Abstract

Objectives. To evaluate and compare the efficacy and acceptability of the antidepressants amitriptyline (AMT), duloxetine (DLX) and milnacipran (MLN) for FM syndrome (FMS).

Methods. Cochrane Library, MEDLINE, SCOPUS, www.clinicalstudyresults.org and www.clinicalTrials.gov were searched for randomized pharmacological placebo-controlled trials until 30 May 2010. Outcomes of interest were symptom reduction [pain, fatigue, sleep disturbance and reduced health-related quality of life (HRQOL)] and acceptability (total drop-out rates). We performed a meta-analysis of each drug vs placebo using a random-effects model and adjusted indirect analyses of the three drugs. Methodological quality was assessed by the Cochrane risk of bias tool.

Results. Ten AMT studies (612 patients), four DLX studies (1411 patients) and five MLN studies (4129 patients) met the inclusion criteria. The reported methodological quality of most AMT trials was poor, that of DLX and MLN were high. The three drugs were superior to placebo except DLX for fatigue, MLN for sleep disturbance and AMT for HRQOL. The significant effects of AMT and DLX were small and those of MLN not substantial. In adjusted indirect comparisons, AMT was superior to DLX and MLN in reduction of pain, sleep disturbances, fatigue and limitations of HRQOL. DLX was superior to MLN in reducing pain, sleep disturbances and limitations of HRQOL. MLN was superior to DLX in reducing fatigue. There were no significant differences in acceptability of the three drugs.

Conclusions. AMT cannot be regarded as the gold standard of FMS therapy with antidepressants because of the methodological limitations of its trials.

Key words: Fibromyalgia syndrome, Amitriptyline, Duloxetine, Milnacipran, Efficacy, Acceptability.

Introduction

FM syndrome (FMS) can be diagnosed by a history of a typical cluster of key symptoms (chronic widespread pain, physical fatigue, cognitive disturbances and unrestorative sleep) and the exclusion of a somatic disease that sufficiently explains the symptoms [1, 2]. Recent population-based studies found a prevalence of FMS of 2.2–6.6% in Western European countries [3, 4]. FMS patients in clinical settings report high levels of disability and cause high direct and indirect health costs [5]. Therefore, effective and safe treatment options for FMS are warranted. Antidepressants such as amitriptyline (AMT), duloxetine (DLX) and milnacipran (MLN) have been recommended by the evidence-based guidelines of the European League against Rheumatism (EULAR) and the Association of Medical Scientific Societies in Germany (AWMF) for pharmacological therapy of FMS [6, 7].

However, the evidence for pharmacological therapy of FMS with antidepressants is faced with some problems. First, there is a controversy about the gold standard of pharmacological FMS therapy. AMT was recommended as first-line pharmacological therapy of FMS by an AWMF guideline based on a qualitatitive and quantitative analysis of randomized controlled trials (RCTs) of AMT in FMS, the long-term clinical experience with this drug by most physicians and its availability in Germany [7]. In contrast, another systematic review could not make a definitive clinical recommendation regarding the efficacy of AMT [8]. Secondly, in Europe no drug has been approved specifically for FMS. AMT is approved for the treatment of chronic pain syndromes in some European countries such as Austria, Germany, Belgium and Norway (personal communication: GmbH, Hamburg, Germany). DLX and MLN were rejected by the European Medical Agency (EMEA) for the treatment of FMS because the effects of DLX and MLN were considered to be too small to be relevant for patients [9, 10]. In contrast, both drugs have been approved by the US Food and Drug Administration for FMS [11]. Thirdly, there is uncertainty about the acceptability of antidepressant therapy of FMS to patients. The assumption that new drugs, such as DLX and MLN, are better tolerated than old drugs, such as AMT [12], has to our knowledge not been tested by direct head-to-head comparisons between these drugs.

Recently, comparative effectiveness research has been promoted in the USA to assess whether new (and more expensive) drugs provide an additional benefit compared with standard therapies [13]. In the case of head-to-head comparisons, formal adjusted indirect methods summarize combined estimates and incorporate additive variance from both sets of trials. Recently, we compared the efficacy and safety of DLX, MLN and PGB in FMS [14]. Meanwhile, two other large RCTs with MLN in FMS had been published [15, 16], which had not been included in the analysis. Therefore, we saw the need to update the formal adjusted indirect comparisons of DLX and MLN and to compare their efficacy in reducing FMS symptoms and their acceptability with those of AMT.

Methods

The review was performed according to the PRISMA (preferred reporting items for systematic reviews and meta-analyses) statement [17]. Meta-analysis was performed following the recommendations of the Cochrane Collaboration [18].

Protocol

Methods of analysis and inclusion criteria were specified in advance. We used the review protocol of our systematic review on DLX, MLN and pregabalin in FMS [14].

