Neuropathic pain after spinal cord injury (SCI) is a. Tramadol in Neuropathic Pain After Spinal Cord Injury

Transcription

1 ORIGINAL ARTICLE Tramadol in Neuropathic Pain After Spinal Cord Injury A Randomized, Double-blind, Placebo-controlled Trial Cecilia Norrbrink, RPT, PhD*w and Thomas Lundeberg, MD, PhD* Objectives: To assess the efficacy and safety of tramadol for relieving neuropathic pain after spinal cord injury (SCI). Methods: Thirty-six patients with SCI and neuropathic pain were enrolled in a randomized, double-blind, placebo-controlled trial, and 35 patients were included in the intention-to-treat analysis based on all patients taking at least 1 dose of study medication. Of the intention-to-treat population, 23 were randomized to treatment with tramadol and 12 to placebo. Patients were given study medication for 4 weeks, starting with 3 times 50 mg tramadol daily or 3 times 1 tablet of placebo. Results: At evaluation, 4 weeks after the treatment initiation, ratings of pain intensity were lower in those randomized to tramadol compared with those on placebo. Pain severity scores on the Multidimensional Pain Inventory had also decreased in patients on tramadol. No between-group differences in pain affect were observed. Adverse events were substantial and caused 43% of those on tramadol and 17% on placebo to withdraw from the study. Discussion: Tramadol might be tried for neuropathic pain after SCI after the use of gabapentin/pregabalin, and tricyclic antidepressants have been found to be insufficient. Titration should be slow and individual, to minimize the risk of adverse events. Key Words: spinal cord injury, neuropathic pain, pain, tramadol (Clin J Pain 2009;25: ) Neuropathic pain after spinal cord injury (SCI) is a difficult condition to treat. Guidelines are often based on studies on peripheral neuropathic pain conditions because few studies have been carried out in patients with SCI and pain. Siddall and Middleton 1 have presented guidelines for treating neuropathic pain after SCI and proposed an algorithm for the management of SCI pain. In their proposal, gabapentin was the first-line agent in the subacute or chronic setting followed by a tricyclic antidepressant (TCA) (amitryptiline or nortryptiline) or a weak opioid, such as tramadol, as second-line treatment. Both drugs could be used alone or together with the firstline agent. Both gabapentin 2,3 and pregabalin 4 have been assessed in patients with SCI and neuropathic pain and found to be Received for publication April 5, 2007; revised May 20, 2008; accepted July 11, From the *Spinalis SCI Unit, Karolinska University Hospital; and wdepartment of Clinical Sciences, Section of Rehabilitation Medicine, Karolinska Institutet, Danderyd Hospital, Stockholm, Sweden. This study was made possible by grants from the Norrbacka-Eugenia Foundation. Reprints: Cecilia Norrbrink, RPT, PhD, Department of Clinical Sciences, Karolinska Institutet, Danderyd Hospital, Bld 39:3, SE Stockholm, Sweden ( cecilia.norrbrink@ki.se). Copyright r 2009 by Lippincott Williams & Wilkins beneficial, whereas amitryptiline 5 was reported to be ineffective in patients with SCI. Opioids in the treatment of neuropathic pain have been an issue of debate for many years, but several recent studies have pointed to the efficacy of these drugs in patients with neuropathic pain of a peripheral origin. 6 In patients with neuropathic pain after SCI, intravenous infusions of alfentanil were found to reduce continuous and evoked pain, 7 and morphine to be effective on some components of central pain. 8 So far, tramadol has not been assessed in patients with SCI and neuropathic pain. Tramadol is a drug with 2 suggested modes of action; in part, it is a low-affinity m-opioid agonist and, in part, a weak monoamine (norepinephrine and serotonin) reuptake inhibitor. Tramadol has been approved for treating moderate to severe acute and chronic pain in Sweden since A Cochrane review on the effectiveness of tramadol for neuropathic pain was published in The authors found 4 placebo-controlled studies that evaluated the effect of tramadol in neuropathic pain and concluded that tramadol is an effective treatment for neuropathic pain. Effectiveness was shown in patients with diabetic neuropathy, in patients with painful polyneuropathy, and in patients with postherpetic neuralgia. Besides a decrease in ratings of pain intensity, 1 study also reported a positive effect on touch-evoked pain. 11 Tramadol has also been shown to be effective in decreasing suggested nociceptive-related behavior 12 (ie, thermal hyperalgesia) and in increasing nociceptive pressure thresholds in rats with a peripheral nerve lesion. 