Summary of findings
Neuropathic pain is defined as "pain initiated or caused by a primary lesion or dysfunction of the nervous system" (Merskey 1994) or more recently as "pain arising as a direct consequence of a lesion or disease affecting the somatosensory system" (Treede 2008). Neuropathic pain is difficult to treat because of its severity, chronicity and resistance to simple analgesics. Neuropathic pain may have its genesis in the brain, spinal cord or peripheral nerves and arises from conditions including radiculopathy (cervical or lumbar), malignancy, infection (for example postherpetic neuralgia and human immunodeficiency virus (HIV) related neuropathy), injury to the spinal cord, diabetic neuropathy, trigeminal neuralgia and complex regional pain syndrome type II (Jensen 2001). The epidemiology of neuropathic pain has not been well described, at least in part because the underlying conditions are so diverse. Current pooled estimates suggest that as much as 3% of the population may be affected by neuropathic pain (Davis 2004; Foley 2003; Heliovaara 1987; Schmader 2002; Verma 2005; Werhagen 2004).
The underlying mechanisms of neuropathic pain are complicated and unclear (Finnerup 2007). It is known that the peripheral and central nervous systems are capable of both structural and functional plastic changes in response to injury and experience. Peripheral changes include the sensitisation of nociceptors, which results in decreased activation thresholds and increased pain in response to a given stimulus, and abnormal neuronal sprouting that leads to expansion of nociceptive receptive fields and ectopic firing of dorsal root ganglia cells. Changes in expression of abnormal sodium and calcium channels are considered to be instrumental in the generation of spontaneous discharges from injured neurons (Callin 2008).
Central changes following peripheral nerve injury include loss of inhibitory effects in the central nervous system and heightened sensitivity of neurons in the spinal cord, which occurs in spite of reduced peripheral input. Moreover, N-methyl-D-aspartic acid (NMDA) receptors are upregulated and activated and play an important role in central sensitisation in neuropathic pain. Other important findings include facilitatory descending pathways and the potential relevance of genetic differences between individuals (Callin 2008; Carpenter 2002; Suzuki 2004).
Neuropathic pain has a profound effect on quality of life and expenditure on health care (Schmader 2002). It remains a major therapeutic challenge despite considerable research efforts during the past few decades. The shared pathophysiology of neuropathic pain and epilepsy supports the rationale for using certain antiepileptic drugs (AEDs) in the treatment of neuropathic pain (Dickenson 2002; Jensen 2002). A number of Cochrane reviews assess the effects of individual AEDs in neuropathic pain, including lamotrigine (Wiffen 2011a), carbamazepine (Wifffen 2011b), gabapentin (Moore 2011), pregabalin (Moore 2009), valproic acid (Gill 2011), phenytoin (Birse 2012) and clonazepam (Corrigan 2012). These replace an original review of AEDs for neuropathic pain (Wiffen 2010), first published in 2005 and now withdrawn.
Oxcarbazepine (10,11-dihydro-10-oxo-5H,dibenz[b,f]azepine-5-carboxamide) is the keto analogue of carbamazepine, a sodium channel modulator used primarily in the treatment of epilepsy as well as trigeminal neuralgia. It has previously been reported to be better tolerated and safer than carbamazepine, with a lower risk of allergic reactions and drug-drug interactions (Beydoun 2002a; Dam 1989). Three double-blind, placebo-controlled trials have evaluated oxcarbazepine in painful diabetic neuropathy. Two of them (Beydoun 2006; Dogra 2005) showed statistically significant pain reduction in participants treated with oxcarbazepine compared with placebo, while the other (Grosskopf 2006) showed no difference in efficacy between oxcarbazepine-treated and placebo-treated participants.
Data on the potential efficacy of oxcarbazepine for treating the pain associated with trigeminal neuralgia were derived from an active-control double-blind trial (Lindstrom 1987) and meta-analyses (Beydoun 2002a; Beydoun 2002b) of three double-blind comparative trials. These showed that oxcarbazepine produced substantial pain relief in the great majority of people suffering from trigeminal neuralgia and that there was no significant difference in pain relief between oxcarbazepine and carbamazepine. Additionally, a pooled analysis was performed of seven open-label clinical trials, sharing the same protocol, of oxcarbazepine in different neuropathic pain conditions (Magenta 2005). The results of this analysis suggest that oxcarbazepine administered as monotherapy can relieve pain associated with neuropathies. We undertook this Cochrane review to systematically gather and assess the evidence from randomised controlled trials (RCTs) on the use of oxcarbazepine for all forms of neuropathic pain.
To determine the benefits and harms of oxcarbazepine for different forms of neuropathic pain.
Criteria for considering studies for this review
Types of studies
We included RCTs or cross-over studies of oxcarbazepine for the treatment of neuropathic pain. Trials were eligible whether published or unpublished. There was no language restriction. We interviewed the study authors by telephone or wrote for clarification if the studies were described as RCTs but there was poor reporting of the methodology. We excluded quasi-randomised trials and trials where the reports described them as randomised but which proved not to be randomised on further examination.
Types of participants
Participants of any age or either sex with any neuropathic pain were eligible.
Types of interventions
Oxcarbazepine versus placebo or any other intervention, regardless of administration route, dosage, or length of treatment. We allowed co-interventions if they were offered equally to both arms of the trial.
Types of outcome measures
We planned to measure outcomes as close to eight weeks (minimum six weeks) as possible. When outcomes were reported after different durations of treatment we extracted the data measured as close to the specified time as possible, but not less than the minimum. In fact we could only obtain 16-week follow-up data from the included trials, so we only performed analysis of results at the 16th week.
The primary outcome measure was reduction in patient-reported pain scores by 50% from the baseline, measured on commonly used pain scales such as the Visual Analogue Scale (VAS), the Verbal Rating Scale (VRS), the Numerical Rating Scale (NRS) (Williamson 2005) and the Faces Pain Scale (FPS) (Bieri 1990).
1. Reduction in patient-reported pain scores by 30% from the baseline, measured on commonly used pain scales.
2. Patients' global impression of their change in pain at eight weeks.
3. Overall quality of life measures, as changes or levels at eight weeks using the Short Form-36 Health Survey (SF-36) (Ware 1992), including both the physical health summary, mental health summary and the pain subscale.