Eligibility criteria

Types of studies and interventions

RCTs comparing AMT, DLX or MLN with pharmacological placebo were included. RCTs in which drugs were combined with any other defined treatment were excluded.

Types of participants

Patients diagnosed with FMS according to recognized criteria, of any age, were included.

Type of outcome measures

Studies should assess at least one key domain of FMS [pain, sleep, fatigue or health-related quality of life (HRQOL)] [19] and report treatment discontinuation rates (number of patients who terminated the study early for any reason during the study). Discontinuation rates are considered to be the most consistently reported estimate of treatment acceptability [20].

Data sources

We expanded our searches used for systematic reviews on antidepressants in FMS [14, 21] in the electronic bibliography databases such as MEDLINE, SCOPUS and the Cochrane Central Register of Controlled trials (until 30 May 2010). We reviewed the reference lists of included articles. We searched the web sites of the US National Institutes of Health (NIH) [22] and the Pharmaceutical Research and Manufacturers of America (PhRMA) [23] for unpublished data. We contacted the drug sponsors’ medical information departments and the first authors of published studies in cases of missing data in a full paper or in the databases detailed above.

Study selection

Two authors independently screened the titles and abstracts of potentially eligible studies identified by the search strategy described above (C.S. and W.H.). The full-text articles were then examined independently by two authors to determine whether they met the inclusion criteria (F.B. and N.Ü.).

Data extraction

Two authors independently extracted the data using standard extraction forms (C.S. and W.H.). Discrepancies were rechecked and consensus achieved by discussion. If needed, a third author reviewed the data to reach a consensus (F.P.).

Where means or s.d.s were missing, attempts were made to obtain these data by contacting five trial authors. Where s.d.s were not available from trial authors, they were calculated from t-values, CIs or s.e.s, where reported in articles [18].

Risk of bias in individual studies

The methodological quality was assessed by the risk of bias tool of the Cochrane Collaboration [18] by two authors (C.S. and N.Ü.). We assessed the following risks of bias: adequacy of randomization, concealment of treatment allocation and adequacy of blinding (patient, physician and outcome assessor), complete reporting of data and analysis of the data according to the intention-to-treat principle (one point each). We arbitrarily defined a total score of 0–2 as poor, a score of 3–4 as moderate and a score of 5 as high methodological quality.

Data synthesis and analysis

Non-parametric tests were used for the comparison of continuous variables and chi-squared tests for the comparison of categorical variables. Data are presented as median (range). A two-sided P ≤ 0.05 was considered statistically significant.

Meta-analyses were conducted using RevMan Analyses software (RevMan 5.0.24) of the Cochrane Collaboration [24] and comprehensive meta-analysis software [25]. Standardized mean differences (SMDs) were calculated by means and s.d.s for each interventions. Examination of the combined results was performed by a random-effects model (inverse variance method), because this model is more conservative than the fixed-effects model and incorporates both within- and between-study variance [26]. SMD used in Cochrane reviews is the effect size known as Hedges’ (adjusted) g. We used Cohen’s categories to evaluate the magnitude of the effect size, calculated by SMD, with g > 0.2–0.5 = small effect size, g > 0.5–0.8 = medium effect size and g > 0.8 = large effect size [27].

Numbers needed to treat (NNT) were calculated using the pooled number of observations. If the number of patients with a 30% pain reduction was not reported, it was calculated by the means and s.d.s of pain scores at baseline and post-treatment by an imputation method with worst case analysis (number of patients imputed at baseline) [28].

We performed adjusted indirect comparisons using the method described by Bucher et al. [29]. We calculated indirect relative risks (RRind) for AMT vs DLX, AMT vs MLN and DLX vs MLN, adjusted by the results of their comparisons against placebo: RRind = RRDrug 1vsPlacebo/RR Drug 2 vs Placebo. The s.e. of RRind was estimated as: s.e. (RRInd) = forumlaSE (RRDrug 1vsPlacebo)2 + SE (RR Drug 2 vs Placebo)2 [30].

Planned methods of analysis

Heterogeneity was tested using the I2 statistic with I2 > 50% indicating strong heterogeneity [18]. The similarity assumption for adjusted indirect comparisons [31] was tested by comparing the clinical (exclusion criteria, patients’ ages, percentage of women and Caucasians, treatment duration) characteristics and methodological quality of the studies.

Additional analyses

Subgroup and sensitivity analysis

Subgroup analyses were prespecified for the continent in which the study was conducted, because the EMEA doubted the long-term efficacy of MLN in European patients [10] and for AMT studies with parallel vs cross-over design because Nishishinya et al. [8] reported that none of the cross-over trials provided separate baseline data before the cross-over therapy [8]. Post hoc we decided to perform a subgroup analysis of AMT studies with and without support from pharmaceutical companies. These subgroup analyses were also used to examine potential sources of clinical heterogeneity.

