Abstract
Background: Tricyclic antidepressants (TCA) are often used in the treatment of painful polyneuropathy. Venlafaxine is a serotonin and weak noradrenaline reuptake inhibitor antidepressant with a different profile of other pharmacologic actions from those of TCA.
Objective: To test if venlafaxine would relieve painful polyneuropathy and compare its possible efficacy with that of the TCA imipramine.
Methods: The study design was randomized, double blind, and placebo controlled, with a three-way crossover. Forty patients were assigned to one of the treatment sequences, and 29 completed all three study periods. The daily doses were venlafaxine 225 mg and imipramine 150 mg. During the three treatment periods, each of 4 weeks’ duration, patients rated pain paroxysms, constant pain, and touch- and pressure-evoked pain by use of 0- to 10-point numeric rating scales.
Results: The sum of the individual pain scores during treatment week 4 was lower on venlafaxine (80% of baseline score; p = 0.006) and imipramine (77%; p = 0.001) than on placebo (100%) and did not show any statistical difference between venlafaxine and imipramine (p = 0.44). The individual pain scores for pain paroxysms, constant pain, and pressure-evoked pain showed a similar pattern, whereas touch-evoked pain was uncommon and was not altered by any of the drugs. Numbers needed to treat to obtain one patient with moderate or better pain relief were 5.2 for venlafaxine and 2.7 for imipramine.
Conclusion: Venlafaxine relieves pain in polyneuropathy and may be as effective as imipramine.
Tricyclic antidepressants (TCA) are first-line drugs for the treatment of pain in polyneuropathy.1 However, contraindications such as cardiac disease and intolerable side effects limit their usefulness in clinical practice. Therefore, effective drugs that are better tolerated are needed.
TCA probably relieve pain mainly by inhibiting presynaptic reuptake of serotonin and noradrenaline and possibly also by an NMDA-antagonistic and a sodium channel-blocking effect.2 Selective serotonin reuptake inhibitors (SSRI) have been found to be either ineffective3 or less effective4,5⇓ than TCA in painful polyneuropathy. Venlafaxine is a serotonin and weak noradrenaline reuptake inhibitor,6 and it thus may provide better pain relief than the SSRI. The serotonin and noradrenaline reuptake inhibition is provided by both the r and the s enantiomers of venlafaxine and their O-desmethylated metabolites.7-9⇓⇓ It appears that r-O-desmethylvenlafaxine, which is present in the highest steady-state plasma levels, is the most potent inhibitor of both noradrenaline and serotonin reuptake. Relating in vitro reuptake inhibitory concentrations to steady-state plasma concentrations suggests that serotonin reuptake inhibition will be maximal at low doses (<100 mg/day), whereas noradrenaline reuptake can be expected to increase over the dose range from 100 to 400 mg/day.
Animal experiments have shown venlafaxine to have an analgesic effect,10,11⇓ and it is indicated that the effect is mediated mainly via adrenergic mechanisms and κ- and δ-opioid receptors.11 In one experiment, venlafaxine relieved thermal hyperalgesia in experimental mononeuropathy in rats,10 and recently we reported an increased threshold at which repetitive electrical stimulation shows pain summation in humans.12
We tested if venlafaxine, in doses expected to induce a balanced inhibition of reuptake of serotonin and noradrenaline, relieved pain in polyneuropathy and if the size of this potential effect was equivalent to that of a TCA.
Methods.
Patients.
