Review Articles Antidepressant Treatment of Fibromyalgia A Meta-Analysis and Review LESLEY M. ARNOLD, M.D. PAUL E. KECK, JR., M.D. JEFFREY A. WELGE, PH.D.

Fibromyalgia is a common musculoskeletal pain disorder associated with mood disorders. Antidepressants, particularly tricyclics, are commonly recommended treatments. Randomized, controlled trials of antidepressants for treatment of fibromyalgia were reviewed by methodology, results, and potential predictors of response. Twenty-one controlled trials, 16 involving tricyclic agents, were identified; 9 of these 16 studies were suitable for meta-analysis. Effect sizes were calculated for measurements of physician and patient overall assessment, pain, stiffness, tenderness, fatigue, and sleep quality. Compared with placebo, tricyclic agents were associated with effect sizes that were substantially larger than zero for all measurements. The largest improvement was associated with measures of sleep quality; the most modest improvement was found in measures of stiffness and tenderness. Further studies are needed utilizing randomized, doubleblind, placebo-controlled, parallel designs with antidepressants administered at therapeutic dose ranges, using standardized criteria for fibromyalgia and systematically assessed for co-occurring psychiatric illness. (Psychosomatics 2000; 41:104–113)

F

ibromyalgia is a common musculoskeletal pain disorder of unknown etiology, characterized by widespread pain and muscle tenderness and often accompanied by fatigue, sleep disturbance, and depressed mood.1,2 With an estimated lifetime prevalence of approximately 2% in community samples,3 fibromyalgia is at least twice as common as rheumatoid arthritis and constitutes a significant public health problem.4 Although a number of pharmacologic treatments have been studied in patients with fibromyalgia, the majority of studies have examined antiReceived March 2, 1999; revised July 20, 1999; accepted July 28, 1999. From the Biological Psychiatry Program, Department of Psychiatry, University of Cincinnati College of Medicine, Cincinnati, Ohio; and the Center for Biostatistical Services, University of Cincinnati Medical Center, Cincinnati, Ohio. Address reprint requests to Dr. Arnold, Department of Psychiatry, University of Cincinnati College of Medicine, P.O. Box 670559, 231 Bethesda Avenue, Cincinnati, OH 45267–0559; e-mail: [email protected]. Copyright q 2000 The Academy of Psychosomatic Medicine.

104

depressant agents.4–6 The rationale for treating this musculoskeletal disorder with antidepressant medications is based on a number of lines of evidence. The initial impetus for examining tricyclic antidepressants as potential treatments for fibromyalgia4,5 was based on the identification of the alpha-delta NREM sleep abnormality in polysomnographic studies of patients with the disorder.7 This sleep abnormality consisted of an inappropriate intrusion of alpha waves (normally seen during wakefulness or REM sleep in sleep electroencephalograms [EEG] ) into deep sleep (characterized by delta waves on sleep EEG).7 The alpha-delta NREM sleep abnormality was postulated to be associated with the chronic musculoskeletal pain and fatigue characteristic of the syndrome and, in turn, to be mediated by an abnormality in central serotonergic neurotransmission.8 This hypothesis was supported by the experimental induction of a diffuse pain condition similar to fibromyalgia following the administration Psychosomatics 41:2, March-April 2000

Arnold et al. of p-chlorophenylalanine (PCPA), a centrally acting serotonin synthesis inhibitor.9 Thus, tertiary amine tricyclics, with higher ratios of serotonin/norepinephrine reuptake inhibition than secondary amine agents, became the initial candidate antidepressants for the treatment of fibromyalgia. Further evidence for examining antidepressants in the treatment of fibromyalgia has come from studies of psychiatric phenomenology and personal and family history of psychiatric disorders in individuals with fibromyalgia.10 Phenomenologic studies in patients with fibromyalgia found elevated rates of depressive symptoms.11–14 Conversely, elevated rates of myalgia and musculoskeletal pain were reported in patients with major depression.15–17 Most,18–23 but not all,24,25 studies that examined lifetime prevalence rates of psychiatric disorders in patients with fibromyalgia found elevated rates of major depression. Furthermore, studies of family history of psychiatric illness in patients with fibromyalgia found elevated rates of major depression in their first-degree relatives.20–22,26 Another rationale for studying antidepressants in the treatment of fibromyalgia is based on studies of tricyclics in chronic pain syndromes.27 In these studies, tricyclics were found to potentiate the effects of analgesics in a variety of conditions, including rheumatoid arthritis and neuropathic pain, and have also been reported to be the primary treatments for a variety of painful conditions.27,28 In most studies, analgesic response was not associated with the presence or absence of co-occurring depression or depressive symptoms.27 Studies of fibromyalgia treatment have found that medications, including antidepressants and cyclobenzaprine, cardiovascular fitness training, and cognitive-behavioral and behavioral therapies may be helpful to patients with fibromyalgia.29 To our knowledge, however, there has been no published meta-analysis of antidepressant treatment of fibromyalgia. We report a systematic review of all studies of antidepressant treatment of fibromyalgia and a meta-analysis of controlled trials of tricyclic agents. We address the efficacy of antidepressants in the treatment of fibromyalgia; whether specific subgroups of patients with fibromyalgia may be particularly responsive to treatment with these agents; and what issues remain to be addressed in further research. METHODS All randomized controlled trials of antidepressant treatment of fibromyalgia were identified by a Paperchaset Psychosomatics 41:2, March-April 2000