We decided post hoc to perform a sensitivity analysis of AMT studies with and without extracting values from figures and of studies with and without intention-to-treat analysis.

Meta-regression analysis

A priori we planned to meta-regress the effect size on pain with the methodological quality score. Meta-regression was performed using the mixed effects model. The τ2 variance was calculated by the method of unrestricted maximum likelihood by Comprehensive Meta-analysis software [25].

Risk of bias across studies

Potential publication bias (i.e. the association of publication probability with the statistical significance of study results) was investigated using visual assessment of the funnel plot (plots of effect estimates against its s.e.) calculated by RevMan Analyses software, if appropriate (at least 10 studies available). Publication bias may lead to asymmetrical funnel plots [18].

Additionally, we performed the Egger test, in which the standardized effect size (effect size calculated by s.e.) is regressed on precision (inverse of s.e.). The intercept value is an estimate of asymmetry of funnel plot.

Positive values (>0) indicate higher levels of effect size in studies with smaller sample sizes [32].

Results

Search results

The search yielded 246 hits. We found no studies with head-to-head comparisons of the three drugs. After article review, five studies with AMT and one study each with DLX and MLN were excluded. In two studies, AMT was combined with another defined treatment (guided imagery, cardiovascular training) [33, 34]. One AMT study used a N-of-1 design [35], one was a double publication [36], one had no pharmacological placebo [37], one study had no placebo group [37] and one was only published as an abstract [38]. The data of one study each with DLX and MLN had not been published [39, 40]. Ten RCTs with AMT [41–50], four with DLX [51–54] and five with MLN [12, 15, 16, 55, 56] were included in qualitative and quantitative analysis (supplementary figure 1, available as supplementary data at Rheumatology Online).

Study characteristics

Data are reported as median (range): five of the AMT trails were (co-)sponsored by pharmaceutical companies. All DLX and MLN trials were sponsored by the manufacturer of the drug. Six of the AMT studies were conducted in North America, one in South America and three in Europe. Two of the DLX studies were conducted in North America and one each in North America/Middle America and North America/Europe. Four MLN studies were conducted in North America and one in Europe. The AMT trials recruited patients from the institutions involved. The MLN trials recruited patients from the institutions involved and by referral, the DLX studies recruited additional patients by advertisement. All trials excluded patients with severe somatic diseases. Patients with severe mental disorders were excluded in three AMT trials and in all MLN trials. The DLX studies excluded patients with severe mental disorders except for major depression. All studies used a parallel design except three AMT studies with a cross-over design. A total of 612 patients were randomized in the AMT trials, 1411 in the DLX trials and 4129 in the MLN trials. The study duration in AMT trials was 8 (4–26), in DLX trials 19 (12–27) and in MLN trials 16 (12–27) weeks (H = 9.2, P = 0.01). The average age in AMT trials was 44.4 (38–53), in DLX trials 51 (50–51) and in MLN trials 48.8 (47–50) weeks (H = 7.8, P = 0.02). The percentage of women was 95 (83–100) in AMT trials, 95 (86–100) in DLX trials and 96 (94.5–97) in MLN trials (H = 0.6, P = 0.74). The percentage of Caucasians was 92 (65–100) in AMT trials, 89 (84–92) in DLX trials and 92 (86–94) in MLN trials (H = 0.9, P = 0.65). The dosage of AMT was between 10 and 50 mg/day, of DLX between 60 and 120 mg/day and of MLN between 100 and 200 mg/day. Only acetaminophen was allowed in the AMT studies for rescue medication. Acetaminophen and ASA were allowed in the DLX studies, one of the studies allowed NSAIDs and narcotics additionally. Hydrocodone was used in the MLN studies for rescue medication and these studies additionally allowed stable doses of ASA, acetaminophen and NSAIDs (Table 1).

Table 1

Main characteristics of randomized, placebo-controlled studies with AMT, DLX and MLN in FMS

Reference; country of the principal investigator; number of research centres; recruitment Mean age, years; Women, %; Caucasian, % Diagnostic criteria of FMS; exclusion criteria Study population
 