Men and women with painful polyneuropathy were recruited from the Departments of Neurology at the University Hospitals in Odense and Aarhus. Inclusion criteria comprised symptoms compatible with polyneuropathy present for >6 months, polyneuropathy diagnosis confirmed by nerve conduction studies, and age between 20 and 70 years. Further, at study entry, the patients had, during 1 week off medication, a median pain rating of at least 4 on a 0- to 10-point numeric scale (0 = no pain, 10 = worst possible pain) for their individual most bothersome pain symptom. We excluded patients with causes of pain other than polyneuropathy, previous allergic reactions to venlafaxine or imipramine, treatment with monoamine oxidase inhibitors or quinidine, cardiac conduction disturbances or recent myocardial infarction, pregnancy, and severe terminal illness. Further, patients genotyped to be poor metabolizers of sparteine/debrisoquine13 were excluded, as both venlafaxine and imipramine are metabolized via this enzyme.14,15⇓ The study was approved by the Regional Ethics Committees (Funen and Aarhus, j.no. 19980124 PMC) and the Danish Medicines Agency (j.no. 2612-513), and all patients consented to participate on the basis of verbal and written information.
Study design and randomization.
For patients already treated with drugs for their neuropathic pain, their treatment was gradually tapered off and discontinued during a prestudy period of maximally 1 week. After 1 further week for baseline observations, the patients entered a double-blind, three-way, crossover treatment sequence for 4 + 4 + 4 weeks with venlafaxine, imipramine, and placebo. A washout period of at least 1 week separated the treatment periods. One week for washout was considered sufficient, as the study included only extensive metabolizers of sparteine and the decline in TCA effect seems to parallel the decline in serum drug concentrations.16⇓
Assignment to one of the six possible treatment sequences was random via a computer-generated randomization code. The randomization plan was generated by one author who was not involved in the conduct of the trial. The study drugs were packed in boxes marked with patient number and treatment period. After the baseline period, the patients were numbered consecutively and were treated with the study drugs with the corresponding randomization number. Sealed envelopes with treatment sequence for each patient were present at the study sites for emergency situations. Double-dummy technique was employed, as venlafaxine and its corresponding placebo (capsules) had a different appearance than imipramine and its corresponding placebo (tablets). The dose of venlafaxine was 37.5 mg b.i.d. in the first week, 75 mg b.i.d. the second week, and 112.5 mg b.i.d for the remaining 2 weeks; the dose of imipramine was 25 mg b.i.d. in the first week, 50 mg b.i.d. the second week, and 75 mg b.i.d. for the remaining 2 weeks. Placebo tablets to venlafaxine and imipramine were dosed similarly in the placebo phase and the treatment periods as necessary to keep blinding according to the double-dummy technique. Study medication was taken at breakfast and at dinner. Up to 6 tablets of 500 mg of paracetamol could be used daily as escape medication during all study phases.
Recording.
Pain measures.
The sum of the patient’s daily separate ratings of constant pain (burning, pressing, or deep aching), pain paroxysms, touch-evoked pain, and pain on pressure on four separate 0- to 10-point numeric rating scales throughout baseline and double-blind treatment periods served as the primary outcome measure with value for week 4 used in the primary statistical analysis. The ratings were performed in the morning and covered the last 24 hours. The stimulus-evoked pains were rated as experienced during activities of daily living, for example, pain on touch by bedclothes and pain under feet on walking. Secondary outcome measures were 1) ratings of specific pain phenomena on the numeric scales, 2) number of paracetamol tablets used per week as escape medication, and 3) patient’s global evaluation of pain relief (“complete,” “good,” “moderate,” “slight,” “none,” or “worse”). The statistical analysis of paracetamol consumption and pain rating included only values from week 4 of each treatment period.
Sensory testing.
Sensory function was assessed by measurements of touch detection thresholds with von Frey hairs, cold and heat detection and pain thresholds with a Termotest (Somedic AB, Stockholm, Sweden), pressure pain thresholds with an electronic pressure algometer (Somedic AB), and rating of pain by repetitive (2-Hz) pinprick17 and electronic toothbrush18 stimulation at the end of baseline and after each of the treatment periods.
Drug concentrations.
Serum concentrations of venlafaxine and imipramine and their metabolites were determined by high-pressure liquid chromatography from blood samples drawn 12 hours after medication during the last week of each treatment period by methods described previously.12,19⇓ The assay method was not enantio-selective.
Side effects.