search that was augmented by reference cross-check. Twenty-one controlled trials, most involving tricyclics (16), were identified. Because tricyclic agents were the most common antidepressant medications studied, all 16 studies of tricyclics were reviewed for suitability for metaanalysis. Fourteen of the trials involving tricyclics compared improvement with tricyclic treatment to a placebo group (parallel group trials) or a placebo period (crossover trials).30–43 Those trials that did not have a placebo control44,45 could not be used to calculate standard effect size measures, which define treatment effects relative to a control. Five of the remaining studies30,34,36,43,44 did not report sufficient statistical data for effect size computations (i.e., means and standard deviations for continuous outcomes, proportions for binary outcomes). Our meta-analysis is based on the nine remaining studies.31–33,37–42 Although this total sample of studies is small, each study reported more than one effect size. Multiple effect estimates from the studies occurred in three ways: more than one treatment condition was compared on a single outcome measure; treatment conditions were compared on multiple outcome measures; and/or measurements were taken at multiple time points. We included seven outcome measures most commonly used in the studies. These measures included the patients’ self-ratings of pain, stiffness, fatigue, and sleep; the patient and the physician global assessment of improvement; and tenderness as measured by tender points. Where repeated measures were taken or multiple treatments were compared, a mean effect size was calculated. Studies of the tricyclic agents that were not included in the meta-analysis and studies of other antidepressants were also reviewed, and the study designs and outcomes of these studies were summarized. RESULTS Review of Studies of Tricyclic Agents Included in the Meta-Analysis The nine studies of tricyclic agents suitable for metaanalysis (Table 1) included two double-blind, placebocontrolled, parallel-group acute treatment trials that examined the efficacy of amitriptyline31 and dothiepin,33 respectively. In the first study,31 70 patients with fibromyalgia diagnosed according to the criteria of Smythe46 were randomized to amitriptyline, titrated to 50 mg/day for the last 5 weeks of a 9-week trial, or placebo. In the second study,33 60 patients who met the criteria of Campbell et 105

AMI v PB CBP v PB CML v MPL v PB

AMI`PB v AMI `FLX v FLX `PB v PB`PB

Carette38 Reynolds42 Bibliotti32

Goldenberg39

19

AMI v PB DTP v PB CBP v PB CBP v PB AMI v CBP v PB Carette31 Caruso33 Bennett40 Quimby41 Carette37

Note: Where repeated measures were taken or multiple treatments were compared, a mean effect size is reported. *Study completers. **Only the results of amitriptyline alone were included in the meta-analysis. NA 4 not assessed; AMI 4 amitriptyline; PB 4 placebo; DTP 4 dothiepin; CBP 4 cyclobenzaprine; CMI 4 chlomipramine; MPL 4 maprotiline; FLX 4 fluoxetine; GBL 4 global.

NA 0.14 0.61 0.22 0.68 0.49 0.68 6

NA NA NA 0.25 10.34 0.73 0.90 0.37 NA 0.88 0.55 NA 0.80 0.17 NA 0.74 NA 0.70 0.70 NA NA 8 4 3

0.30 NA 0.10 0.77 NA 0.27 0.63 0.59 NA 0.03

Tenderness Sleep

0.70 NA 0.31 1.19 0.53 NA NA 0.10 0.31 0.64

Fatigue Pain

0.635 0.45 0.42 0.75 0.45 0.58 1.32 0.44 0.98 0.44 0.58 0.76 NA 0.98 0.27 9 8 12 6 26

AMI 50 DTP 75 CBP 10-40 CBP 10-40 AMI 50 CBP 30 AMI 25 CBP 20-40 CMI 75 MPL 75 AMI 25**

Physician Global Patient Global Duration, weeks Dosage, mg/day Effective N* Design Study

The observed effect sizes for seven outcomes in nine controlled studies of tricyclic medications in the treatment of fibromyalgia TABLE 1.

106

59 52 63 40 98 (AMI) / 86 (CBP) 20 9 18

Stiffness

Antidepressant Treatment of Fribromyalgia al.47 for fibromyalgia were randomized to dothiepin (75 mg/day) or placebo for 8 weeks. Two other parallel-group, double-blind, placebocontrolled acute treatment trials included in the metaanalysis examined the efficacy of the tricyclic, cyclobenzaprine.40,41 Although cyclobenzaprine is not classified as an antidepressant, its pharmacologic properties are very similar to other tricyclics that exert antidepressant effects.48 In particular, cyclobenzaprine appears to affect both noradrenergic and serotonergic neurotransmission in the central nervous system.49–53 In the first study,40 120 patients meeting the criteria of Campbell et al.47 were randomized to cyclobenzaprine (10–40 mg/day) or placebo for 12 weeks. In the second study,41 40 female patients with fibromyalgia as determined by the investigators were randomized to cyclobenzaprine (10–40 mg/day) or placebo for 6 weeks. Another placebo-controlled parallel group trial included in the meta-analysis was a study of 208 patients meeting ACR criteria1 for fibromyalgia who were randomized to amitriptyline titrated to 50 mg/day, cyclobenzaprine titrated to 30 mg/day, or placebo for up to 26 weeks.37 Outcome was assessed according to criteria proposed by Simms et al.,43 which best distinguished patients treated with amitriptyline from placebo in a prior study.30 These criteria included physician global assessment score, patient-rated sleep score, andtenderpoint score. The remaining placebo-controlled trials of tricyclic agents in the treatment of fibromyalgia that were included in the meta-analysis were conducted with crossover designs.32,38,39,42 One placebo-controlled, double-blind study assessed the efficacy of amitriptyline.38 In this study, 22 patients with ACR criteria fibromyalgia received amitriptyline (25 mg/day) or placebo for 8 weeks followed by immediate crossover for 8 more weeks. Another crossover study assessed the efficacy of cyclobenzaprine,42 in which nine patients with fibromyalgia by Campbell et al.47 criteria were randomized to treatment with cyclobenzaprine (20– 40 mg/day) or placebo for 4 weeks followed by a 2-week washout and subsequent 4-week crossover. Bibolotti et al.32 conducted a randomized trial in which 37 patients received chlomipramine (75 mg/day), maprotiline (75 mg/ day), or placebo for 3-week intervals without intervening washout periods. Fibromyalgia was diagnosed according to criteria generated by the authors and patients were required to have a score of $7 on the Hamilton Depression Rating Scale (Ham-D) at baseline. The mean Ham-D at baseline was 23.5, suggesting that most patients were experiencing depressive symptoms of moderate-to-marked Psychosomatics 41:2, March-April 2000