Treatment group
 
Placebo group Funding source Outcome measures used for meta-analysis 
Screened/randomized, n (%) Randomized/completing, n (%) Randomized/completing, n (%) Duration treatment; drugs and dosage; study design; rescue medication Randomized/ completing, n (%) 
AMT          
    Carette et al. [41]; Canada; three university centres 42; 92.6; NR Smythe; severe somatic diseases, inflammatory rheumatoid disease NR 70/59 (84.3) 34/27 (79.4) 9 weeks; AMT 10–50 mg/day; parallel; only acetaminophen allowed 36/32 (88.9) Canadian Arthritis Society Pain VAS 0–10; fatigue NA; sleep NA; HRQOL NA 
    Carette et al. [42]; Canada; nine university centres and two private practices 44; 95.5; NR ACR 1990; severe somatic diseases, inflammatory rheumatoid disease NR 208/184 (88.5) 82/78 (95.1) 24 weeks; AMT 10–50 mg/day; parallel; only acetaminophen allowed 42/36 (85.7) Canadian Arthritis Society and Merck Frosst Pain NRS 0–3; fatigue NA; sleep NA; HRQOL SIP total 
    Carette et al. [43]; five Canada; one university department 44; 95.5; NR ACR 1990; severe somatic diseases, inflammatory rheumatoid disease NR 22/20 (90.9) 22/20 (90.9) 8 weeks; AMT 25 mg/day; cross-over; only acetaminophen allowed 22/20 (90.9) Canadian Arthritis Society and Merck Frosst Pain VAS 0–10; fatigue VAS 0–10; sleep VAS 0–10; HRQOL NA 
    Ginsberg et al. [44]; Belgium; not specified (probably three university departments) 46; 83; 92 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease NR 51/46 (90.2) 26/24 (92.3) 8 weeks; sustained-release AMT 25 mg/day; parallel; only acetaminophen allowed 25/22 (88) Sanofi Pain VAS 0–10; fatigue VAS 0–10 ; sleep VAS 0–10; HRQOL NA 
    Goldenberg et al. [45]; USA; not specified (probably one university department) 44 years; 95; 87 Yunus; history of peptic ulcer, cardiac arrhythmia NR 62/56 (90.3) 15/14 (93.3) 6 weeks; AMT 25 mg/day and placebo; parallel; only acetaminophen for rescue medication allowed 15/14 (93.3) Arthritis Foundation Syntex Co. Pain VAS 0–10;a fatigue VAS 0–10;a sleep VAS 0–10;a HRQOL NA 
    Goldenberg et al. [46]; USA; one university department 43; 90; 100 ACR 1990; age, somatic diseases, major depression NR 31/19 (61.2) 21/19 (90.5) 6 weeks; AMT 25 mg/day; cross-over; only acetaminophen (up to 4 g/day) allowed 19/18 (94.7) Arthritis Foundation Pain VAS 0–100; fatigue VAS 0–100; sleep VAS 0–100; quality of life FIQ total score 
    Hannonen et al. [47]; Finland; one university department and one register of FMS patients 50; 100; NR ACR 1990; age, severe somatic diseases, severe mental disorders 184/130 (70.6) 130/92 (70.8) 42/32 (76.2) 12 weeks; AMT 12.5 mg/day; parallel ; only acetaminophen (up to 4 g/day) allowed 45/30 (66.7) Roche Pain VAS 0–10; fatigue VAS 0–10 ;sleep VAS 0–10 ; quality of life SDS work 
    Heymann et al. [48]; Brazil; one university department 53; 100; 65 ACR 1990 age; severe somatic diseases; inflammatory rheumatoid disease NR 118/106 (89.8) 40/37 (92.5) 8 weeks; AMT 25 mg/day; parallel; not specified 40/33 (82.5) Not specified Pain VAS 0–10 NP; fatigue VAS 0–10 NP; sleep VAS 0–10 NP; quality of life FIQ total score 
    Kempenaers et al. [49]; Belgium; not specified (probably one university department) 38; 100; NR Yunus; severe somatic diseases and mental disorders, inflammatory rheumatoid disease NR 36/23 (63.9) 12/6 (50.0) 8 weeks; AMT 50 mg/day; parallel; only acetaminophen (dosage not reported) allowed 12/8 (66.7) Serolab Pain VAS 0–100; fatigue NA; sleep VAS 0–100; quality of life NA 
    Scudds et al. [50]; Canada; one university department 40; 88.9; NR Smythe; severe somatic diseases NR 39/36 (92.3) 39/36 (92.3) 10 weeks; AMT 10–50 mg/day; cross–over; NR 39/36 (92.3) Arthritis Society Pain MPQ;a fatigue NA ; sleep NA ; quality of life NA 
DLX          
    Arnold et al. [51]; USA; 18 research centres in USA; physician referral and advertisement for medication trial 50; 88.5; 88.5 ACR 1990; age, severe somatic diseases; inflammatory rheumatoid disease, mental disorders without major depression, disability reviews 555/207 (37.3) 207/124 (59.9) 104/58 (55.7) 12 weeks; DLX 120 mg/day; parallel; acetaminophen up to 2 g/day and ASA up to 325 mg/day 103/66 (64.1) Eli Lilly Pain FIQ VAS 0–10; sleep NA ; fatigue FIQ VAS 0–10; quality of life FIQ total score 