4. Adverse effects measures occurring in the eight weeks after start of treatment. Adverse events were classified as any adverse effect, adverse effects which led to withdrawal from treatment, and serious adverse effects which were life threatening, required hospitalisation or were fatal.
Search methods for identification of studies
We searched the Cochrane Neuromuscular Disease Group (CNMDG) Specialized Register (30 October 2012) for randomised trials using the following search terms: (oxcarbazepine or Trileptal or Trexapin) AND (neuropathic pain or cervical radiculopathy or lumbar radiculopathy or diabetic neuropathy or peripheral neuropathy, neuropathic pain or postherpetic neuralgia or HIV-related neuropathy or spinal cord injury or trigeminal neuralgia or complex regional pain syndrome type II AND pain). We adapted this strategy to search the following databases: CENTRAL (2012, Issue 10), MEDLINE (January 1966 to October 2012), EMBASE (January 1980 to October 2012) and the Chinese Biomedical Retrieval System (January 1978 to October 2012). We also searched the National Institutes of Health (NIH) databases and the World Health Organization (WHO) International Clinical Trials Registry Platform for current ongoing registered trials.
Searching other resources
We wrote to the companies (for example Novartis Pharmaceuticals) who made oxcarbazepine and to pain experts asking for information about other or ongoing trials. We examined the Novartis clinical trials results database with the assistance of the CNMDG Managing Editor. We also searched reference lists of review articles and included studies.
Data collection and analysis
Two review authors (MZ and NC) independently selected the studies for inclusion, assessed their risk of bias and extracted relevant data. The CNMDG Managing Editor also assisted in the identification of trial reports by performing an independent review of English language search results. One author (NC) entered the data into the Cochrane statistical software Review Manager 5 (RevMan 5) and the other one (MZ) checked it. The review authors resolved disagreements through discussion.
Selection of studies
Two review authors scrutinised all the titles and abstracts identified by the searches to determine which might fulfil the selection criteria. We obtained full reports of all the potentially eligible studies (when available) to determine if they met the inclusion criteria for the review. We included randomised trials reporting the analgesic effects of oxcarbazepine with subjective assessment of pain of neuropathic origin. We excluded studies that included participants with chronic headache and migraine.
Data extraction and management
We extracted the data from included trials using a specially designed data extraction form. The following data items were extracted.
- Participants: number, age range, gender, type of neuropathic disorder, setting.
- Intervention: dosing regimen, duration, route of administration.
- Control: placebo, other intervention.
- Outcomes: analgesic outcome measures and results, withdrawals and adverse effects (minor and major).
- Design: methods of randomisation, study design (parallel group, cross-over), treatment duration and duration of study follow-up, and whether specifically designed to measure pain.
We also included risk of bias criteria on the data extraction form: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other concerns about bias.
Assessment of risk of bias in included studies
Two authors (MZ and NC) independently assessed risk of bias according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008b; Higgins 2011). The authors resolved any disagreement by discussion.
The authors considered six specific domains (namely sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting and other issues). They assessed the adequacy of each included study in relation to the entry as at low, high or unclear risk of bias.
Measures of treatment effect
Only dichotomous variables regarding our outcome measures were available from the included trials, so we calculated the relative benefit (expressed as risk ratio (RR)) in RevMan with the 95% confidence interval (CI).
Unit of analysis issues
We would have analysed cross-over trials in two ways: firstly using only the data from the first treatment period, and secondly using data from both treatment periods provided that the washout period was adequate. We would have analysed the results with the generic inverse variance (GIV) method.
In addition, we recorded any data relating to patient preference.
Dealing with missing data
We dealt with missing data, such as missing standard deviations, by attempting to contact the authors via email or telephone. If participant dropout led to missing data we conducted an intention-to-treat (ITT) analysis whenever possible. For dichotomous outcomes, we regarded participants with missing outcome data as treatment failures and included these in the analysis. For continuous outcomes, we carried forward the last recorded value for participants with missing outcome data (Higgins 2009).
Assessment of heterogeneity
We performed formal statistical testing of heterogeneity between the trials using the RevMan software.
We presented the pooled outcomes data from all similar studies using risk ratios (RRs) and 95% confidence intervals (CIs).
We also reported results as number needed to treat for an additional beneficial outcome (NNTB) for the specified percentage change in pain scores and number needed to treat for an additional harmful outcome (NNTH) for mild and serious adverse drug reactions.
We planned to perform analyses using a fixed-effect model unless significant heterogeneity was found, when we used a random-effects model. Where appropriate, we combined data from included studies using RevMan.
We also included a 'Summary of findings' (SoF) table, which presented the main findings of the review including information about the quality of the evidence, the magnitude of effects and the sum of the available data on the main outcomes (Schünemann 2008). We included the following outcomes.
- Reduction in patient-reported pain scores by 50% from baseline.
- Reduction in patient-reported pain scores by 30% from baseline.
- Patients with obvious or significant improvement after 16 weeks of treatment.
- Serious adverse events.
Subgroup analysis and investigation of heterogeneity
Where data were available, we planned to undertake subgroup analysis according to the types of neuropathic pain (peripheral neuropathic pain and central neuropathic pain).
We planned to conduct sensitivity analyses to examine the effects of excluding poor quality studies, defined as those with a moderate or high risk of bias as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008). Since all the included trials were rated as of moderate quality, we did not perform any sensitivity analyses in the present review.
Description of studies
Results of the search
The electronic searches retrieved a number of references: 12 from the Cochrane Neuromuscular Disease Group Specialized Register, 71 from MEDLINE, 95 from EMBASE, 14 from CENTRAL, 3 from DARE and 18 from the Chinese Biomedical Retrieval System. From the searches in other resources, including replies from the companies, company databases and the reference lists of found articles, we retrieved five additional references, one of which is an ongoing trial (NCT01302275) and another is a study awaiting classification (Brainer-Lima 2003). From the total of 186 references (excluding duplicates) we identified 36 records reporting the efficacy or safety, or both, of oxcarbazepine for the treatment of different neuropathic pain conditions. Most of them were judged not to meet the entry criteria of the present review because the titles or abstracts indicated that they were open-label clinical trials comparing the efficacy and safety outcomes at the end of studies with the baseline conditions. Of the 21 remaining possibly appropriate records we excluded nine references (see Characteristics of excluded studies). We included four trials (described in nine references) (Beydoun 2006; Dogra 2005; Grosskopf 2006; Novartis 2004) which met the eligibility criteria (see Characteristics of included studies) and two studies are awaiting classification (Brainer-Lima 2003; Liebel 2001). An additional study is ongoing (NCT01302275). See Figure 1 for a flow chart illustrating study selection.
|Figure 1. Study flow diagram.|
Although we tried to identify studies pertaining to various kinds of neuropathic pain three of the included studies, which involved a total of 634 participants, investigated people with painful diabetic neuropathy (Beydoun 2006; Dogra 2005; Grosskopf 2006) and one trial involved 145 participants with neuropathic pain due to radiculopathy (Novartis 2004). We did not identify any trial eligible for inclusion that investigated other kinds of neuropathic pain.