Arnold et al. severity. Because of high dropout and noncompliance rates, data were analyzed only for the initial randomization (Weeks 0–3). The final study to be included in the metaanalysis was another multicomparison, double-blind, crossover study,39 in which 19 patients with fibromyalgia by ACR criteria received four, 6-week trials of fluoxetine (20 mg/day), amitriptyline (25 mg/day), combination amitriptyline (25 mg/day) and fluoxetine (20 mg/day), or placebo. However, only the results of amitriptyline alone were included in the meta-analysis. Effect Sizes For all effects, Cohen’s measure of effect size was used.54 This measure is defined as the difference between the treatment and control group means, divided by the pooled within-group standard deviation. In the case of ordinal outcomes, counts in categories of moderate and marked improvement were collapsed into a single “improved” category, and the effect size was computed as if the variable were binary. The relevant means are actually differences from pretrial to the time of the outcome measurement. However, the standard deviation of these differences was usually not reported. Instead, the standard deviation at baseline was the only measure of variability available in most cases. Because the standard deviation of the difference score depends on the correlation of the preand posttreatment measures, which was rarely available in published reports, the standard deviation at baseline was used. The likely effect of this compromise is to deflate the true size of the effect (assuming a positive correlation between multiple measures on the same individual), and as a result, our estimates of effect may tend to be conservative. Table 1 summarizes the observed effect sizes for seven outcomes in the nine trials that were suitable for meta-analysis. Figure 1 is a histogram of the observed effect sizes (including all multiple effect sizes from the same study). It appears that effect size is approximately normally distributed, with a mean treatment effect of almost one-half (0.44) of a standard deviation. Under the widely used guidelines of Cohen,54 an effect of this magnitude would be classified as “medium,” and would be recognized clinically. The histogram also suggests that the treatment effect is approximately normally distributed. This distribution is very nearly symmetrical, as supported by the similarity of the median, unweighted mean, and weighted (by number of study completers) mean. The observed effects are shown by type of outcome as a box plot in Figure 2. The dark line in each box marks Psychosomatics 41:2, March-April 2000

the median, the width of the box corresponds to the interquartile range, and the “whiskers” mark the 5th and 95th percentiles. Observations lying beyond the 5th and 95th percentiles are indicated by diamond-shaped markers. These descriptive displays suggest moderate improvement in patient and physician global assessment, pain, and sleep, and more modest improvement in fatigue, tenderness, and stiffness. To determine the precision of the effect estimates while accounting for the dependence of effect sizes from the same study, we used the mixed model framework developed by Kalaian and Raudenbush.55 Each effect size was weighted by the number of study completers. This model incorporates the hierarchical structure of the data (multiple effect size estimates nested with each study) by treating each study as a random effect. The mixed model approach is also suitable for unbalanced data (different number of effect sizes per study) and takes advantage of the correlation between outcome measures. The latter approach results in confidence intervals that are shorter (i.e., more precise) than if each outcome were treated in a separate analysis. We estimated the effect size for each type of outcome separately to determine what type of improvement contributed to the overall effect. The parameter estimates for this model are shown in Table 2. It is typical in meta-analysis to test for homogeneity of effect sizes before proceeding to search for possible effect moderators. Unfortunately, the usual tests for this purpose are not valid for correlated effect sizes. In any case, heterogeneity of effect sizes may be due to either real variability in population effects, differences in methodological quality, or both. The interpretation of a significant heterogeneity test, therefore, is open to question even for independent effect size estimates.56 More complicated models were unable to detect significant variation due to type of tricyclic, treatment duration, or dose. The latter was expressed as a percentage of a usual therapeutic antidepressant doses to equate dosage across treatments. Given the low power to detect these effects, these findings should not be taken as evidence that these effects do not exist. Summary of Other Controlled Tricyclic Studies Studies of tricyclic agents that could not be included in the analysis included a placebo-controlled, parallelgroup, acute treatment trial,30 in which 62 patients meeting the criteria of Yunus et al.57 were randomized to combinations of amitriptyline (25 mg/day) and naproxen (500 mg bid), amitriptyline (25 mg/day) and placebo, placebo 107

Antidepressant Treatment of Fribromyalgia

FIGURE 1.

Histogram of 102 observed effect sizes from nine controlled studies of tricyclic medication treatment of fibromyalgia

Median Effect Size = 0.44

20

Unweighted Mean Effect Size = 0.46 Weighted Mean Effect Size = 0.43

Frequency

15

10

5

0 –0.34 –0.18 –0.03 0.13 0.29 0.44 0.60 0.75 0.91 1.06 1.22 1.37 1.53 Effect Size (Standard Deviations)

FIGURE 2.