13 The aim of this study was to assess the efficacy and safety of tramadol for relieving neuropathic pain after SCI in a randomized, double-blind, placebo-controlled trial. METHODS AND PATIENTS Patients Patients were recruited from the spinal out-patient clinic in Stockholm, Sweden, and from 3 regional hospitals in areas near Stockholm (Sandviken, O rebro, and Falun). Patients registered at the spinal unit in Stockholm with a traumatic or nontraumatic injury and pain diagnosed in a previous annual health control as being of at or below level neuropathic pain in origin were selected from the database. From this selection, patients between 18 and 70 years of age with no known cognitive dysfunction and currently not using tramadol were sent information about the study. Information was also disseminated in advertisements at the spinal unit. At the 3 regional hospitals, information was sent to all patients who fulfilled the inclusion criteria. Inclusion criteria were a signed informed consent, an SCI for more than 12 months, pain classified as neuropathic Clin J Pain Volume 25, Number 3, March April

2 Norrbrink and Lundeberg Clin J Pain Volume 25, Number 3, March April 2009 pain at or below the level of lesion 14 and of a duration of more than 6 months, fluency in Swedish, and no known cognitive impairment. Pain intensity ratings of either the general or the worst pain intensity in the last week were required to be equivalent to or more than 3 on Borg s Category Ratio (CR-10) scale, a combined numerical and verbal rating scale. 15 Patients who were pregnant or lactating were excluded. The patients had to be naive to treatment with tramadol and have had no signs of intolerance to treatment with opioids in the past. If patients were currently using opioids or antidepressants, inclusion was decided on an individual basis. Patients were allocated consecutively to the study during the study period (23 mo); 36 patients fulfilled the inclusion criteria and consented to participate in the study. Study Design The study was a double-blind, randomized, placebocontrolled trial of tramadol for the treatment of at or below level neuropathic pain after SCI. Patients included in the study were informed about the study in writing and signed a written consent form at the time of inclusion/assessment. Patients were thereafter randomized in a 2:1 ratio (tramadol/placebo) and subsequently given study medication for a treatment period of 4 weeks. The study medications (both active and placebo drugs) were imported in containers from Lannacher Heilmittel GesmbH in Lannach, Austria. The active and the placebo drugs were identical in appearance. The drugs were thereafter blistered and randomized by Apoteket AB in Stockholm, Sweden. Apoteket AB is a state-owned limited company that has the exclusive right to sell pharmaceuticals to the public. Apoteket AB provided sealed, coded envelopes that were opened after each participant had completed his or her treatment or when participants reported adverse events that caused them to drop out or be withdrawn from the study (hereafter the term withdrawal will be used). During the study, patients were allowed to continue stable pain medication and asked not to make any changes in their current dosages. Patients were given the drugs in a double-blinded fashion together with information about drug titration, a pain diary, and whom to contact with questions about dosages or adverse events. The diary contained a copy of the CR-10 scale for rating daily pain intensity and had space for noting side-effects in a fixed schedule. The most commonly reported adverse events listed for tramadol 16 were given in the patient diary. Four weeks (28 to 30 d) after the baseline visit, patients were scheduled for an evaluation of efficacy and safety at the spinal unit according to the protocol. The initial dose was 50-mg tramadol (1 tablet) or 1 placebo tablet 3 times daily according to recommendations in The Swedish Drug Compendium 16 published by the Swedish Association of the Pharmaceutical Industry. Patients were thereafter instructed to increase the dose every 5 days by 50 mg (1 placebo tablet) until a maximum dose of 400 mg daily (8 placebo tablets) was reached. If optimal pain relief was obtained or if adverse events became intolerable before maximum dose was reached, patients were instructed to stop increasing their dose. Patients experiencing intolerable adverse events were withdrawn from the study and if they were receiving the active drug, they were asked to phase out the drug by 50 mg/d. Outcome Measures At baseline, a pain specialist physician collected data on the patient s medical history and assessed the patient s pain. A bedside assessment of the different sensory modalities was made (light touch, pressure, pin prick, vibration, heat, and cold). Responses were rated as intact, decreased, or absent. The assessment was made in the most painful area at or below the level of lesion. Further, an electrical tooth-brush (Braun type 4728, 50 to 60 Hz) was used to assess dynamic mechanical allodynia. 