Following all three treatment periods, patients were asked to categorize side effects as “not present,” “light,” “bothersome,” or “unacceptable.” If side effects were present, patients were requested to name them.
Data analysis and statistics.
For daily numeric scale ratings, a mean value for each week was determined, as was the total weekly paracetamol consumption. The values of week 4 were used for statistical analysis. If the patient stopped the treatment period prematurely, the last week with observations was carried forward and used in the analysis. The analysis of treatment effects was based on ordinary least-squares regression with patient, treatment, and period as categorical covariates.20 Existence of carryover effects was assessed by adding the treatment of the previous period as a covariate to the model. We used PROC GLM of SAS version 8.1 (SAS Institute, Cary, NC). A significance level of p = 0.05 was used for overall analysis and Bonferroni-corrected significance level of p = 0.05/3 = 0.017 for interdrug and placebo comparisons. A similar procedure was employed for ratings of the four individual pain phenomena, paracetamol consumption, and measures of pain processing. Pain relief and side effects (categorical data) were analyzed with Friedman tests followed by Wilcoxon’s signed rank tests when appropriate. Number needed to treat (NNT) to obtain at least 50% pain relief was calculated from pain relief data, with this degree of pain relief corresponding to at least moderate pain relief.21
Exploratory analysis of different factors’ influence on response was done by dichotomizing patients as responders and nonresponders and calculating an odds ratio (OR) for the factor or using Mann–Whitney tests (plasma drug concentrations). Responders for each drug separately were defined as patients with at least moderate pain relief (pain relief scale).
It was calculated that a sample size of approximately 40 patients would suffice to obtain an acceptable risk of type 1 error of 1.7% (corresponding to 5%/3 according to the Bonferroni correction for multiple comparisons) and type 2 error of 10% detecting a difference in the summary pain score (0 to 40 points) of 3 points assuming a correlation coefficient of 0.8 for the correlation between pretreatment and treatment values of pain measure and an SD of the measure of 8 (estimated from previous trials).
Results.
Seventy patients with polyneuropathy and pain of >6 months’ duration were screened for participation in the study; 40 of these entered the study, with 33 patients completing (periods 1 through 3) or partially completing (two periods) the study (figure 1). One patient was excluded after having completed the trial but before the study was unblinded, owing to unacceptable high continuous concomitant intake of tramadol during all study phases. Protocol deviations or violations for patients included in the data analysis (n = 32) were as follows: One patient had pain rating missing for study period 3 (imipramine); one patient stopped medication after 2 weeks in period 1 (venlafaxine) because of side effects (last observation carried forward); results from period 2 were omitted for one patient owing to noncompliance and hospitalization for pronounced urinary tract infection; one patient had period 2 (placebo) shortened to 12 days owing to lack of pain relief; and one patient had treatment period 3 (imipramine) stopped prematurely owing to skin rash. Patients completing only two periods (n = 2) or with data only from two periods (n = 2) were included in data analysis when possible. Details on the patients completing the trial are given in table 1. Treatment sequence assignment for patients included in data analysis was venlafaxine–placebo–imipramine four, venlafaxine–imipramine–placebo five, placebo–venlafaxine–imipramine six, placebo–imipramine–venlafaxine six, imipramine–venlafaxine–placebo five, and imipramine–placebo–venlafaxine six.
Figure 1. Patient flow in study.
Table 1 Clinical data on patients completing or partially completing the trial
Tablet counts after each treatment period ensured compliance with treatment, which was also confirmed by measurements of serum drug concentrations (see below). The mean percentages of compliance were venlafaxine 97% (SD 8%), imipramine 96% (9%), and placebo 98% (6%).
The study was stopped before the stipulated number of patients had completed the trial, as the study drug had expired and new supplies were not available.
In none of our analyses did we find a significant period or carryover effect.
Main outcome measures.