Box plot of effect size by type of outcome measure in nine controlled studies of tricyclic medication treatment of fibromyalgia

1.0

Stiffness

Tenderness

Sleep

Fatigue

–0.5

Pain

0.0

M.D. Global Assessment

0.5

Patient Global Assessment

Effect Size (Standard Derivations)

1.5

Outcome Measure

108

Psychosomatics 41:2, March-April 2000

Arnold et al. and naproxen (500 mg bid), and placebo with placebo for 6 weeks. Patients in both amitriptyline arms displayed significantly greater improvement in pain, sleep, fatigue, tender-point score, and patient and physician global assessment. The combination of amitriptyline and naproxen was not significantly better than amitriptyline plus placebo; however, the sample size limited the ability of this study to detect such a difference if one existed. Four other studies were conducted with crossover designs.34–36,43 The first study of this design35 compared amitriptyline (10 mg/day Week 1, 25 mg/day Week 2), temazepam (15 mg/day Week 1, 30 mg/day Week 2), and placebo in a double-blind trial consisting of three consecutive 2-week treatment intervals separated by 2-week washout periods. Ten patients who met criteria for fibromyalgia derived by the investigators and who had the NREM alpha-delta sleep anomaly on screening polysomnography were randomized. Amitriptyline produced significant improvement in severity and number of tender points, but the results of this study were limited by the small sample size and brief duration of treatment. In another study,34 36 patients with fibromyalgia according to the criteria of Smythe and Moldofsky58 received amitriptyline or placebo for 4 weeks followed by a 2-week washout and subsequent crossover. Amitriptyline was titrated at 10 mg/day for Week 1, 25 mg/day for Week 2, and 50 mg/ day for Weeks 3 and 4. Treatment with amitriptyline was associated with significant improvement in pain, tender-point sensitivity, and patient global assessment compared with placebo. In the third study, Simms et al.43 analyzed their data from an unpublished double-blind, placebo-controlled trial of 24 patients who met Yunus et al.57 criteria for fibromyalgia randomized to cyclobenzaprine (10–20 mg/day) or placebo, each for 4 weeks separated by a 2-week washout. According to response criteria developed by the authors as part of this study, 30% of cyclobenzaprine-treated patients responded. The response rate to placebo treatment was not presented. TABLE 2.

Model-estimated mean effect sizes for each outcome measure in nine controlled studies of tricyclic medication treatment of fibromyalgia

Outcome Measure Patient global Physician global Pain Fatigue Sleep Tenderness Stiffness Study variance Residual variance

Effect Size Estimate

Standard Error

0.506 0.636 0.565 0.521 0.690 0.358 0.301 0.026 3.388

0.097 0.091 0.084 0.092 0.088 0.084 0.109 0.019 0.511

Psychosomatics 41:2, March-April 2000

Finally, Jaeschke et al.36 reported on the results of 23 doubleblind, randomized multiple crossover (N-of-1) trials using amitriptyline (5–50 mg/day) or placebo for intervals ranging from 3 to 12 weeks in patients with fibromyalgia diagnosed clinically. The authors estimated that, overall, 25% of patients responded significantly to amitriptyline. Of the two controlled trials that did not use a placebo group,44,45 the first44 compared the combination of fluoxetine (20 mg/day) and cyclobenzaprine (10 mg/day) with cyclobenzaprine (10 mg/day) alone in a parallel group 12week trial involving 21 women with fibromyalgia by ACR criteria. In the second study,45 two doses (10 mg/day and 30 mg/day, respectively) of cyclobenzaprine were compared in a crossover study lasting 15 days separated by a 15-day washout involving 40 patients with fibromyalgia by Yunus et al.57 criteria. Although patients in treatment groups in both studies displayed significant reductions in most outcome measures from baseline to endpoint, the contribution of placebo response could not be assessed in either trial. Controlled Studies of Selective Serotonin Reuptake Inhibitors Two randomized controlled trials have examined the selective serotonin reuptake inhibitor (SSRI) antidepressants, fluoxetine59 and citalopram,60 respectively, in the treatment of patients with fibromyalgia. In the first study,59 42 women with fibromyalgia by ACR criteria were randomized to fluoxetine (20 mg/day) or placebo for a 6-week acute treatment trial. At the end of 6 weeks, there were no significant differences between the two groups. The results of this study were limited by a very high (57%) placebo dropout rate, insufficient sample size, brief trial duration, and restriction of fluoxetine dose to 20 mg/day. In the second study,60 43 patients with fibromyalgia by ACR criteria were randomized to citalopram (20 mg/ day) or placebo for 4 weeks. Patients who did not display a marked improvement by 4 weeks (17/19 on citalopram, all placebo-treated patients) received citalopram (40 mg/ day) or placebo for the last 4 weeks of the trial. At the end of 8 weeks, there were no significant differences in efficacy measures between the two groups. However, citalopram (40 mg/day) appears to be a more optimal antidepressant dose61 and a 4-week trial at this dose is of insufficient duration to detect possible therapeutic effect. Moreover, although both studies used psychometric rating scales to measure the severity of depressive symptoms at entry into 109