11 If patients were aware of a hypersensitive area, the assessment was made in this region; if not, the assessment was made in the most painful area at or below the level of lesion. Stimulus was applied until patients reported pain or for a maximum of 60 seconds. If pain was generated by the stimulus, patients rated the stimulus-evoked pain intensity on the CR-10 scale and were considered to have dynamic mechanical allodynia. The pain protocol included drawing a pain chart and rating pain intensity and pain unpleasantness on the CR-10 scale. The patient recorded mildest, general, and worst pain intensities; present pain intensity; and pain unpleasantness (the affective component of pain). Pain intensity was also estimated using the PainMatcher, 17 where patients matched the magnitude of their present pain against an electrical stimulus produced by the apparatus. Present pain intensity was also recorded daily in a pain diary, as were adverse events. Part I of the Swedish language version of the Multidimensional Pain Inventory (MPI-S) 18 was used to assess psychosocial consequences of pain. Section 1 comprises 22 items rated on a 7-point numerical scale in 5 subdomains: pain severity (2 items), interference pain-related interference in everyday life (11 items), perceived life control (4 items), affective distress (3 items), and social support perceived support from a spouse or significant others (2 items). The Hospital Anxiety and Depression (HAD) 19 scale was used to assess mood. This instrument consists of 7 questions on anxiety and 7 on depression. Each item has outcomes ranging from 0 to 3. Life satisfaction was assessed using the LiSat-9, 20,21 which is a patient self-rating, life satisfaction instrument consisting of the global item life as a whole and 8 domain-specific items. The 9 items are rated on an ordinal scale ranging from 1 to 6 where 1 represents very dissatisfying and 6 very satisfying. In this study, we analyzed the global item life as a whole. Quality of sleep was rated on a sleep questionnaire, 22 which consisted of 10 items on sleep quality (difficulties initiating sleep, difficulties waking up, frequent awakenings with difficulty returning to sleep, nightmares, unrefreshing sleep, waking too early, disturbed night-time sleep, excessive day-time sleepiness, irritated or tired eyes, and mental tiredness) that are rated on an ordered categorical scale: never, rarely, sometimes, most often, and always. The categories are numbered from 1 to 5. The outcome of the 10 categories is summarized and divided by 10 to obtain a global value. At the evaluation, patients also rated treatment effect concerning pain using the Patient Global Impression of Change (PGIC). 23,24 The study was approved by the Ethics committee of Karolinska Institutet in Stockholm, Sweden [KI registration numbers (D-nr) and 2005/ ] and by the Medical Products Agency (151:2003/ 17345). 178 r 2009 Lippincott Williams & Wilkins

3 Clin J Pain Volume 25, Number 3, March April 2009 Tramadol in Neuropathic Pain After Spinal Cord Injury STATISTICS Patient and baseline characteristics are presented using descriptive statistics, that is, number of observations, mean, standard deviation, median, and range (minimum to maximum). Outcome variables are further presented using median and interquartile range (IQR) or range before and after treatment by group. Data analyses were performed on the intention-to-treat population defined as all patients who were randomized and had taken at least 1 dose of study medication. Data from those patients who dropped out or were withdrawn prematurely were replaced using the lastobservation-carried-forward approach. Differences between treatments concerning proportion of patients who experienced adverse events were tested using the w 2 test. Determination of the between-treatment difference measured with the PGIC was tested with the Mann-Whitney U test. The rank-invariant method introduced by Svensson 25,26 was used to estimate systematic changes in outcome variables (pain intensity, pain affect, mood, sleep quality, and psychosocial consequences) 4 weeks after treatment compared with baseline. Systematic group changes are explained by relative change in position (RP) that is, the proportion of individuals with a higher level minus the proportion of patients with a lower level in the outcome variable. Values of RP range from 1 (all patients decreased) to +1 (all patients increased) and a value close to 0 indicates a negligible systematic group change in the outcome measure. When RP a0, the