    Arnold et al. [52]; USA; 21 research centres in USA; physician referral and advertisement for drug trial 50; 100; 89.5 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease, mental disorders without major depression, disability reviews 745/354 (47.5) 354/215 (60.7) 234/148 (63.2) 12 weeks; DLX 60 mg/day or DLX 120 mg/day; parallel; acetaminophen up to 2 g/day and ASA up to 325 mg/day 120/68 (56.7%) Eli Lilly Pain average BPI; sleep BPI; fatigue SF-36 vitality NP; quality of life FIQ total score 
    Chappell (2008); USA; 36 study centres in five countries; physician referral and advertisement for drug trialb 51; 95.1; 92.2 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease, mental disorders except major depression NR 330/204 (61.8) 162/101 (62.3) 27 weeks; DLX 60 or 120 mg; parallel; ASA up to 325 mg/day, NSAIDS or narcotics up to 3 days for acute injury, acetaminophen up to 2 g/day 168/103 (61.3) Eli Lilly Pain average BPI; sleep NA; fatigue MFI; quality of life FIQ total 
    Russel et al. [53]; USA; 38 research centres in USA and Puerto Rico; physician referral and advertisement for medication trial 51; 94.8; 84.2 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease, mental disorders except major depression, disability reviews, refractory to treatment 1010/520 (51.8) 520/323 (62.1) 376/205 (54.5) 26 weeks; DLX 20–60 mg or DLX 60 mg or DLX 120 mg; parallel; acetaminophen up to 2 g/day and ASA up to 325 mg/day 144/72 (50) Eli Lilly Pain average BPI; sleep NA; fatigue MFI; quality of life FIQ total score 
MLN          
    Arnold et al. [16]; USA/Canada; 68 outpatient clinical/research centres in Canada and USA; own practices, referral; radio, newspaper and flyer advertisement 49; 95.3; 90.9 ACR 1990; age, somatic diseases, severe mental disorder including major depression, disability or engaged in litigation related to FMS 1947/1025 (52.6) 1025/710 (69.2) 516/353 (68.4) 18–20 weeks; flexible dose escalation to 100 mg/day; parallel; tramadol and hydrocodone allowed; acetaminophen, aspirin, NSAIDs, sedating antihistamines and chloral hydrate, no diazepam hypnotics allowed 509/357 (70.1) Cypress Pain weekly mean PED 24-h morning fatigue MFI; sleep NA, quality of life FIQ total 
    Branco et al. [3]; Portugal; 89 outpatient clinical/research centres in 13 European countries; rheumatologists, pain therapists and primary physiciansb 49; 94.5; NR ACR 1990; age, somatic diseases, severe mental disorder including general anxiety disorder or current major depression 1406/884 (62.9) 884/678 (76.7) 435/308 (70.1) 16 weeks; MLN 200 mg/day; parallel; analgesics (no details reported) allowed; anticonvulsants not allowed 449/370 (82.4) Pierre Fabre Pain weekly mean PED 24-h morning; fatigue MFI; sleep SF-36 sleep index II; quality of life FIQ total 
    Clauw et al. [12]; USA; 86 US-American outpatient clinical/research centres; rheumatologists, pain therapists and primary physiciansb 50; 96; 94.0 ACR 1990; age, somatic diseases, severe mental disorder including major depression 2270/1207 (53.2) 1200/803 (66.9) 795/513 (64.5) 15 weeks; MLN 100 or 200 mg/day; parallel; hydrocodone up to 60 mg/day 405/290 (71.6) Cypress Pain weekly mean PED 24-h morning recall pain score; fatigue MFI; sleep SF-36 sleep problem index II; quality of life FIQ total score 
    Mease et al. [19]; USA; 59 US-American outpatient clinical/research centres; rheumatologists, pain therapists and primary physiciansb 49; 95.6; 93.2 ACR 1990; age, somatic diseases, severe mental disorder including major depression 1639/888 (54.1) 888/512 (57.6) 665/367 (55.2) 27 weeks; MLN 100 or 200 mg/day; parallel; acetaminophen, ASA and NSAIDs in stable dose and hydrocodone as rescue allowed up to 60 mg/day 223/145 (65.0) Cypress Pain weekly mean PED 24-h morning recall pain score; fatigue MFI; sleep SF-36 sleep problem index II; quality of life FIQ total score 
    Vitton et al. [55]; USA; 14 US-American outpatient sites; rheumatologists, pain therapists and primary physiciansb 46–48; 96–98; 79–89 ACR 1990; age, somatic diseases, severe mental disorders 184/125 (67.9) 125/90 (72) 97/69 (71.1) 12 weeks; MLN 25–200 mg/day; parallel; stable doses of ASA, acetaminophen, NSAIDs allowed, dosage NR 28/21 (75) Cypress Pain VAS 0–10; sleep Jenkins scale;b fatigue FIQ VAS 0–10 ;b quality of life FIQ totalb 
Reference; country of the principal investigator; number of research centres; recruitment Mean age, years; Women, %; Caucasian, % Diagnostic criteria of FMS; exclusion criteria Study population
 