Antidepressant Treatment of Fribromyalgia the trial, neither study determined whether patients met criteria for a current or past history of major depression. Other Antidepressant Trials Three other controlled trials have examined medications with putative antidepressant properties in the treatment of patients with fibromyalgia.62–64 Alprazolam, a triazolobenzodiazepine with antidepressant effects conferred by its triazolo moiety65 (0.5–3.0 mg/day), was compared with ibuprofen (600 mg qid), the combination, or placebo in 78 patients with fibromyalgia by the criteria of Russell et al.66 in an 8-week trial.62 At the conclusion of the study, there were no significant differences in improvement between the four treatment groups. Two controlled trials assessed the efficacy of Sadenosylmethionine (SAMe),63,64 a medication with putative antidepressant activity.67 In the first study, 17 patients with fibromyalgia by investigator-derived criteria received SAMe (200 mg/day im) or placebo for 21 days in a crossover design, separated by a 2-week washout.63 Baseline Ham-D scores correlated with the number of trigger points and painful anatomic sites. Treatment with SAMe was associated with significant reductions in the number of trigger points, painful anatomic sites, and Ham-D scores compared with placebo. In the second study, 44 patients with fibromyalgia by criteria of Yunus et al.57 were randomized to SAMe (800 mg/day) or placebo for 6 weeks in a doubleblind, parallel-design trial.64 Treatment with SAMe was associated with significant improvement in pain, fatigue, morning stiffness, and global improvement compared with placebo. There was no significant difference between SAMe and placebo in reduction of tender-point number. Predictors of Antidepressant Treatment Response A number of studies have attempted to identify clinical features that might be associated with antidepressant response. These attempts have focused mainly on polysomnographically identified sleep abnormalities, in particular the NREM alpha-delta anomaly,35,38 and severity of depressive symptoms as measured by various psychometric rating scales at baseline.32,37,39,41,45,59,60,63,64 Neither study that assessed for the presence of the NREM alpha-delta sleep anomaly found it to be predictive of antidepressant treatment response.35,38 Similarly, although nine studies used changes in validated rating scales for depressive symptoms (e.g., Beck Depression Inventory, Ham-D) as potential predictors of fibromyalgia antidepressant re110

sponse, in none of these studies was such an association evident. However, psychometric rating scales do not allow for the diagnosis of a past or current episode of depression. Furthermore, some symptoms of depression and fibromyalgia overlap (e.g., sleep disturbance, fatigue), and changes in depression rating scores may be confounded by improvement in these nonspecific symptoms. The potential relevance of a current or past history of major depressive disorder as a predictor of antidepressant response is supported by the results of a recent study68 that did, indeed, find an association between such a history and response to venlafaxine. DISCUSSION The meta-analysis of nine placebo-controlled trials of tricyclic agents, including tertiary amine tricyclic antidepressants and cyclobenzaprine, in the treatment of fibromyalgia showed that the overall effect of these agents was moderate. The largest effect was found in measures of sleep quality, whereas the most modest changes were noted in measures of tenderness and stiffness. Thus, the most consistently observed improvement in these studies may have been attributed, in part, to the sedative properties of these agents. The seven other studies of tricyclics that could not be included in the meta-analysis because they lacked a placebo control group44,45 or sufficient statistical data for effect size computations30,34–36,43 had results that were consistent with the findings of the meta-analysis. Significant clinical response to tricyclic agents was observed in 25%–37% of patients with fibromyalgia,31,36–38,40,43 and the overall degree of efficacy was modest in most studies.4–6 There were important limitations to the tricyclic studies. First, no studies examined higher or standard therapeutic doses of tricyclics effective in the treatment of depression. Similarly, no studies obtained plasma concentrations of tricyclics that would be useful in determining compliance and whether the low doses used were producing plasma concentrations within the therapeutic range for depression, at least in some patients. The need for monitoring of tricyclic plasma concentrations is especially relevant to the studies that reported greater efficacy for combined amitriptyline and fluoxetine38 and combined cyclobenzaprine and fluoxetine.43 The findings of greater efficacy with combined treatment was limited by the probable pharmacokinetic interaction between fluoxetine and tricyclics, an interaction that typically elevates tricyclic plasma levels substantially.69 Considering the possible asPsychosomatics 41:2, March-April 2000

Arnold et al. sociation between mood disorder and fibromyalgia, it may be that standard antidepressant doses of tricyclics with documented therapeutic plasma concentrations are needed to obtain more than moderate improvement. Furthermore, secondary amine tricyclics (e.g., desipramine, nortriptyline), which have fewer side effects than tertiary amines, have not been examined in controlled trials in patients with fibromyalgia, but would likely be better tolerated. Second, six studies were ,4 weeks in duration.32,34,35,42,43,45 Studies lasting at least 6 weeks are required to detect significant improvement if these agents are ameliorating fibromyalgia in any way related to their antidepressant mechanism.70 Virtually no data are available regarding the intermediate and long-term efficacy of tricyclics in the treatment of fibromyalgia.4–6 In the only study extending to 26 weeks,37 neither amitriptyline nor cyclobenzaprine exerted significantly greater efficacy than placebo. Third, despite evidence of elevated prevalence rates of depression in patients with fibromyalgia and the possible prognostic significance of past or current depression, no controlled study systemically evaluated patients for a diagnosis of a mood disorder. Only five studies32,37,39,41,45 administered standardized psychometric instruments to measure depressive symptoms, but these are not adequate to establish diagnoses. Last, interpretation of the results of at least four studies was limited by small sample sizes35,38,42,44 and over half of all studies employed a crossover design.32,34–36,38,39,42–45 The frequent and noticeable side effects associated with tertiary amine tricyclics may have affected the blindedness of these trials upon switch to or from placebo. In addition, two studies32,38 omitted washout intervals that may have produced discontinuation syndromes after abrupt cessation of tricyclics,71 contributing to insomnia and myalgias during placebo treatment and reducing placebo response rates. From the available studies, the overall response of patients with fibromyalgia to treatment with nontricyclic antidepressants has not been compelling. However, the two trials that examined SSRIs59,60 were limited by relatively small sample sizes and use of only low doses or brief duration of treatment at higher doses. Although the efficacy of alprazolam was not significantly greater than placebo, the antidepressant activity of this agent may not be robust in general.65 Finally, although studies of SAMe63,64 were promising, this agent is not licensed for clinical use in the United States, and its efficacy as an antidepressant has not been definitively established.