values after treatment are systematically higher (+) or lower ( ) for the group than the initial levels. RP was estimated together with the corresponding 95% confidence interval. Standard error was calculated using the Jack-knife method. Differences between tramadol and placebo in RP were also estimated together with the 95% confidence intervals. Negative differences between treatments (tramadol-placebo) indicate that a larger proportion of individuals in the tramadol group decreased (or increased less) from baseline to 4 weeks after treatment compared with the individuals in the placebo group in the outcome variable. The difference in RP between treatments was also tested using the t test for independent samples. Tests were 2-sided, and a significance level of 0.05 was chosen. All descriptive statistics were produced in STATISTICA 7.0 (StatSoft Inc, Tulsa, OK) and RP values were calculated in SYSRAN 1.0 (JK Biostatistics, Stockholm, Sweden) for Matlab 6.0 (The MathWorks Inc, Natick, MA). RESULTS Thirty-six patients were enrolled in the study. One of these patients never began drug treatment and was therefore not included in the intention-to-treat population or in the evaluation of safety. Of the 35 patients who did begin medication, 23 received the active drug and 12 the placebo. Table 1 lists patient characteristics. Patients randomized to the active drug and placebo groups differed regarding level of injury with more patients with paraplegia being randomized to treatment with placebo. Patients in the placebo group also reported higher levels of pain and some associated variables at baseline (Table 2). Twenty of the 35 (57%) patients were on concomitant analgesic medication (including antiepileptic drugs or antidepressants); 14 (61%) in the tramadol group and 6 (50%) in the placebo group (Table 1). Treatment All patients were treated a mean of 21.0 days (SD 10.1), range 1 to 30 days; patients in the tramadol group a mean 18.8 days (SD 10.8), range 1 to 30 days; and patients in the placebo group a mean 25.1 days (SD 10.1), range 7 to TABLE 1. Patient Characteristics Total Group Tramadol Placebo n Sex 28#; 7~ 19#; 4~ 9#; 3~ Age in y (SD) 51.3 (10.8) 51.4 (11.6) 51.2 (9.7) Years postinjury (SD) 14.6 (11.3) 14.0 (12.4) 15.7 (9.4) Traumatic injury (78%) 9 (75%) Tetraplegia/paraplegia 16/19 12/11 (52%/48%) 4/8 (33%/67%) Employment status Working 13 9 (39%) 4 (33%) Studying 1 1 (4%) 0 (0%) Sick-pension, etc (57%) 7 (58%) Retired 1 0 (0%) 1 (8%) Educational level University 11 6 (26%) 5 (42%) High school (48%) 4 (33%) Other 8 5 (22%) 3 (25%) Marital status Single/divorced (52%) 6 (50%) Married/unmarried cohabitant/living apart together (LAT) (48%) 6 (50%) Concomitant overall medication 28 (80%) 19 (83%) 9 (75%) Concomitant pain medication 20 (57%) 14 (61%) 6 (50%) Opioids 7 (20%) 5 (22%) 2 (17%) AED 9 (26%) 5 (22%) 4 (33%) AD 6 (17%) 4 (17%) 2 (17%) NSAIDs/COX-2 4 (11%) 2 (9%) 2 (17%) AD indicates antidepressant drugs; AED, antiepileptic drugs; COX-2, cyclooxygenase-2-inhibitors; NSAIDs, nonsteroidal antiinflammatory drugs. r 2009 Lippincott Williams & Wilkins 179

4 Norrbrink and Lundeberg Clin J Pain Volume 25, Number 3, March April 2009 TABLE 2. Primary and Secondary Outcome Measures at Baseline and After 4 Weeks of Treatment With Tramadol or Placebo Tramadol Placebo Baseline Week 4 Baseline Week 4 MD (IQR) MD (IQR) MD (IQR) MD (IQR) Present pain intensity* 3 (2.5;5) 3 (2;4) 5 (4.5;5.5) 5.5 (3.5;7) General pain intensity* 4 (3;5) 3 (2.5;5) 7 (4.5;7) 6.5 (5;7.25) Worst pain intensity* 7 (5;8) 5 (4;7) 9 (7;10) 8 (7;10) Pain unpleasantness 5 (2.5;7) 4 (2;5.5) 7 (5.5;9) 6.5 (4;9.5) Matched pain (PainMatcher) 14 (8;21) 10 (6;15) 15 (12;18) 15 (12;17) Anxiety* 7 (2;9) 6 (1:8) 9 (5.75;13) 9 (5.5;12) Depression 4 (2;8.5) 3 (2;6) 4.5 (3;13.5) 5 (2;4.5) Sleep quality* 2.7 (2.1;3.2) 2.7 (2.3;3.2) 2.8 (2.7;3.4) 2.9 (2.4;3.4) Global life satisfaction* 4 (4;5) 4 (4;5) 3 (2;4) 4 (2;4.5) MPI pain severity* 3 (2;4) 2.5 (2;3.5) 4 (3.5;4.38) 4.13 (3.5;4.75) MPI pain interference 2.36 (1.7;4.18) 2.45 (1.55;3.55) 3.64 (2.13;4.67) 3.64 (1.65;5.34) MPI perceived life control 4 (2.5;4.75) 4.25 (3.25;5.0) 3 (2.38;3.88) 3.63 (2.38;4.5) MPI affective distress 2.33 (1.33;3.33) 2.0 (0.67;3.33) 2.83 (1.83;3.33) 2.5 (1.83;3.33) MPI social support 3.5 (2.5;5.0) 3.5 (2.5;5.0) 3.25 (1.5;4.25) 3.5 (1.25;4.75) The significant between-group differences found in the analysis and reported in Figure 1 are marked with an *. Higher ratings on the MPI scales perceived life control and social support and lower ratings on the other MPI scales indicate improvement. IQR indicates interquartile range; MD, median; MPI, Multidimensional Pain Inventory. 