and 7.9%, respectively).

A poor-quality, 12-week, open-label trial of pregabalin compared with duloxetine, reported only in a poster from a conference presentation to date, found duloxetine noninferior to the combination.50 The study was rated poor quality in part because the limited study report does not provide details on randomization and allocation concealment or on the number of patients included in each of the multiple analyses reported. The unusually high percentage of patients who withdrew from this short-term study (31%) and the open-label nature of this study (given the subjective, patient-assessed outcomes) increase the risk for bias.

Demographics, socio-economic status, or comorbidities

No evidence was found.

Footnotes

a

A modified PRISMA diagram was used.36

b

Numbers in parentheses are results of the literature search new to Update 1.

c

There are 12 trials that were included in original report but excluded now owing to change in eligibility criteria.

d

There are 8 systematic reviews included in original but excluded in Update 1.

Copyright © 2011 by Oregon Health & Science University.
Bookshelf ID: NBK61827