Psychosomatics 41:2, March-April 2000

Clinical Recommendations

Although the data are limited, preliminary recommendations about the pharmacological treatment of fibromyalgia can be made. There is some evidence that patients with a history of mood and anxiety disorders may be more responsive to antidepressant treatment of fibromyalgia. Thus, a trial of any antidepressant using therapeutic doses and adequate duration of treatment should be considered for these patients. The tricyclic agents have been most well studied in patients with fibromyalgia. Although the studies have focused mostly on the tertiary amine tricyclics, secondary amine agents may be just as effective and better tolerated, allowing for titration to higher doses. Doses at the therapeutic level for the treatment of depression should be attempted if the patient does not respond to, but otherwise tolerates, lower doses.

Further Research Issues

Because there is promising, though modest, overall efficacy of most antidepressants in the treatment of fibromyalgia but significant methodologic limitations in many of the available studies, a number of design issues are worth considering for future studies. First, double-blind, placebo-controlled, parallel-group studies of sufficient sample size and adequate duration are needed. Candidate agents include tricyclic agents (including secondary amines) administered at therapeutic antidepressant doses and confirmed by measurement of serum concentrations, dose ranging studies of newer (e.g., SSRIs, venlafaxine, bupropion, nefazodone, and mirtazapine) as well as older (e.g., trazodone) antidepressants that have different pharmacologic mechanisms. Ideally, study populations meeting standardized (e.g., ACR) criteria for fibromyalgia and assessed for history of mood disorder using standardized structured interviews (e.g., SCID-IV72) should be recruited for study. The use of standardized, operationally defined outcome measures of improvement would enhance the comparative value of treatment trials in fibromyalgia. Actual impact on level of functioning and quality of life should also be included in outcome measures. Finally, studies designed to extend beyond acute treatment are needed to establish the efficacy of antidepressants as maintenance treatment for this chronic musculoskeletal disorder.

111

Antidepressant Treatment of Fribromyalgia

References

1. Wolfe F, Smythe HA, Yunus MB, et al: The American College of Rheumatology 1990 criteria for the classification of fibromyalgia. Report of the Multienter Criteria Committee. Arthritis Rheum 1990; 33:160–172 2. Hudson JI, Pope HG, Jr: The relationship between fibromyalgia and major depressive disorder. Rheum Dis Clin North Am 1996; 22:285–303 3. Wolfe F, Ross K, Anderson J, et al: The prevalence and characteristics of fibromyalgia in the general population. Arthritis Rheum 1995; 38:19–28 4. Simms RW: Controlled trials of therapy in fibromyalgia syndrome. Baillieres Clin Rheumatol 1994; 8:917–934 5. Goldenberg DL: A review of the role of tricyclic medications in the treatment of fibromyalgia syndrome. J Rheumatol 1989; 16[Suppl 19]:137–139 6. Godfrey RG: A guide to the understanding and use of tricyclic antidepressants in the overall management of fibromyalgia and other chronic pain syndromes. Arch Intern Med 1996; 156:1047– 1052 7. Moldofsky H, Scarisbrick P, England R, et al: Musculoskeletal symptoms and non-REM sleep disturbance in patients with “fibrositis syndrome” and healthy subjects. Psychosom Med 1975; 37:341–345 8. Moldofsky H, Scarisbrick P: Induction of neurasthenic musculoskeletal pain syndrome by selective sleep stage deprivation. Psychosom Med 1975; 38:35–44 9. Sicuteri F: Headache as a possible expression of deficiency of brain 5-hydroxytryptamine (central denervation sensitivity). Psychosom Med 1972; 12:69–72 10. Gruber AJ, Hudson JI, Pope HG, Jr: The management of treatmentresistant depression in disorders on the interface of psychiatry and medicine: fibromyalgia, chronic fatigue syndrome, migraine, irritable bowel syndrome, atypical facial pain, and premenstrual dysphoric disorder. Psychiatr Clin North Am 1996; 19:351–369 11. Buchwald D, Garrity D: Comparison of patients with chronic fatigue syndrome, fibromyalgia and multiple chemical sensitivities. Arch Intern Med 1994; 154:2049–2053 12. Buchwald D, Goldenberg DL, Sullivan JL, et al: The “chronic, active Epstein-Barr virus infection” syndrome and primary fibromyalgia. Arthritis Rheum 1987; 20:1132–1136 13. Goldenberg DL: Fibromyalgia and other chronic fatigue syndromes: is there evidence for chronic viral disease? Semin Arthritis Rheum 1988; 18:111–120 14. Yunus MB, Masi AT, Aldag JC: A controlled study of primary fibromyalgia syndrome: clinical features and association with other functional syndromes. J Rheumatol 1989; 16:62–71 15. Cassidy WL, Flanagan NB, Spellman M, et al: Clinical observations in manic-depression illness: a quantitative study on one hundred manic-depressive patients and fifty medically sick controls. JAMA 1957; 153:1535–1546 16. Mezzich JE, Raab ES: Depressive symptomatology across the Americas. Arch Gen Psychiatry 1980; 37:818–823 17. Stoeckle JD, Davidson GE: Bodily complaints and other symptoms of depressive reaction. JAMA 1962; 180:134–139 18. Burckhardt CS, Wiens AN, O’Reilly CA, et al: Psychological status of women with fibromyalgia (FMS). Arthritis Rheum 1990; 33(suppl):S136 19. Goldenberg SL: Psychological studies in fibrositis. Am J Med 1986; 81(suppl 3A):67–70 20. Hudson JI, Goldenberg DL, Pope HG, Jr, et al: Comorbidity of