We found that the four included trials were from a series of studies supported by the manufacturer, so they had similar design, inclusion and exclusion criteria, efficacy and safety assessments (Beydoun 2006; Dogra 2005; Grosskopf 2006; Novartis 2004).
The four studies were all placebo-controlled, double-blind, parallel-group trials and were conducted in multiple clinics and research centres. The reported study periods differed: a total of 72 weeks for the Dogra 2005 study, and 18 weeks for the Beydoun 2006, Grosskopf 2006 and Novartis 2004 studies. The Dogra 2005 trial consisted of a two-week pre-randomisation screening phase, an 18-week double-blind treatment phase and a 52-week open-label extension phase; while the latter three trials consisted of the screening phase (two weeks) and the double-blind treatment phase (16 weeks) (Beydoun 2006; Grosskopf 2006; Novartis 2004). The double-blind treatment phase was further divided into a titration period of four weeks and a maintenance period of 12 weeks in each study, and an additional follow-up period of two weeks for safety monitoring in the Dogra 2005 study.
Inclusion and exclusion criteria were explicitly stated. The three studies investigating treatment for painful diabetic neuropathy recruited men and non-pregnant, non-lactating women aged 18 years or older with an established clinical diagnosis of type 1 or 2 diabetes mellitus and with a history of neuropathic pain for six months to five years prior to study entry. Stable diabetic control and a certain pain severity and variability were also required. The other trial investigating participants with radiculopathy enrolled participants aged 18 years or older, of both genders, with a diagnosis of radiculopathy and evidence of motor deficit or reflex loss on examination, a history of neuropathic pain for six months or longer, and a VAS score of 50 or higher. Exclusion criteria were: patients with other types of pain, significant medical or psychiatric illnesses, a prior history of hyponatraemia, drug or alcohol abuse, amputations other than the toes, previous treatment with oxcarbazepine or a history of sensitivity to carbamazepine or its metabolites (see Characteristics of included studies). The baseline characteristics of participants were reported to be well matched between groups in each trial in terms of demography, duration of the primary disorders, and duration and severity of neuropathic pain. In total, the studies in this review included 421 males and 358 females, of whom 469 received oral oxcarbazepine and 310 received placebo. The mean duration of neuropathic pain (2.4 to 7.7 years) and average VAS score (100-point scale) of participants before treatment (70.7 to 76.9) in each group were also reported and no significant differences between any groups were found.
After the screening phase, eligible participants were randomised to placebo or to oxcarbazepine treatment with a given target dose. There was only one active group in the Dogra 2005 (oxcarbazepine 1800 mg/day), Grosskopf 2006 (oxcarbazepine 1200 mg/day) and Novartis 2004 (oxcarbazepine 1800 mg/day) studies, while in the Beydoun 2006 study one of the three doses of oxcarbazepine (300 mg, 600 mg, 900 mg) was given twice a day to participants in the different experimental groups. For the present analyses, 83 participants with painful diabetic neuropathy were analysed in the oxcarbazepine 600 mg/day group, 158 in the 1200 mg/day group and 157 in the 1800 mg/day group; and 71 participants with pain due to radiculopathy were analysed in the oxcarbazepine 1800 mg/day group. At the beginning of the treatment phase, participants took a small initial dose of study medication (300 mg/day) and then the oxcarbazepine was titrated over four weeks according to tolerability or the given maximum dose. During the subsequent maintenance period, treatment remained at the dose reached at the end of the titration period. Meanwhile participants in the control group received matching placebo tablets. The percentages of participants who achieved their target dose at the end of the Beydoun 2006 study were 82% (600 mg/day), 62% (1200 mg/day) and 56% (1800 mg/day); and 55% in the active group in the Dogra 2005 study. The mean oxcarbazepine dose during the maintenance period was 1455 ± 389 mg/day for the Dogra 2005 study and 1091 ± 222 mg/day for Grosskopf 2006. Acetaminophen was allowed in all trials as rescue medication, some other analgesics or drugs like benzodiazepines, non-steroidal anti-inflammatory drugs, opioids or stable doses of selective serotonin reuptake inhibitors were permitted during the study phase in one trial (Novartis 2004) but not allowed in the other three studies (Beydoun 2006; Dogra 2005; Grosskopf 2006).
Variables were evaluated after treatment for 16 consecutive weeks (Beydoun 2006; Grosskopf 2006) or 18 consecutive weeks (Dogra 2005) and withdrawals were recorded in detail. There were high percentages of dropouts, with the overall percentages of participants completing the studies being 67% (Beydoun 2006), 72% (Dogra 2005) and 67% (Grosskopf 2006). Common reasons for discontinuation included adverse effects, protocol violations and unsatisfactory responses to treatment. Most efficacy analyses were performed with an ITT analysis, which comprised all randomised participants regardless of completeness of treatment or follow-up, based on the last observation carried forward (LOCF) principle.
The outcome measures were also similar among the four trials. They all used a VAS for participants to report their pain severity. For the three trials investigating painful diabetic neuropathy (Beydoun 2006; Dogra 2005; Grosskopf 2006), the primary outcome was the change in average daily VAS scores between baseline and the end of double-blind treatment. The fourth trial compared the average VAS score between groups during the last week of double-blind treatment (Novartis 2004). Other efficacy variables used in the four trials included: the patient's global assessment of therapeutic effect (GATE), onset of therapeutic effect, durability of treatment effect, and sleep disturbances. Quality of life was also assessed by the SF-36 Health Survey and the Profile of Mood States (POMS) on up to five occasions throughout the double-blind phase. In addition, all included trials reported adverse events during the treatment phase and serious adverse events and those leading to discontinuation were clearly described (Beydoun 2006; Dogra 2005; Grosskopf 2006; Novartis 2004).