Treatment group
 
Placebo group Funding source Outcome measures used for meta-analysis 
Screened/randomized, n (%) Randomized/completing, n (%) Randomized/completing, n (%) Duration treatment; drugs and dosage; study design; rescue medication Randomized/ completing, n (%) 
AMT          
    Carette et al. [41]; Canada; three university centres 42; 92.6; NR Smythe; severe somatic diseases, inflammatory rheumatoid disease NR 70/59 (84.3) 34/27 (79.4) 9 weeks; AMT 10–50 mg/day; parallel; only acetaminophen allowed 36/32 (88.9) Canadian Arthritis Society Pain VAS 0–10; fatigue NA; sleep NA; HRQOL NA 
    Carette et al. [42]; Canada; nine university centres and two private practices 44; 95.5; NR ACR 1990; severe somatic diseases, inflammatory rheumatoid disease NR 208/184 (88.5) 82/78 (95.1) 24 weeks; AMT 10–50 mg/day; parallel; only acetaminophen allowed 42/36 (85.7) Canadian Arthritis Society and Merck Frosst Pain NRS 0–3; fatigue NA; sleep NA; HRQOL SIP total 
    Carette et al. [43]; five Canada; one university department 44; 95.5; NR ACR 1990; severe somatic diseases, inflammatory rheumatoid disease NR 22/20 (90.9) 22/20 (90.9) 8 weeks; AMT 25 mg/day; cross-over; only acetaminophen allowed 22/20 (90.9) Canadian Arthritis Society and Merck Frosst Pain VAS 0–10; fatigue VAS 0–10; sleep VAS 0–10; HRQOL NA 
    Ginsberg et al. [44]; Belgium; not specified (probably three university departments) 46; 83; 92 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease NR 51/46 (90.2) 26/24 (92.3) 8 weeks; sustained-release AMT 25 mg/day; parallel; only acetaminophen allowed 25/22 (88) Sanofi Pain VAS 0–10; fatigue VAS 0–10 ; sleep VAS 0–10; HRQOL NA 
    Goldenberg et al. [45]; USA; not specified (probably one university department) 44 years; 95; 87 Yunus; history of peptic ulcer, cardiac arrhythmia NR 62/56 (90.3) 15/14 (93.3) 6 weeks; AMT 25 mg/day and placebo; parallel; only acetaminophen for rescue medication allowed 15/14 (93.3) Arthritis Foundation Syntex Co. Pain VAS 0–10;a fatigue VAS 0–10;a sleep VAS 0–10;a HRQOL NA 
    Goldenberg et al. [46]; USA; one university department 43; 90; 100 ACR 1990; age, somatic diseases, major depression NR 31/19 (61.2) 21/19 (90.5) 6 weeks; AMT 25 mg/day; cross-over; only acetaminophen (up to 4 g/day) allowed 19/18 (94.7) Arthritis Foundation Pain VAS 0–100; fatigue VAS 0–100; sleep VAS 0–100; quality of life FIQ total score 
    Hannonen et al. [47]; Finland; one university department and one register of FMS patients 50; 100; NR ACR 1990; age, severe somatic diseases, severe mental disorders 184/130 (70.6) 130/92 (70.8) 42/32 (76.2) 12 weeks; AMT 12.5 mg/day; parallel ; only acetaminophen (up to 4 g/day) allowed 45/30 (66.7) Roche Pain VAS 0–10; fatigue VAS 0–10 ;sleep VAS 0–10 ; quality of life SDS work 
    Heymann et al. [48]; Brazil; one university department 53; 100; 65 ACR 1990 age; severe somatic diseases; inflammatory rheumatoid disease NR 118/106 (89.8) 40/37 (92.5) 8 weeks; AMT 25 mg/day; parallel; not specified 40/33 (82.5) Not specified Pain VAS 0–10 NP; fatigue VAS 0–10 NP; sleep VAS 0–10 NP; quality of life FIQ total score 
    Kempenaers et al. [49]; Belgium; not specified (probably one university department) 38; 100; NR Yunus; severe somatic diseases and mental disorders, inflammatory rheumatoid disease NR 36/23 (63.9) 12/6 (50.0) 8 weeks; AMT 50 mg/day; parallel; only acetaminophen (dosage not reported) allowed 12/8 (66.7) Serolab Pain VAS 0–100; fatigue NA; sleep VAS 0–100; quality of life NA 
    Scudds et al. [50]; Canada; one university department 40; 88.9; NR Smythe; severe somatic diseases NR 39/36 (92.3) 39/36 (92.3) 10 weeks; AMT 10–50 mg/day; cross–over; NR 39/36 (92.3) Arthritis Society Pain MPQ;a fatigue NA ; sleep NA ; quality of life NA 
DLX          
    Arnold et al. [51]; USA; 18 research centres in USA; physician referral and advertisement for medication trial 50; 88.5; 88.5 ACR 1990; age, severe somatic diseases; inflammatory rheumatoid disease, mental disorders without major depression, disability reviews 555/207 (37.3) 207/124 (59.9) 104/58 (55.7) 12 weeks; DLX 120 mg/day; parallel; acetaminophen up to 2 g/day and ASA up to 325 mg/day 103/66 (64.1) Eli Lilly Pain FIQ VAS 0–10; sleep NA ; fatigue FIQ VAS 0–10; quality of life FIQ total score 