112

fibromyalgia with medical and psychiatric disorders. Am J Med 1992; 92:363–367 21. Hudson JI, Hudson MS, Pliner LF, et al: Fibromyalgia and major affective disorder: a controlled phenomenology and family history study. Am J Psychiatry 1985; 142:441–446 22. Tariot PN, Yocum D, Kalin NH: Psychiatric disorders in fibromyalgia. Am J Psychiatry 1986; 143:813–814 23. Walker EA, Keegan D, Gardner G, et al: Psychosocial factors in fibromyalgia and rheumatoid arthritis, I: psychiatric diagnoses and functional disability. Psychosom Med 1997; 59:565–571 24. Ahles TA, Khan SA, Yunus MB, et al: Psychiatric status of patients with primary fibromyalgia, patients with rheumatoid arthritis, and subjects without pain: a blind comparison of DSM-III diagnoses. Am J Psychiatry 1991; 148:1721–1726 25. Kirmayer JL, Robbins JM, Kapusta MA: Somatization and depression in fibromyalgia syndrome. Am J Psychiatry 1985; 145:950– 954 26. Alfici S, Sigal M, Landau M: Primary fibromyalgia syndrome—a variant of depressive disorder? Psychother Psychosom 1989; 51:156–161 27. Bouckons AJ: Chronic pain: neuropsychopharmacology and adjunctive psychiatric treatment, in Textbook of Consultation-Liaison Psychiatry. Edited by Rundell JR, Wise MG. Washington, DC, American Psychiatric Press, 1996, pp 1006–1037 28. Frank RG, Kashani JH, Parker JC, et al: Antidepressant analgesia in rheumatoid arthritis. J Rheumatol 1988; 15:1632–1638 29. Alarcon GS, Bradley LA: Advances in the treatment of fibromyalgia: current status and future directions. Am J Med Sci 1998; 315:397–404 30. Goldenberg DL, Felson DT, Dinerman H: A randomized, controlled trial of amitriptyline and naproxen in the treatment of patients with fibromyalgia. Arthritis Rheum 1986; 29:1371–1377 31. Carette S, McCain GA, Bell DA, et al: Evaluation of amitriptyline in primary fibrositis: a double-blind, placebo-controlled study. Arthritis Rheum 1986; 29:655–659 32. Bibolotti E, Borghi C, Pasculli E, et al: The management of fibrositis: a double-blind comparison of maprotiline (Ludiomil), chlorimipramine, and placebo. J Clin Trials 1986; 23:269–280 33. Caruso I, Puttini PCS, Boccassini L, et al: Double-blind study of dothiepin versus placebo in the treatment of primary fibromyalgia syndrome. J Int Med Res 1987; 15:154–159 34. Scudds RA, McCain GA, Rollman GB, et al: Improvements in pain responsiveness in patients with fibrositis after successful treatment with amitriptyline. J Rheumatol 1989; 16[Suppl 19]:98–103 35. Hench PK, Cohen R, Mitler MM: Fibromyalgia: effects of temazepam, amitriptyline and placebo on pain and sleep. Sleep Res 1989; 18:335 36. Jaeschke R, Adachi J, Guyatt G, et al: Clinical usefulness of amitriptyline in fibromyalgia: the results of 23 N-of-1 randomized controlled trials. J Rheumatol 1991; 18:447–451 37. Carette S, Bell MJ, Reynolds WJ, et al: Comparison of amitriptyline, cyclobenzaprine, and placebo in the treatment of fibromyalgia: a randomized, double-blind clinical trial. Arthritis Rheum 1994; 37:32–40 38. Carette S, Oakson G, Guimont C, et al: Sleep electroencephalography and the clinical response to amitriptyline in patients with fibromyalgia. Arthritis Rheum 1995; 38:1211–1217 39. Goldenberg DL, Mayskiy M, Mossey C, et al: A randomized double-blind crossover trial of fluoxetine and amitriptyline in the treatment of fibromyalgia. Arthritis Rheum 1996; 39:1852–1859