We excluded nine studies (Beydoun 2002; Beydoun 2007; Gong 2010; Hu 2010; Li 2011; Lindström 1987; Rémillard 1994; Venancio-Ramirez 2004; Zhou 2010) on inspection of the text because they: did not refer to any of the outcome measures specified in our protocol and the sponsor company could not find the full text or any useful data (Beydoun 2002); were open-label studies only evaluating the changes between the pre- and post-treatment conditions, without control groups (Beydoun 2007; Rémillard 1994); did not use a blind design (Gong 2010; Hu 2010; Li 2011; Zhou 2010); or involved a short follow-up duration, less than the minimum we specified in our protocol (Lindström 1987; Venancio-Ramirez 2004). See Characteristics of excluded studies.
Another study was only available as an abstract (Liebel 2001), which did not refer to any of the outcome measures specified in our protocol. We identified an additional abstract from the reference list of a review (Brainer-Lima 2003). The study randomised 20 participants with cancer-related neuropathic pain to oxcarbazepine or amitriptyline. Opiates were then gradually discontinued. It was unclear from the abstract whether each group received the same opioid regime. We have not received any reply from the authors of the abstracts, the two trials are therefore awaiting classification.
We also found a trial which is still ongoing (NCT01302275). It is a phase 4 clinical trial with a randomised, double-blind, placebo-controlled design aiming to determine if the effect of oxcarbazepine on chronic peripheral nerve pain depends on the supposed mechanism of the pain. It includes participants with definite or probable neuropathic pain due to polyneuropathy, postherpetic neuralgia or peripheral nerve injury; the estimated number of participants to be enrolled is 120. Inclusion and exclusion criteria are explicitly stated in the online report. Participants in the experimental group receive oxcarbazepine capsules orally and the dose is gradually increased over a period of 21 days from 300 mg daily to 2400 mg/day and maintained at that dose for three weeks, while placebo capsules are increased from one per day to eight per day. The primary outcome measure is total pain rated on a numeric scale (zero to 10 points). Secondary outcomes are response rate, neuropathic pain symptom inventory, patients' global impression of change, rating of evoked pain, sleep disturbance, quality of life and use of escape medication. The trial started in February 2011, and is estimated to be completed in December 2013. See Characteristics of ongoing studies.
Risk of bias in included studies
The four included trials all had a large sample of participants and standardised protocols. The methodological quality of the trials was assessed according to the Cochrane 'Risk of bias' tool (Higgins 2008b; Higgins 2011). All four trials were rated as at a high risk of bias mainly because of a large and imbalanced proportion of missing outcome data across groups (Figure 2).
|Figure 2. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
All included studies were multicentre, placebo-controlled, double-blind, parallel-group trials. Based on the published methods, we considered allocation sequence generation and concealment, blinding, and outcome reporting to be well performed to avoid bias for the three trials involving patients with diabetic neuropathy (Beydoun 2006; Dogra 2005; Grosskopf 2006). For the only trial involving patients with radiculopathy, we were unable to obtain detailed information about the methods of allocation (Characteristics of included studies).
All four trials clearly reported withdrawals after randomisation. The main reasons for discontinuation, in both the oxcarbazepine and placebo groups, included adverse events, lack of efficacy and protocol violations, but the numbers of participants with incomplete outcome data and the reasons for withdrawal were imbalanced across groups. To deal with missing outcome data, the trial authors performed analyses based on the ITT population using the LOCF approach, where all withdrawn participants had the average weekly VAS score from their last week of treatment carried forward for assessment. However, participants might have withdrawn before they found any obvious effect, so imputing outcomes in this way for missing participants (especially for early withdrawals) could lead an underestimation of the true effect. We attempted to contact the authors and company to obtain the missing data but failed. There were no detailed descriptions of the results for completers and dropouts in the reports, so we could not conduct an ITT analysis by regarding participants with missing outcome data as failures with no response. Since the dropout rates were high in all groups (19.1% to 56.3%), and they were significantly higher in the oxcarbazepine groups than the control groups (Beydoun 2006; Dogra 2005; Grosskopf 2006; Novartis 2004), a potential risk of bias against oxcarbazepine due to incomplete outcome data should be considered (Higgins 2008b). All outcomes specified in the protocols were reported in the published reports, so the three studies investigating participants with diabetic neuropathy were probably free of bias from selective reporting (Beydoun 2006; Dogra 2005; Grosskopf 2006). We also found a trial that was only published in abstract form (Novartis 2004) and which fulfilled the inclusion criteria of the present review.All expected outcomes in the protocol were reported online. To try to avoid publication bias we included the unpublished data from this trial.
Although sharing a similar design, the included trials reported some outcomes in different ways, which made it impossible for us to combine data from the three trials investigating the same condition, diabetic neuropathy (Beydoun 2006; Dogra 2005; Grosskopf 2006) in a meta-analysis. In fact, their results (for example reduction in pain intensity) were inconsistent. For the primary outcome, for example, we could only obtain relevant data from the single positive trial (Dogra 2005); the two negative trials did not provide data for inclusion in the primary outcome meta-analysis.
Effects of interventions
We planned to evaluate the effects of treatment after eight weeks, but we could not extract the relevant data because the included trials all reported outcomes 16 weeks after starting the treatment and the rates of change in the measured outcomes per week were not constant (Beydoun 2006; Dogra 2005; Grosskopf 2006; Novartis 2004). We could therefore only extract and analyse data as close as possible to 16 weeks after the beginning of treatment. In this review, we analysed the effects of three daily doses (600 mg, 1200 mg and 1800 mg) of oral oxcarbazepine versus placebo for painful diabetic neuropathy or neuropathic pain due to radiculopathy. There were no eligible trials of oxcarbazepine in other dosages, by other administration routes, compared to any other intervention or for other kinds of neuropathic pain.