Pain summation during the study is shown in figure 2. Analysis of variance showed change in this relative measure of total pain (p = 0.0011; table 2). Further analysis revealed lower pain score during venlafaxine than during placebo (p = 0.004; Bonferroni-corrected significance level p = 0.017) and no difference between venlafaxine and imipramine (see table 2). Pain scores were lower on imipramine than placebo (p = 0.0005).
Figure 2. Summation of pain scores given as percentage of the corresponding score during baseline period for placebo (○), venlafaxine (▴), and imipramine (•) during the three treatment periods of 4 weeks’ duration.
Table 2 Mean (SD) pain scores and paracetamol consumption from week 4 of each treatment period or last observation carried forward
Secondary effect variables.
Venlafaxine and imipramine reduced pain paroxysms, pressure-evoked pain, and constant pain (see table 2). Touch-evoked pain was uncommon (see table 1) and was not altered by any of the drugs (see table 2).
Paracetamol consumption in treatment week 4 was influenced by treatment (see table 2), but the interdrug differences in consumption did not reach significance, although consumption during imipramine was lower than during placebo.
Distribution with respect to pain relief category for the treatments is detailed in table 3. There was a trend of more relief with venlafaxine than with placebo (p = 0.073) and no difference between the two active drugs (p = 0.159). Complete, good, or moderate pain relief was obtained more frequently with imipramine than placebo (p = 0.001) but not with venlafaxine, which tended to induce this degree of pain relief less often than imipramine (p = 0.07). NNT for these pain relief categories were imipramine 2.7 (95% CI 1.8 to 5.5) and venlafaxine 5.2 (95% CI 2.7 to 5.9).
Table 3 Effect as evaluated by categorical pain relief scale
Exploratory data analysis.
Patients with diabetic neuropathy tended to be more likely to obtain clinically relevant pain relief (moderate, good, or complete pain relief) than nondiabetic patients with imipramine (OR 2.7, 95% CI 0.6 to 11.3) and venlafaxine (OR 5.6, 95% CI 0.9 to 34.6).
None of the sensory tests (touch detection, cold detection, heat detection) or experimental pain tests (heat pain, cold pain, pressure pain, pain on 2-Hz repetitive stimulation, toothbrush pain) differed between the three treatments, either as tested in the total sample with responders and nonresponders together (analysis of variance, p = 0.25 to 0.92) or as tested separately within the subgroup of venlafaxine responders (p = 0.09 to 1.0) and within the subgroup of imipramine responders (p = 0.058 to 0.70).
Serum concentrations of venlafaxine (n = 21) and imipramine (n = 26) were obtained in only a subset of patients owing to logistic problems. Mean (range) serum concentrations were venlafaxine 190 (18 to 860) nM, O-desmethylvenlafaxine 236 (31 to 502) nM, imipramine 353 (30 to 2,377) nM, and desipramine 210 (0 to 862) nM. The serum concentrations of venlafaxine and the sum of venlafaxine and O-desmethylvenlafaxine were higher in responders to venlafaxine than in nonresponders, whereas there was no difference in O-desmethylvenlafaxine concentrations (figure 3). Imipramine (mean 422 versus 310 nM), desipramine (mean 208 versus 211 nM), and sum of imipramine and desipramine (mean 630 versus 521 nM) did not differ between responders and nonresponders (p = 0.57 to 0.98).
Figure 3. Serum drug concentrations in venlafaxine responders (filled columns) and nonresponders (open columns). Open circles = outliers; Venla = venlafaxine; O-d-venla = O-desmethylvenlafaxine; Sum = venlafaxine plus O-desmethylvenlafaxine; *p = 0.02; **p = 0.006.
Side effects.
Side effect rating and side effects experienced with each of the two active treatments and placebo are detailed in table 4. Side effect ratings were not different between the three treatments, although there was a trend toward more side effects during the two active treatments. The pattern of side effects is also given in table 4, which shows higher incidence of dry mouth and sweating during imipramine and tiredness during venlafaxine administration.
Table 4 Side effect rating by categorical scale and specific side effects reported
Discussion.