30 days. Median maximum dose in the tramadol group was 250 mg (range: 100 to 400 mg) and in the placebo group 8 tablets (corresponding to 400 mg) (range: 3 to 8 tablets). Withdrawals Thirteen patients (37%) were withdrawn from the trial; 2 patients (17%) in the placebo and 11 (48%) in the active drug group. One withdrawal was for drug-unrelated factors (rib fracture), but the remaining patients stopped medication because of adverse events. Adverse Events Twenty-one (91%) patients experienced 1 or more adverse events on active medication and 7 (58%) on placebo (P = 0.02). More moderate to severe adverse events were reported for the active drug than for the placebo (Table 3). The most commonly reported adverse events of tramadol were tiredness (74%), dry mouth (52%), and dizziness (52%). Constipation was the most commonly reported side-effect of the placebo (33%). Primary Outcome Measures Pain Intensity Between-group differences (tramadol vs. placebo) in ratings of present pain, general pain, and worst pain intensity were significant (P<0.05) (Fig. 1). The proportion of patients who reported decreased pain intensity on tramadol was larger than for placebo. A significant decrease was also seen on the MPI subscale pain severity (P<0.05). Assessments of general and of worst pain intensity decreased compared with placebo, and changes in the tramadol group were decreased compared with baseline (P<0.05). Levels of present pain intensity in the tramadol group were decreased compared with the placebo group (P<0.05). Daily pain intensity was rated and noted in a diary. The median value for patients on the active drug who completed the study was 4.5 (IQR 2.5;5.5) at baseline and 3.5 (1.75;5.0) on the last treatment day; for patients on the active drug who withdrew/dropped out (missing data on 1 patient), median pain intensity was 3 at both baseline (2.5;3) and on their last day (2;3). Patients in the placebo group had a median pain intensity of 6 on day 1 (5;7) and 5.75 (4.75;7) on their last treatment day. Patient Global Impression of Change Those who completed the study rated the painrelieving effect of the treatment on the PGIC (Fig. 2). Seven patients on active drug (30%) rated an improvement (minimally improved much improved), but only 4 (17%) TABLE 3. Frequency and Percentage of Various Adverse Events by Treatment All Levels n (%) Mild to Moderate n (%) Moderate to Severe n (%) Tramadol (%) Placebo (%) Tramadol (%) Placebo (%) Tramadol (%) Placebo (%) Tiredness 17 (74) 2 (17) 6 (26) 0 (0) 11 (48) 2 (17) Dry mouth 12 (52) 3 (25) 5 (22) 1 (8) 8 (35) 2 (17) Dizziness 12 (52) 3 (25) 4 (17) 1 (8) 8 (35) 2 (17) Sweating 9 (39) 3 (25) 5 (22) 2 (17) 4 (17) 1 (8) Constipation 8 (35) 4 (33) 2 (9) 1 (8) 6 (26) 3 (25) Nausea 9 (39) 3 (25) 3 (13) 1 (8) 6 (26) 2 (17) Voiding dysfunction 1 (4) 0 (0) 1 (4) 0 (0) 0 (0) 0 (0) In total, 21 of 23 (91%) patients on the active drug and 7 of 12 (58%) on placebo experienced adverse events. The table lists the adverse events reported by all patients. 180 r 2009 Lippincott Williams & Wilkins

5 Clin J Pain Volume 25, Number 3, March April 2009 Tramadol in Neuropathic Pain After Spinal Cord Injury Present pain intensity General pain intensity Worst pain intensity Pain unpleasantness Matched pain Anxiety Depression Global life satisfaction Sleep quality MPI - pain severity MPI - pain interference MPI - perceived life control MPI - Social support MPI - Affective distress FIGURE 1. Differences in outcome variables between tramadol and placebo in relative change in position (RP), that is, systematic change on a group level, and corresponding 95% confidence intervals. Values on the left-hand side (negative) indicate that a larger proportion of tramadol-treated patients had lowered their values (or increased their values less) from baseline to week 4 compared with patients treated with placebo rated their pain to be much improved. One patient in the placebo group reported minimal improvement (8%). No patients in either group reported their pain to be very much improved. The between-group difference was significant (P = 0.04). Secondary Outcome Measures Significant (P<0.05) between-group differences were seen in ratings of anxiety, global life satisfaction, and sleep quality (Fig. 1). Median values in the tramadol group did not change regarding ratings of global life satisfaction or n Active drug Placebo Very much worse Much worse Minimally worse No change Minimally improved Much improved Very much improved FIGURE 2. Ratings of treatment effect on the Patient Global Impression of Change (PGIC). Twelve of 23 patients (52%) in the active drug group and 10 of 12 (83%) in the placebo group completed the trial. r 2009 Lippincott Williams & Wilkins 181