Primary outcome measure
Reduction in patient-reported pain scores by 50% from the baseline
The included trials measured pain severity on a VAS, a commonly used pain scale that is 100 units in length with anchors of 'no pain' at the left end and 'worst pain imaginable' at the right end. The primary efficacy variable of this series of studies was the average daily VAS score during the last week (the 16th week) of double-blind treatment compared with baseline, and the average changes were compared between groups to evaluate the analgesic efficacy of oxcarbazepine. Authors did not always report the proportion of participants whose pain scores were reduced by 50% from the baseline, which was the primary focus of our review. We extracted relevant data from only one included study for each condition (Dogra 2005; Novartis 2004).
For participants with painful diabetic neuropathy, the results showed that the proportion of the oxcarbazepine (1800 mg/day) group whose pain scores were reduced by 50% from the baseline was statistically higher than that of the placebo group (24/69 (34.8%) versus 14/77 (18.2%)), with a RR of 1.91 (95% CI 1.08 to 3.39; P = 0.03; Analysis 1.1) and a NNTB of 6.0 (95% CI 3.3 to 41.0). There were no data available regarding our primary outcome measure (50% reduction in pain intensity) or the first secondary outcome measure (30% reduction in pain intensity) from the two negative trials, so we were unable to resolve the discrepancies between the included trials. For neuropathic pain due to radiculopathy, oxcarbazepine showed insignificant efficacy in reducing pain intensity by 50% from baseline comparing to placebo (7/71 (9.9%) versus 14/77 (18.9%); RR 0.52, 95% CI 0.22 to 1.22; P = 0.13; Analysis 2.1).
To evaluate the efficacy of oxcarbazepine, investigators calculated the mean changes in VAS scores from baseline to the end of double-blind treatment, but the conclusions were inconsistent. Dogra and colleagues found that participants treated with oxcarbazepine (1800 mg/day) experienced a significantly larger decrease from baseline in mean VAS score compared with placebo (-24.3 versus -14.7 units; P = 0.0108) (Dogra 2005) but in the other three trials (Beydoun 2006; Grosskopf 2006; Novartis 2004) the three doses of oxcarbazepine (600 mg/day, 1200 mg/day, 1800 mg/day) all failed to produce statistically significant pain relief.
Secondary outcome measures
Reduction in patient-reported pain scores by 30% from the baseline
We were able to extract the proportion of participants with a 30% reduction in VAS score from baseline to the 16th treatment week only from the Dogra 2005 study. A statistically higher percentage of participants in the oxcarbazepine group experienced this level of pain relief compared with placebo (31/69 (44.9%) versus 22/77 (28.6%); RR 1.57, 95% CI 1.01 to 2.44; P = 0.04; Analysis 1.2). The NNTB value was 6.1 (95% CI 3.1 to 113.6).
Patients' global impression of their change in pain
Investigators of this series of studies used GATE to assess patients' global impression of their change in diabetic neuropathy; GATE was measured on a seven-point Likert scale with scores ranging from -3 (very much improved) to 3 (very much deteriorated). In the present review, we compared the proportions of participants who reported all degrees of improvement (-3 to -1 point for GATE) and significant or obvious improvement (-3 to -2 point for GATE) after 16 weeks of treatment in the oxcarbazepine and placebo groups ( Analysis 1.3; Analysis 2.2; Figure 3). We could obtain relevant data from three included studies (Beydoun 2006; Dogra 2005; Novartis 2004). The other study (Grosskopf 2006) concluded only that no significant difference in this measure was found. From the meta-analysis for global impression of the change in pain of participants with painful diabetic neuropathy, 72.5% (50/69) of participants who received oxcarbazepine felt globally improved, significantly more than for those who received placebo (40.3% (31/77); RR 1.80, 95% CI 1.32 to 2.45). By combining relevant data from the Beydoun 2006 and Dogra 2005 studies, the meta-analysis ( Analysis 1.3) also showed a statistically significant difference between the experimental and control groups in producing significant or obvious improvement in participants with painful neuropathy (150/327 (45.9%) versus 50/166 (30.1%), respectively; RR 1.46, 95% CI 1.13 to 1.88). The NNTB with all dosages of oxcarbazepine was 6.4 (95% CI 4.1 to 14.4) for significant or obvious global improvement. For participants with neuropathic pain due to radiculopathy, there was no significant difference between the oxcarbazepine and placebo groups in either all degrees of improvement (35/71 (49.3%) versus 27/74 (36.5%); RR 1.35, 95% CI 0.92 to 1.98) or significant or obvious improvement (17/71 (23.9%) versus 11/74 (14.9%); RR 1.61, 95% CI 0.81 to 3.20) ( Analysis 2.2).
|Figure 3. Forest plot of comparison: 1 Oxcarbazepine versus placebo for painful diabetic neuropathy, outcome: 1.3 Patients' global impression of their change in pain.|
Overall quality of life measures
The SF-36 Health Survey, consisting of both physical and mental health summaries, was used to assess the overall quality of life in the included studies. It was performed on different occasions during the double-blind phase including at the eighth week, which was the time we planned according to our protocol (Zhou 2009) to measure this outcome. However, scores were not recorded for each follow-up period and some information could only be obtained at week 16 or at the termination visit. At the end of double-blind treatment in the Dogra 2005 trial, a significant difference was seen in the aggregate mental health score (P = 0.03) between oxcarbazepine- and placebo-treated participants (in favour of oxcarbazepine), but not in any SF-36 scales or the aggregate physical health score. In reports of the other two included trials (Beydoun 2006; Grosskopf 2006), the authors stated only that there were no significant differences between the oxcarbazepine groups and placebo groups on the quality of life questionnaire, but we were unable to obtain data for meta-analysis. Nor were we able to obtain relevant data from the trial published only as an abstract (Novartis 2004).
Adverse events often started during the titration period and resolved prior to study end. The frequently reported adverse events included dizziness, headache, nausea, somnolence, fatigue, vomiting, back pain, diarrhoea, tremor and blurred vision; most were mild to moderate in intensity and resulted in dosage adjustment. The number of participants with any adverse event could be extracted only from the Novartis 2004 study, but adverse events leading to withdrawal or serious outcomes were reported in detail in each trial.