    Arnold et al. [52]; USA; 21 research centres in USA; physician referral and advertisement for drug trial 50; 100; 89.5 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease, mental disorders without major depression, disability reviews 745/354 (47.5) 354/215 (60.7) 234/148 (63.2) 12 weeks; DLX 60 mg/day or DLX 120 mg/day; parallel; acetaminophen up to 2 g/day and ASA up to 325 mg/day 120/68 (56.7%) Eli Lilly Pain average BPI; sleep BPI; fatigue SF-36 vitality NP; quality of life FIQ total score 
    Chappell (2008); USA; 36 study centres in five countries; physician referral and advertisement for drug trialb 51; 95.1; 92.2 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease, mental disorders except major depression NR 330/204 (61.8) 162/101 (62.3) 27 weeks; DLX 60 or 120 mg; parallel; ASA up to 325 mg/day, NSAIDS or narcotics up to 3 days for acute injury, acetaminophen up to 2 g/day 168/103 (61.3) Eli Lilly Pain average BPI; sleep NA; fatigue MFI; quality of life FIQ total 
    Russel et al. [53]; USA; 38 research centres in USA and Puerto Rico; physician referral and advertisement for medication trial 51; 94.8; 84.2 ACR 1990; age, severe somatic diseases, inflammatory rheumatoid disease, mental disorders except major depression, disability reviews, refractory to treatment 1010/520 (51.8) 520/323 (62.1) 376/205 (54.5) 26 weeks; DLX 20–60 mg or DLX 60 mg or DLX 120 mg; parallel; acetaminophen up to 2 g/day and ASA up to 325 mg/day 144/72 (50) Eli Lilly Pain average BPI; sleep NA; fatigue MFI; quality of life FIQ total score 
MLN          
    Arnold et al. [16]; USA/Canada; 68 outpatient clinical/research centres in Canada and USA; own practices, referral; radio, newspaper and flyer advertisement 49; 95.3; 90.9 ACR 1990; age, somatic diseases, severe mental disorder including major depression, disability or engaged in litigation related to FMS 1947/1025 (52.6) 1025/710 (69.2) 516/353 (68.4) 18–20 weeks; flexible dose escalation to 100 mg/day; parallel; tramadol and hydrocodone allowed; acetaminophen, aspirin, NSAIDs, sedating antihistamines and chloral hydrate, no diazepam hypnotics allowed 509/357 (70.1) Cypress Pain weekly mean PED 24-h morning fatigue MFI; sleep NA, quality of life FIQ total 
    Branco et al. [3]; Portugal; 89 outpatient clinical/research centres in 13 European countries; rheumatologists, pain therapists and primary physiciansb 49; 94.5; NR ACR 1990; age, somatic diseases, severe mental disorder including general anxiety disorder or current major depression 1406/884 (62.9) 884/678 (76.7) 435/308 (70.1) 16 weeks; MLN 200 mg/day; parallel; analgesics (no details reported) allowed; anticonvulsants not allowed 449/370 (82.4) Pierre Fabre Pain weekly mean PED 24-h morning; fatigue MFI; sleep SF-36 sleep index II; quality of life FIQ total 
    Clauw et al. [12]; USA; 86 US-American outpatient clinical/research centres; rheumatologists, pain therapists and primary physiciansb 50; 96; 94.0 ACR 1990; age, somatic diseases, severe mental disorder including major depression 2270/1207 (53.2) 1200/803 (66.9) 795/513 (64.5) 15 weeks; MLN 100 or 200 mg/day; parallel; hydrocodone up to 60 mg/day 405/290 (71.6) Cypress Pain weekly mean PED 24-h morning recall pain score; fatigue MFI; sleep SF-36 sleep problem index II; quality of life FIQ total score 
    Mease et al. [19]; USA; 59 US-American outpatient clinical/research centres; rheumatologists, pain therapists and primary physiciansb 49; 95.6; 93.2 ACR 1990; age, somatic diseases, severe mental disorder including major depression 1639/888 (54.1) 888/512 (57.6) 665/367 (55.2) 27 weeks; MLN 100 or 200 mg/day; parallel; acetaminophen, ASA and NSAIDs in stable dose and hydrocodone as rescue allowed up to 60 mg/day 223/145 (65.0) Cypress Pain weekly mean PED 24-h morning recall pain score; fatigue MFI; sleep SF-36 sleep problem index II; quality of life FIQ total score 
    Vitton et al. [55]; USA; 14 US-American outpatient sites; rheumatologists, pain therapists and primary physiciansb 46–48; 96–98; 79–89 ACR 1990; age, somatic diseases, severe mental disorders 184/125 (67.9) 125/90 (72) 97/69 (71.1) 12 weeks; MLN 25–200 mg/day; parallel; stable doses of ASA, acetaminophen, NSAIDs allowed, dosage NR 28/21 (75) Cypress Pain VAS 0–10; sleep Jenkins scale;b fatigue FIQ VAS 0–10 ;b quality of life FIQ totalb 