6 Norrbrink and Lundeberg Clin J Pain Volume 25, Number 3, March April 2009 sleep quality; instead, changes in median values were seen in the placebo group where ratings of global life satisfaction improved and sleep quality deteriorated. No between-group changes were seen in pain unpleasantness, in depression, or on the MPI scales pain interference, perceived life control, affective distress, or social support. Stimulus-evoked Pain Ten of 23 patients in the tramadol group (43%) were diagnosed with stimulus-evoked pain (dynamic mechanical allodynia) as assessed with a Braun tooth-brush compared with 4 of 12 (33%) in the placebo group. Of those in the tramadol group who completed the trial (n = 5), an increased latency to pain during stimulation of the skin and decreased pain intensity ratings of 1.8 units or more 27 were seen in 1 patient. Another patient s pain intensity ratings decreased 1.5 units, but latency to pain was unchanged. In the placebo group (n = 3), 1 patient s ratings of pain intensity also decreased 1.5 units, but latency to pain was unchanged. DISCUSSION In our study, patients with SCI and neuropathic pain who were randomized to treatment with tramadol significantly improved regarding pain intensity ratings (including pain severity on the MPI-S) and anxiety ratings compared with those randomized to placebo. Improvements measured as changes in median scores were rather small. Seven of 12 patients in the tramadol group who completed the study reported a favorable effect on the PGIC compared with only 1 of 11 in the placebo group. But adverse events were substantial in our study; 91% of those on tramadol compared with 58% on placebo experienced side-effects, which caused 10 of the 23 patients (43%) on the active drug to withdraw from the study. Altogether, 11 patients (48%) withdrew from active treatment. This is a much higher withdrawal rate than was reported in any of the 4 published placebo-controlled trials on tramadol for neuropathic pain where 9% to 17% left the trials owing to adverse events. 11,28 30 In a study comparing the use of tramadol versus clomipramin for postherpetic neuralgia, Go bel and Stadler 31 reported a withdrawal rate similar to ours owing to adverse events for those on tramadol (41%). Reported withdrawal rates (from the active drug group) from trials for SCI pain have varied: 10% on lamotrigine, 32 18% on amitryptilin (both nociceptive and neuropathic pain), 5 30% on pregabalin, 4 and 50% on gabapentin. 12 The reason behind the sometimes high withdrawal rates can be linked to the complexity of an SCI, but in our study, withdrawal was nearly always linked to adverse events. In 2 of the studies on tramadol for neuropathic pain, 28,29 the initial dose was lower than in our study (50 and 100 mg, respectively), and slow-dose titration was found to increase tolerability. 33 Our dosing regime might have contributed to the adverse events experienced and thus, to compliance. Of the 10 patients on tramadol who did not fulfil the treatment protocol in our study owing to adverse events, 1 patient withdrew after taking 1 tablet and 2 patients during days 2 to 4, that is, while they were still on the initial 150-mg daily dose. Most patients, though, withdrew at higher dosages. We might have seen fewer patients withdrawing if the initial dose had been lower. Another reason for the number of adverse events could be the allowance of concomitant analgesic drugs in our study, which was not allowed in most other tramadol studies. 11,28,29,31 Owing to the high withdrawal rate seen in our study, the results should be interpreted with care. It seems that patients who chose to cease medication with tramadol in our study owing to adverse events experienced no pain relief. The median values of their diary ratings at baseline compared with their last day in the trial were equal (3) with similar IQRs. Opioids have long been considered to have no or very little effect on neuropathic pain, 34 but during the last few years this has changed and studies have shown opioids to be effective in at least peripheral neuropathic pain. 6,35 The number of patients needed to treat (NNT), to achieve 50% pain relief was reported to be 2.5 for opioids (2.5 for morphine and 2.6 for oxycodone) and 3.9 for tramadol. 6 In comparison, an NNT of 4.7 was reported for gabapentin/pregabalin and 3.1 for TCAs, 6 which suggests that opioids and gabapentin/ pregabalin and TCAs may be similarly effective in decreasing pain intensity. Some have argued that it is mainly the affectivemotivational component of pain that opioids alleviate, not the pain intensity itself. 