In total, 25.6% (102/398) of participants with painful diabetic neuropathy in the oxcarbazepine group and 6.8% (16/236) in the placebo group discontinued treatment prematurely due to adverse effects, resulting in an RR of 3.83 (95% CI 2.29 to 6.40) with a significant difference between groups (P < 0.00001) ( Analysis 1.4; Figure 4). Serious adverse events occurred in a small number of participants. Some were considered related to treatment, such as erythema multiforme, severe fatigue, weakness, lightheadedness, headache leading to hospitalisations. Meta-analysis of these results revealed a statistically significant difference: the RR for serious adverse events comparing the experimental group (33/398, 8.3%) with the control group (6/236, 2.5%) was 3.65 (95% CI 1.45 to 9.20) with a P value of 0.006 ( Analysis 1.4; Figure 4). Thus we calculated the NNTH as 17.4 (95% CI 11.0 to 42.0) for serious adverse drug reactions. In the single trial investigating patients with neuropathic pain due to radiculopathy, the proportions of participants who experienced any kind of adverse events (68/71 (95.8%) versus 58/74 (78.4%); RR 1.22, 95% CI 1.07 to 1.39; P = 0.002) and who withdrew due to adverse events (30/71 (42.3%) versus 11/74 (14.9%); RR 2.84, 95% CI 1.55 to 5.23; P = 0.0008) were significantly higher in the oxcarbazepine group than in the placebo group. Additionally, six serious adverse events (8.5%) occurred in the oxcarbazepine group and two (2.7%, including one death) in the control group, showing a non-significant difference (RR 3.13, 95% CI 0.65 to 14.98; P = 0.15) ( Analysis 2.3).
|Figure 4. Forest plot of comparison: 1 Oxcarbazepine versus placebo for painful diabetic neuropathy, outcome: 1.4 Adverse effects.|
Lower and higher daily dose (600 mg/day, 1200 mg/day, 1800 mg/day)
Experimental participants with painful diabetic neuropathy could be divided into groups according to the target dosage of oxcarbazepine (600, 1200 or 1800 mg/day) used in each trial, and a dose-effect relationship was suggested in a dose-ranging study (Beydoun 2006). We conducted subgroup analyses of these three daily doses for outcomes where relevant data were available.
The percentage of participants who reported a significant or obvious improvement (-3 to -2 point for GATE) in pain was statistically greater in the 1800 mg/day oxcarbazepine group versus placebo (76/157 (48.4%) versus 50/166 (30.1%); RR 1.60, 95% CI 1.21 to 2.11). There was no statistically significant difference from placebo with oxcarbazepine at 600 mg/day (30/83 (36.1%) versus 33/89 (37.1%); RR 0.97, 95% CI 0.66 to 1.45) or 1200 mg/day (44/87 (50.6%) versus 33/89 (37.1%); RR 1.36, 95% CI 0.97 to 1.92) ( Analysis 3.1; Figure 5). The NNTB could not be calculated for the 600 mg/day oxcarbazepine group because the event rate was lower than that of the placebo group, while it was 7.4 (95% CI 3.6 to 97.1) in the 1200 mg and 5.5 (95% CI 3.5 to12.8) in the 1800 mg daily oxcarbazepine groups.
|Figure 5. Forest plot of comparison: 2 Subgroup analysis: different daily doses of oxcarbazepine versus placebo for painful diabetic neuropathy, outcome: 2.1 Patients with obvious or significant improvement after 16-week treatment.|
As for adverse effects, events leading to discontinuation occurred more frequently with the higher daily doses of oxcarbazepine (38/158 on 1200 mg; and 55/157 on 1800 mg) compared to the placebo group (10/159, RR 3.83, 95% CI 1.97 to 7.41; 12/166, RR 4.83, 95% CI 2.68 to 8.70, respectively). The difference between 600 mg/day oxcarbazepine and placebo was non-significant (RR 1.61, 95% CI 0.60 to 4.32) ( Analysis 3.2). Although the rates of adverse effects were increasing and dose-dependent, the confidence intervals were wide and overlapping. The comparisons of serious side effects between the experimental and control groups showed a similar trend toward a dose-dependent relationship: the highest daily dose (1800 mg/day) of oxcarbazepine treatment led to the highest percentage of serious adverse events (17/157, 10.8%) while the corresponding values were 14/158 (8.9%) for 1200 mg/day and 2/83 (2.4%) for 600 mg/day oxcarbazepine. Compared to the placebo group, the two higher doses of oxcarbazepine caused significantly more serious side effects ( Analysis 3.3; Figure 6). For serious adverse events, the NNTH was 77.5 (95% CI 19.0 to 37.5) for 600 mg/day, 14.3 (95% CI 8.4 to 48.5) for 1200 mg/day and 11.9 (95% CI 7.2 to 33.0) for 1800 mg/day oxcarbazepine.
|Figure 6. Forest plot of comparison: 2 Subgroup analysis: different daily doses of oxcarbazepine versus placebo for painful diabetic neuropathy, outcome: 2.3 Serious adverse effects.|
Despite attempts to develop more rational therapeutic approaches, treatment for some patients with neuropathic pain is still unsatisfactory. As a second-generation antiepileptic drug, oxcarbazepine was found to have some effect in treating several types of neuropathic pain in some open-label studies (Magenta 2005) and controlled trials, but the evidence was not convincing. We collected and extracted data from relevant RCTs for a more complete systematic analysis of the efficacy and safety of oxcarbazepine for neuropathic pain.
Summary of main results
We included four RCTs, all of which compared oral oxcarbazepine with placebo for the systematic treatment of neuropathic pain; three trials investigated people with painful diabetic neuropathy and the other involved participants with neuropathic pain due to radiculopathy. Since the trials were from the same series of studies, similar designs, methods, treatment and follow-up periods, and efficacy and safety variables allowed the combination of some but not all relevant data. The results of meta-analyses indicated a slight effect in relieving the pain (both with VAS reduction by 50% and 30% from baseline) due to diabetic neuropathy and good improvement of patients' global impression of their change (especially with significant or obvious improvement) ( Summary of findings for the main comparison). However, such results were mainly based on data from the single positive trial since the two negative trials did not provide enough data that could be included in the meta analyses. So the efficacy of oxcarbazepine for painful diabetic neuropathy is still uncertain. Results from the subgroup analyses, according to different target daily dose, suggested trends toward dose-effect relationships both for global recovery and adverse events. The highest daily dose (1800 mg/day) of oxcarbazepine had the greatest effect on pain relief but also led to the most adverse events, especially serious ones. For participants with radiculopathy, data from the only trial showed that oxcarbazepine could neither significantly relieve the pain nor improve patients' global impression of change compared to placebo.