Unblinding is always an issue in placebo-controlled trials with antidepressants, because side effects may unmask the current treatment. In the current study, this seems not to have been a major problem, as a considerable number of patients reported side effects on placebo and the overall severity of side effects was quite similar between the two active drugs. However, the higher incidence of dry mouth during imipramine and tiredness on venlafaxine may have unmasked treatment in some patients. This especially holds for previously TCA-treated patients (n = 21).
The study shows that venlafaxine relieves pain in polyneuropathy, and the primary efficacy variable shows no statistical difference between venlafaxine and imipramine. Upon examination of effect size using NNT for >50% pain relief, it is suggested that imipramine is more efficacious than venlafaxine, although the confidence intervals for the NNT are wide. The NNT for imipramine of 2.7 is within the range calculated for TCA in a recent meta-analysis.1 The NNT for venlafaxine of 5.2 is outside the range found for TCA. However, there is some uncertainty about these estimates, and the assay of serum concentrations of venlafaxine and its metabolite in relation to efficacy suggested that efficacy may be improved by employing higher doses or concentration-guided dosing.
Venlafaxine appears in our study to be a little weaker than in another recent study with venlafaxine 150 to 225 mg daily against placebo in painful diabetic polyneuropathy, in which an NNT of 4.2 can be calculated from preliminary data.22 However, it is, in all cases, better than the NNT of 6.7 for SSRI.1 It may be that the weaker effect of venlafaxine in the current study is caused by the nonhomogeneous study group including about 50% nondiabetic individuals, as responder frequency may be higher in diabetic patients.
The serum concentration of venlafaxine and its metabolites obtained in most patients in the current study seems to be within the range that is expected to provide inhibition of serotonin reuptake, whereas it is suboptimal for obtaining noradrenaline reuptake. The relation between drug level and effect of venlafaxine may reflect the suboptimal levels obtained to induce noradrenaline reuptake. In all, it is strongly indicated that reuptake inhibition of both serotonin and noradrenaline is contributing to the action of TCA in painful polyneuropathy. If the lower NNT for imipramine than for venlafaxine in the current study reflects a real higher efficacy of imipramine, the additional effect of TCA may be obtained by other mechanisms, namely, the NMDA-antagonistic effect and sodium channel blockade. Venlafaxine apparently does not interfere with NMDA receptors, and it will probably not block ion channels at clinically relevant drug concentrations.23
The trend of better response in diabetic than in nondiabetic persons was seen with both venlafaxine and imipramine. This may indicate that the pain mechanisms in diabetic neuropathy differ from those seen in nondiabetic neuropathy. However, a difference in drug response between diabetic and nondiabetic subjects has not been obvious in prior studies on painful polyneuropathy.1
Venlafaxine appears not to be superior to imipramine with respect to tolerability, as a higher number of patients withdrew because of side effects to venlafaxine than to imipramine and the severity of side effects in patients completing the trial was quite similar with the two drugs. Venlafaxine could still be useful in patients with contraindications to TCA, although its efficacy as compared with other drugs for this condition, such as, for example, gabapentin and tramadol,1 needs to be explored in larger face-to-face trials.
In clinical practice, TCA are often combined with gabapentin, although such combination therapy is not evidenced by controlled trials. Combination of venlafaxine and gabapentin can, of course, also be tried in case of insufficient response to each of the drugs given as monotherapy. A recent controlled trial has shown that the addition of venlafaxine in patients unresponsive to gabapentin provides additional pain relief in a number of patients.24 Combination of venlafaxine with TCA or tramadol should not be used because monoaminergic action would be provided by both drugs in the combination.
Acknowledgments
Supported by the Danish National Research Council (NASTRA grant no. 42820) and the local research foundation at Odense University Hospital. Study medication was provided by Wyeth Lederle and Nycomed.
Acknowledgment
The authors thank research nurses Anne Lee and Anders Due Kristensen for technical assistance.
- Received December 21, 2001.
- Accepted January 13, 2003.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within the last 8 weeks.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.