Psychosomatics 41:2, March-April 2000

Arnold et al. 40. Bennett RM, Gatter RA, Campbell SM, et al: A comparison of cyclobenzaprine and placebo in the management of fibrositis. Arthritis Rheum 1988; 31:1535–1542 41. Quimby LG, Gratwick GM, Whitney CD, et al: A randomized trial of cyclobenzaprine for the treatment of fibromyalgia. J Rheumatol 1989; 16(suppl 19):140–143 42. Reynolds WJ, Moldofsky H, Saskin P, et al. The effects of cyclobenzaprine on sleep physiology and symptoms in patients with fibromyalgia. J Rheumatol 1991; 18:452–454 43. Simms RW, Felson DT, Goldenberg DL: Development of preliminary criteria for response to treatment in fibromyalgia syndrome. J Rheumatol 1991; 18:1558–1563 44. Cantini F, Bellandi F, Niccoli L, et al: Fluoxetine associata a ciclobenzaprina nel trattamentao della fibromialgia. Minerva Med 1994; 85:97–100 45. Santandrea S, Montrone F, Sarzi-Puttini P, et al: A double-blind crossover study of two cyclobenzaprine regimens in primary fibromyalgia syndrome. J Int Med Res 1993; 21:74–80 46. Smythe HA: Fibrositis and other diffuse musculoskeletal syndromes, in Textbook of Rheumatology, 1st Edition. Edited by Kelley WN, Harris ED Jr, Ruddy S, et al. Philadelphia, Saunders, 1981, pp 485–493 47. Campbell SM, Clark S, Tindall EA, et al: Clinical characteristics of fibrositis, I: a “blinded” controlled study of symptoms and tender points. Arthritis Rheum 1983; 26:817–824 48. Cedarbaum JM, Schleifer LS: Drugs for Parkinson’s disease, spasticity, and acute muscle spasms, in The Pharmacological Basis of Therapeutics. Edited by Gilman AG, Rall TW, Nies AM, et al. Elmsford, NY, Persimmon, 1990, pp 463–484 49. Harsch HH. Mania in two patients following cyclobenzaprine. Psychosomatics 1984; 25:791–793 50. Barnes CD, Fung SJ, Gintautus J: Brain stem noradrenergic system depression by cyclobenzaprine. Neuropharmacology 1980; 19:221–224 51. Share NN, McFarlane CS: Cyclobenzaprine: a novel centrally acting skeletal muscle relaxant. Neuropharmacology 1975; 14:675– 684 52. Kobayashi H, Hasegawa Y, One H: Cyclobenzaprine, a centrally acting muscle relaxant, acts on descending serotonergic systems. Eur J Pharmacol 1996; 311:29–35 53. Lang IM: Cyclobenzaprine effects on locus coeruleus cells in tissue slice. Neuropharmacology 1983; 22:249–252 54. Cohen J: Statistical Power Analysis for the Behavioral Sciences, 2nd Edition. Hillsdale, NJ, Erlbaum, 1988 55. Kalaian HA, Raudenbush SW: A multivariate mixed linear model for meta-analysis. Psychological Methods 1996; 1:227–235 56. Matt GE, Cook TD: Threats to the validity of research syntheses, in The Handbook of Research Synthesis. Edited by Cooper H, Hedges LV. New York, Russell Sage Foundation, 1988

Psychosomatics 41:2, March-April 2000

57. Yunus M, Masi AT, Calabro JJ, et al: Primary fibromyalgia (fibrositis): clinical study of 50 patients with matched normal controls. Semin Arthritis Rheum 1981; 11:151–172 58. Smythe HA, Moldofsky H: Two contributions to understanding of the “fibrositis” syndrome. Bull Rheum Dis 1997; 28:928–931 59. Wolfe F, Cathey MA, Hawley DJ: A double-blind placebo controlled trial of fluoxetine in fibromyalgia. Scand J Rheumatol 1994; 23:255–259 60. Njrregard J, Volkmann H, Danneskiold-Samsde B: A randomized controlled trial of citalopram in the treatment of fibromyalgia. Pain 1995; 61:445–449 61. Ekselius L, VonKnorring L, Eberhard G: A double-blind multicenter trial comparing sertraline and citalopram in patients with major depression. Int Clin Psychopharmacol 1997; 12:323–331 62. Russell IJ, Fletcher EM, Michalek JE, et al: Treatment of primary fibrositis/fibromyalgia syndrome with ibuprofen and alprazolam. Arthritis Rheum 1991; 34:552–560 63. Tavoni A, Vitali C, Bombardieri S, et al: Evaluation of S-adenosylmethionine in primary fibromyalgia. Am J Med 1987; 83[Suppl 5A]:107–110 64. Jacobsen S, Danneskiold-Samsje B, Andersen RB: Oral S-adenosylmethionine in primary fibromyalgia: double-blind clinical evaluation. Scan J Rheumatol 1991; 20:294–302 65. Rickels K, Chung HR, Csanalosi IB, et al: Alprazolam, diazepam, imipramine, and placebo in outpatients with major depression. Arch Gen Psychiatry 1987; 44:862–866 66. Russell IJ, Vipraio GA, Morgan WW, et al: Is there a metabolic basis for the fibrositis syndrome? Am J Med 1986; 81(suppl 3A):50–54 67. Lipinski JF, Cohen BM, Frankenburg F, et al: Open trial of Sadenosylmethlonine for treatment of depression. Am J Psychiatry 1984; 141:448–450 68. Dwight MM, Arnold LM, O’Brien H, et al: An open clinical trial of venlafaxine in fibromyalgia. Psychosomatics 1998; 39:14–17 69. Tollefson GD, Rosenbaum JF: Selective serotonin reuptake inhibitors, in Textbook of Psychopharmacology, 2nd Edition. Edited by Schatzberg AF, Nemeroff CB. Washington, DC, American Psychiatric Press, 1998, pp 219–237 70. Burke MJ, Preskorn SH: Short-term treatment of mood disorders with standard antidepressants, in Psychopharmacology: the Fourth Generation of Progress. Edited by Bloom FE, Kupfer DJ. New York, Raven, 1995, pp 1053–1066 71. Potter WZ, Manji HK, Rudorfer MV: Tricyclics and tetracyclics, in Textbook of Psychopharmacology, 2nd Edition. Edited by Schatzberg AF, Nemeroff CB. Washington, DC, American Psychiatric Press, 1998, pp 199–218 72. First MB, Spitzer RL, Gibbon M, et al: Structured Clinical Interview for the DSM-IV Axis I Disorders: Patient Edition (SICD-I/P, Version 2.0). New York, Biometrics Research Department, New York State Psychiatric Institute, 1995

113