36 Tramadol is a m-opioid agonist, but its pain-relieving effect is probably also owing to its monoaminergic action. In our study, we observed a significant decrease in pain intensity ratings in those on the active drug compared with placebo, but we detected no decrease in pain affect ratings. The greatest decreases were actually seen in ratings of worst pain intensity. The daily diary ratings also showed that patients who completed the study on the active drug decreased their pain intensity ratings (median value and IQRs) compared with those who withdrew from the active drug group and to those on placebo. We thus conclude that tramadol might have a beneficial effect on neuropathic pain after an SCI and that this effect consists primarily of a decrease in pain intensity, not pain affect. But the adverse events were substantial and caused many patients to withdraw from our study. The risk of dependence and abuse is considered low for tramadol, and the drug is believed to cause fewer opioidlike adverse reactions and less pronounced withdrawal symptoms than traditional opioids. 37 The Swedish Registry for adverse events (Swedis) reported in November 2006 problematic withdrawal reactions from tramadol. 9 Abstinence and withdrawal symptoms typical for opioids (shivering, pain, perfuse sweating, nausea, and insomnia) and symptoms atypical for opioids, such as worry/anxiety, hallucinations, sense of unreality, tremors, and muscle spasms, were listed. These withdrawal symptoms were reported to be present already at low doses (50 mg daily) and from treatment periods as short as 1 week. Analgesic efficacy of tramadol has previously been shown in peripheral neuropathic pain and now in patients with SCI who mainly have central neuropathic pain. The drug option has been included in recent treatment algorithms. Concerning pain relief and side-effects in the treatment of peripheral neuropathic pain, Finnerup et al 6 suggest use of tramadol (or oxycodone) when gabapentin/ pregabalin and TCAs, and possibly serotonin and norepinephrine reuptake inhibitors, have been tried. Siddal and Middleton 1 recommend tramadol (or another weak opioid) or TCAs as second-line treatment after gabapentin in patients with SCI. 182 r 2009 Lippincott Williams & Wilkins

7 Clin J Pain Volume 25, Number 3, March April 2009 Tramadol in Neuropathic Pain After Spinal Cord Injury An important question is whether tramadol has a place in treating SCI-related neuropathic pain with regard to the effect and the adverse events seen in our trial. Tramadol can decrease pain intensity and is a reasonably cost-effective drug. But like other opioids, it is associated with negative side-effects and can cause dependence and withdrawal reactions. We agree with Finnerup et al 6 that gabapentin/pregabalin, TCAs, and serotonin and norepinephrine reuptake inhibitors should be tried before considering tramadol or other opioid drugs. Tramadol, or other opioids, may then be tried as an adjuvant or as a solitary drug. We do recommend, however, a low initial dose with slow and individual dose titration with respect to adverse events. Tramadol may also be used as a rescue medication for decreasing the worst pain intensity. As with other opioids, dependence needs to be closely monitored and the drug needs slow phasing out even when used for short periods of time. We also observed a significant difference in sleep quality and global life satisfaction between the tramadol and the placebo groups. The median values seem to indicate that the tramadol group ratings were stable over time and that differences between the groups were owing to changes in the placebo group ratings. No effect on these parameters was reported in 2 other studies on tramadol. 28,29 One weakness in our study was the limited number of patients included despite involving regional rehabilitation departments for SCI rehabilitation in the mid-sweden region. This stresses the need for national and international multicentre trials in SCI populations in the future. Another limitation in our study was that patients differed to a large extent regarding pain ratings despite that they were randomized to either active or placebo drug. Likely, this could have been avoided with a larger study sample. It is, however, difficult to speculate if these differences have influenced the results. ACKNOWLEDGMENTS The authors thank statistician Jan Kowalski for carrying out the statistical analysis and Anders Ha ggstro m, O rebro, Birgitta Norell-Ka gstro m, Sandviken, and Marie Ko hler, Falun for their collaboration. REFERENCES 1. Siddall PJ, Middleton JW. A proposed algorithm for the management of pain following spinal cord injury. Spinal Cord. 2005;44: Tai Q, Kirshblum S, Chen B, et al. Gabapentin in the treatment of neuropathic pain after spinal cord injury: a prospective, randomized, double-blind, crossover trial. J Spinal Cord Med. 2002;25: Levendoğlu F, O ğ+un CÖ, Özerbil Ö, et al. Gabapentin is a first line drug for the treatment of neuropathic pain in spinal cord injury. Spine. 