The benefits of oxcarbazepine were largely owing to the positive results of one study (Dogra 2005), which investigated a higher dose (1800 mg/day) of oxcarbazepine. The other two included trials (Beydoun 2006; Grosskopf 2006) both failed to produce statistically significant differences between each dose of experimental drug and placebo, although there was a statistically significant trend favouring the two higher dose (1200 and 1800 mg/day) groups (Beydoun 2006). Neither of these studies was included in the meta-analysis, which is thus enormously biased towards the positive report. The investigators themselves put forward a number of factors to explain the differences in outcome among the trials. Besides the higher dose that was used in the positive trial, its lower placebo response might at least partially contribute to the statistically significant findings. In the placebo group of the Dogra 2005 study, the mean change of VAS scores from the baseline to the treatment end was -14.7 units, and about 22% of participants experienced an obvious or significant improvement; whereas the corresponding values were -19.1 units and 37.3% for the Beydoun 2006 study and -22.0 units for the Grosskopf 2006 study (percentage of improved participants was not reported). This phenomenon can be explained by the different experimental populations and the established conclusion that the placebo effect varies substantially among people with painful diabetic neuropathy (Backonja 2003; Beydoun 2006; Eisenberg 2007). Further, the effect of early discontinuations due to adverse events should also be considered. In the positive study (Dogra 2005) 7.8% (6/77) of participants withdrew from the placebo group because of adverse events, compared to 27.5% (19/69) from the oxcarbazepine group (1800 mg/day); a comparable percentage of participants withdrew in the placebo group of the Beydoun 2006 study (6.7% (6/89)) but the percentage of withdrawals in the 1800 mg oxcarbazepine group was much higher (40.9% (36/88)). Although analyses were based on ITT populations, the impact of different withdrawal rates on the treatment-effect estimation cannot be overlooked for reasons mentioned in the Risk of bias in included studies section.
Although the reports of trials included in this review state that most side effects related to the study drugs were mild to moderate in severity, both the proportions of any events and those leading to withdrawals were statistically higher in the oxcarbazepine group than the placebo group. The frequency of such adverse events was evidently dose related. So the speed and period of titration, and the observation and management of intolerance, should be rescheduled in further studies, and the most appropriate target dose might be further investigated for patients with neuropathic pain. Furthermore, it has been suggested that some human leukocyte antigen (HLA) alleles (for example HLA-B*1502 and HLA-A*3101) are associated with carbamazepine-induced skin-related adverse drug reactions that are sometimes serious and potentially life-threatening in certain populations. This association was not investigated in the included trials of the present review. Reports about the association of HLA alleles with oxcarbazepine are limited. Because of the close structural relationship with carbamazepine, if suggestive signs and symptoms appear then treatment should be withdrawn immediately (Drug Safety Update 2012).
Overall completeness, applicability and quality of the evidence
The four included trials were well designed and conducted in multiple centres. However, there are still unacceptable problems mainly due to imbalances in the completeness of outcome data. The considerable inconsistencies in placebo responses could also undermine the reliability of the combined results. Although the placebo response in neuropathic pain varies considerably among trials, and is difficult to regulate and control, RCTs are still the most powerful evidence on which to base conclusions (GRADE). The mean pain severity at randomisation differs from trial to trial but has not been further analysed in this review because of limited data. These differences could also impact the results to some extent. All of the included trials reported relevant outcomes for pain relief, patients' quality of life and safety, but data were not always available for meta-analysis for each outcome. Most of our results came from one or two eligible trials, or from descriptive analyses. So the evidence from this review is incomplete and may not be widely applicable.
The efficacy of oxcarbazepine has also been investigated for types of neuropathic pain other than diabetic neuropathy and radiculopathy (for example trigeminal neuralgia and postherpetic neuralgia). Several studies (Beydoun 2002a; Beydoun 2002b; Carrazana 2003; Nasreddine 2007), which were RCTs, compared oxcarbazepine with carbamazepine in participants with trigeminal neuralgia; all the results consistently showed that both drugs had comparable effectiveness in relieving pain and improving the quality of life. These studies have only been published as a meta-analysis in abstract form (Beydoun 2002a; Beydoun 2002b) or in reviews (Carrazana 2003; Nasreddine 2007), where detailed descriptions and data are unavailable. Thus they could not be included in the current review. A postherpetic neuralgia study showed a significant decrease in VAS score and improvements in participants' quality of life (Criscuolo 2005) but had an open-label design. As a result, we cannot draw any firm conclusions about oxcarbazepine for the treatment of non-diabetic neuropathic pain.
Agreements and disagreements with other studies or reviews
In agreement with the conclusions of our systematic review, to date the European Federation of Neurological Societies (EFNS) guidelines (Attal 2006; Attal 2010) have not recommended oxcarbazepine for people with diabetic neuropathic pain because of insufficient and conflicting evidence. The guidelines confirmed oxcarbazepine as one of the first-line drugs for classical trigeminal neuralgia, whereas with no eligible trials we are unable to draw any conclusion in this review about the use of oxcarbazepine for this type of neuropathic pain.
Implications for practice
On the basis of moderate quality evidence from one trial in diabetic peripheral neuropathy, oxcarbazepine is effective in reducing pain for this condition. However, this conclusion does not take into account negative results from other trials of oxcarbazepine in diabetic peripheral neuropathy that could not be included in our analysis. We did not find any evidence from randomised controlled trials to determine the efficacy or safety of oxcarbazepine for other kinds of neuropathic pain other than one low quality trial in people with radiculopathy. Most adverse effects related to oxcarbazepine are rated as mild to moderate in severity, but adverse events leading to discontinuation of drug administration or serious adverse events are not uncommon.
Implications for research
More well designed, large, multicentre, randomised placebo- or active-controlled trials investigating oxcarbazepine are needed. The response in neuropathic pain of other causes should be investigated. A dose-ranging design should be considered to determine whether a dose-response relationship is present, and the most appropriate target dose might be studied further.
We are grateful for the assistance of the Cochrane Neuromuscular Disease Group.
The editorial base of the Cochrane Neuromuscular Disease Group is supported by the MRC Centre for Neuromuscular Diseases.
The Trials Search Co-ordinator of the Cochrane Neuromuscular Disease Group, Angela Gunn, provided literature searches.
Data and analyses
- Top of page
- Summary of findings [Explanations]
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Differences between protocol and review
- Index terms
Appendix 1. MEDLINE (OvidSP) search strategy
Database: Ovid MEDLINE(R) <1946 to October Week 3 2012>
1 randomized controlled trial.pt. (339605)
2 controlled clinical trial.pt. (85425)
3 randomized.ab. (242346)
4 placebo.ab. (135629)
5 drug therapy.fs. (1578525)
6 randomly.ab. (173863)
7 trial.ab. (251056)
8 groups.ab. (1136962)
9 or/1-8 (2939707)
10 exp animals/ not humans.sh. (3797751)
11 9 not 10 (2497281)
12 (oxcarbazepine or trileptal$ or trexapin).mp. (1224)
13 Radiculopathy/ (3272)
14 Diabetic Neuropathies/ (11395)
15 Neuralgia, Postherpetic/ (513)
16 Spinal Cord Injuries/ (26088)
17 Trigeminal Neuralgia/ (5363)
18 exp Complex Regional Pain Syndromes/ (4142)
19 (neuropath$ pain or cervical radiculopathy or lumber radiculopathy or diabetic neuropath$).mp. (21012)
20 (peripheral neuropath$ or postherpetic neuralgia or HIV neuropath$ or spinal cord injur$ or trigeminal neuralgia or complex regional pain syndrome).mp. (50650)
21 or/13-20 (74261)
22 11 and 12 and 21 (71)
23 22 and 20111101:20121030.(ed). (1)
Appendix 2. EMBASE (OvidSP) search strategy
Database: Embase <1980 to 2012 Week 43>
1 crossover-procedure.sh. (35377)
2 double-blind procedure.sh. (111601)
3 single-blind procedure.sh. (16569)
4 randomized controlled trial.sh. (331618)
5 (random$ or crossover$ or cross over$ or placebo$ or (doubl$ adj blind$) or allocat$).tw,ot. (908624)
6 trial.ti. (136873)
7 or/1-6 (1037646)
8 (animal/ or nonhuman/ or animal experiment/) and human/ (1217670)
9 animal/ or nonanimal/ or animal experiment/ (3327400)
10 9 not 8 (2754501)
11 7 not 10 (951313)
12 limit 11 to embase (739107)
13 (oxcarbazepine or trileptal$ or trexapin).mp. (6675)
14 radiculopathy/ (6452)
15 radicular pain/ (1836)
16 diabetic neuropathy/ (16378)
17 postherpetic neuralgia/ (3462)
18 spinal cord injury/ (33492)
19 trigeminus neuralgia/ (7885)
20 exp complex regional pain syndrome/ (6871)
21 (neuropath$ pain or cervical radiculopathy or lumbar radiculopathy or diabetic neuropath$ or peripheral neuropath$).mp. (68480)
22 (postherpetic neuralgia or HIV neuropath$ or spinal cord injur$ or trigeminal neuralgia or complex regional pain syndrome).mp. (52944)
23 or/14-22 (129539)
24 12 and 13 and 23 (95)
25 24 and 20111101:20121030.(dd). (11)
Appendix 3. CENTRAL search strategy
#1 oxcarbazepine OR trileptal* OR trexapin
#2 "neuropathic pain" OR "neuropathy pain"
#3 "cervical radiculopathy"
#4 MeSH descriptor Radiculopathy, this term only
#5 "lumber radiculopathy"
#6 "diabetic neuropathy" OR "diabetic neuropathies"
#7 MeSH descriptor Diabetic Neuropathies explode all trees
#8 "peripheral neuropathy" OR "peripheral neuropathies"
#9 "postherpetic neuralgia"
#10 MeSH descriptor Neuralgia, Postherpetic, this term only
#11 "HIV neuropathy" OR "HIV neuropathies"
#12 MeSH descriptor Spinal Cord Injuries explode all trees
#13 "spinal cord injury" OR "spinal cord injuries"
#14 "trigeminal neuralgia" OR "trigeminus neuralgia"
#15 MeSH descriptor Trigeminal Neuralgia explode all trees
#16 "complex regional pain syndrome"
#17 MeSH descriptor Complex Regional Pain Syndromes explode all trees
#18 (#2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17)
#19 (#1 AND #18)
Appendix 4. Chinese Biomedical Retrieval System Database search strategy
(NB. all of the search terms were translated to Chinese terms when we conducted the searches)
1. neuropathic pain
3. cervical radiculopathy
4. lumber radiculopathy
5. diabetic neuropathy
6. peripheral neuropathy
7. postherpetic neuralgia or PHN
8. HIV neuropathy
9. Spinal Cord Injuries
11.complex regional pain syndrome or CRPS
15.qulai or renao or wanyi (trade names of oxcarbazepine in Chinese)
23.12 and 16 and 22
Last assessed as up-to-date: 30 October 2012.
Contributions of authors
Mi Yang and Ning Chen searched for studies and extracted data and Cairong Zhu carried out analyses. All authors were involved with writing of the protocol and the full review.
Declarations of interest
Differences between protocol and review
We searched the Chinese Biomedical Retrieval System in addition to other databases. We did not search the Trials Register of the Cochrane Pain, Palliative and Supportive Care Group. We added a 'Summary of findings' table to the review, which was not mentioned in the protocol.
For the 'Risk of bias' assessments we used 'High', 'Low' or 'Unclear' risk of bias, according to Higgins 2011.
In our protocol we stated that for continuous data we would calculate the weighted mean difference (WMD) or standardised difference (SMD). Only dichotomous data were available from the included trials. There were no cross-over trials for analysis.
We planned sensitivity analyses on the basis of trial quality and subgroup analyses on the basis of type of neuropathic pain. We did not perform them because all trials were of moderate quality; and subgroup analysis on the basis of type of neuropathic pain was not possible.
Changes in authorship have taken place since publication of the protocol: Guo J, Li Q, Yang X, Yang J withdrew and Wu F became an author.
Medical Subject Headings (MeSH)
Analgesics, Non-Narcotic [*therapeutic use]; Carbamazepine [*analogs & derivatives; therapeutic use]; Diabetic Neuropathies [*drug therapy]; Neuralgia [*drug therapy]; Radiculopathy [*drug therapy]; Randomized Controlled Trials as Topic
MeSH check words
* Indicates the major publication for the study