2004;29: Siddall PJ, Cousins MJ, Otte A, et al. Pregabalin in central neuropathic pain associated with spinal cord injury: a placebocontrolled trial. Neurology. 2006;67: Cardenas DD, Warms CA, Turner JA, et al. Efficacy of amitryptiline for relief of pain in spinal cord injury: results of a randomized controlled trial. Pain. 2002;96: Finnerup NB, Otto M, McQuay HJ, et al. Algorithm for neuropathic pain treatment: an evidence based proposal. Pain. 2005;118: Eide PK, Stubhaug A, Stenehjem AE. Central dysesthesia pain after traumatic spinal cord injury is dependent on M-Methyl- D-Aspartate receptor activation. Neurosurgery. 1995;37: Attal N, Guirimand F, Brasseur L, et al. Effects of IV morphine in central pain: a randomized placebo-controlled study. Neurology. 2002;58: Persson M, Thurelius J, Korkmaz S, et al. Withdrawal reactions from tramadol a larger problem than expected (Swedish)? Available at: NewsPage 5449.aspx. 10. Hollingshead J, D+uhmke RM, Cornblath DR. Tramadol for neuropathic pain (review). Cochrane Database Syst Rev. 2006;3: CD Sindrup SH, Andersen G, Madsen C, et al. Tramadol relieves pain and allodynia in polyneuropathy: a randomised, doubleblind, controlled trial. Pain. 1999;83: Tsai Y-C, Sung Y-H, Chang P-J, et al. Tramadol relieves thermal hyperalgesia in rats with chronic constriction injury of the sciatic nerve. Fundam Clin Pharmacol. 2000;14: Apaydin S, Uyar M, Karabay NU, et al. The antinociceptive effect of tramadol on a model of neuropathic pain in rats. Life Sci. 2000;66: Siddall PJ, Yezierski RP, Loeser JD. Pain following spinal cord injury: clinical features, prevalence, and taxonomy. IASP Newsletter Borg G. Borg s Perceived Exertion and Pain Scales. Champaign IL: Human Kinetics; FASS. The Swedish Association of the Pharmaceutical Industry. 2003: Lundeberg T, Lund I, Dahlin L, et al. Reliability and responsiveness of three pain assessments. J Rehabil Med. 2001; 33: Bergstro m G, Jensen IB, Bodin L, et al. Reliability and factor structure of the Multidimensional Pain Inventory Swedish language version (MPI-S). Pain. 1998;75: Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand. 1983;67: Fugl-Meyer A, Bra nholm I, Fugl-Meyer K. Happiness and domain-specific life satisfaction in adult northern Swedes. Clin Rehabil. 1991;5: Fugl-Meyer AR, Melin R, Fugl-Meyer KS. Life satisfaction in 18- to 64-year-old Swedes: in relation to gender, age, partner and immigrant status. J Rehabil Med. 2002;34: Åkerstedt T, Torsvall L. Medical, psychological, and social aspects of shift work at the specialised steel-making plants in So derfors. National Institute for Psychosocial Medicine. Rep Stress Res Guy W. ECDEU Assessment Manual for Psychopharmacology, Revised Rockville, Maryland: US Department of Health, Education and Welfare; Farrar JT, Young JP Jr, LaMoreaux L, et al. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain. 2001;94: Svensson E. Ordinal invariant measures for individual and group changes in ordered categorical data. Stat Med. 1998;17: Svensson E. Guidelines to statistical evaluation of data from ratings scales and questionnaires. J Rehab Med. 2001;33: Hanley MA, Jensen MP, Ehde DM, et al. Clinically significant change in pain intensity rating in persons with spinal cord injury or amputation. Clin J Pain. 2006;22: Harati Y, Gooch C, Swenson M, et al. Double-blind randomized trial of tramadol for the treatment if the pain of diabetic neuropathy. Neurology. 1998;50: Boureau F, Legallicier P, Kabir-Ahmadi M. Tramadol in postherpetic neuralgia: a randomized, double-blind, placebocontrolled trial. Pain. 2005;118: Arbaiza D, Vidal O. Tramadol in the treatment of neuropathic cancer pain. A double-blind, placebo-controlled study. Clin Drug Invest. 2007;27: Go bel H, Stadler T. Treatment of pain due to postherpetic neuralgia with tramadol. Results of an open, parallel pilot r 2009 Lippincott Williams & Wilkins 183

8 Norrbrink and Lundeberg Clin J Pain Volume 25, Number 3, March April 2009 study vs clomipramine with or without levomepromazine. Clin Drug Invest. 1995;10: Finnerup NB, Sindrup SH, Bach FW, et al. Lamotrigine in spinal cord injury pain: a randomized controlled trial. Pain. 2002;96: Ruoff GE. Slowing the initial titration rate of tramadol improves tolerability. Pharmacotherapy. 1999;19: Arnér S, Meyerson B. Lack of analgesic effect of opioids on neuropathic and idiopathic forms of pain. Pain. 1988;33: Dworkin RH, O Connor AB, Backonja M, et al. Pharmacological management of neuropathic pain: evidence-based recommendations. Pain. 2007;132: Kupers RC, Konings H, Adriaensen H, et al. Morphine differentially affects the sensory and affective pain ratings in neurogenic and idiopathic forms of pain. Pain. 1991;47: Desmeules JA. The tramadol option. Eur J Pain. 2000;4 (suppl A): r 2009 Lippincott Williams & Wilkins