aData abstracted from figures. bData not included in publication; not provided on request. cData not included in publication; provided on request. BDI: Beck depression inventory; BPI: brief pain inventory; HDRS: Hamilton depression rating scale; MFI: multidimensional fatigue inventory; MFA: multidimensional assessment of fatigue score global index; NR: not reported; NRS: numerical rating scale; SDS: Sheehan disability scale; PED: pain electronic diary; SF-36: short-form health survey; SIP: sickness impact profile; VAS: visual analogue scale.

Outcome measures

There was great variability between the studies in the outcome measures used and in the domains assessed. The AMT and DLX studies used questionnaires with visual or numerical pain ratings and four MLN studies used electronic diaries. HRQOL was assessed only by three AMT studies. Sleep quality was measured only in one DLX study. The same outcome measures were used in the four large MLN trials (Table 1). The imputing method for ITT analysis for the outcomes meta-analysed in DLX and MLN trials was the last observation carried forward method. The two AMT studies with ITT did not report on the imputing method.

Methodological quality

The reported methodological quality of seven AMT trials was poor and two were moderate. Only two studies analysed the data per ITT. We did not succeed in getting to know the details missing in the publications from the authors. The reported methodological quality of three DLX trials and two MLN trials was low; of one DLX trial and two MLN trials moderate; and of one MLN trial high. The methodological quality of all DLX and MLN studies was high for all studies after the provision of details of the methodology and some outcome data on request (Table 2). I2 statistics revealed substantial heterogeneity in the outcomes pain and HRQOL in the AMT trials; the outcome HRQOL in the DLX trials; and the outcome drop-out in the MLN trials (Table 3).

Table 2

Methodological quality of the studies

References Reported adequacy of randomization Reported adequacy of concealment of treatment allocation Reported adequacy of blinding Free of selective reporting Intention-to- treat analysis 
AMT      
    Carette et al. [41Unclear Unclear Unclear Yes No 
    Carette et al. [42Unclear Unclear Unclear Yes Noa 
    Carette et al. [43Yes Unclear Unclear Yes No 
    Ginsberg (1995) Unclear Unclear Unclear Yes Yes 
    Goldenberg et al. [45Unclear Unclear Unclear Yes No 
    Goldenberg et al. [46Yes Yes Yes Yes No 
    Hannonen et al. [47Yes Yes Unclear Yes Yes 
    Heymann et al. [36Yes Yes Unclear No No 
    Kempenaers et al. [49Unclear Unclear Unclear Yes No 
    Scudds et al. [50Unclear Unclear Yes Yes No 
DLX      
    Arnold et al. [51Yes Unclear; on request: yes Unclear; on request: yes Yes Yes 
    Arnold et al. [52Unclear; on request: yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
    Chappell (2008) Yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
    Russel et al. [53Yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
MLN      
    Arnold et al. [16Yes Yes Yes Yes Yes 
    Branco et al. [3Unclear; on request: yes Unclear; on request: yes Unclear; on request: yes Yes Yes 
    Clauw et al. [12Yes Yes Unclear; on request: yes No; complete data provided on request Yes 
    Mease et al. [19Unclear; on request: yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
    Vitton et al. [55Yes Yes Unclear; on request: yes No; complete data provided on request Yes 
References Reported adequacy of randomization Reported adequacy of concealment of treatment allocation Reported adequacy of blinding Free of selective reporting Intention-to- treat analysis 
AMT      
    Carette et al. [41Unclear Unclear Unclear Yes No 
    Carette et al. [42Unclear Unclear Unclear Yes Noa 
    Carette et al. [43Yes Unclear Unclear Yes No 
    Ginsberg (1995) Unclear Unclear Unclear Yes Yes 
    Goldenberg et al. [45Unclear Unclear Unclear Yes No 
    Goldenberg et al. [46Yes Yes Yes Yes No 
    Hannonen et al. [47Yes Yes Unclear Yes Yes 
    Heymann et al. [36Yes Yes Unclear No No 
    Kempenaers et al. [49Unclear Unclear Unclear Yes No 
    Scudds et al. [50Unclear Unclear Yes Yes No 
DLX      
    Arnold et al. [51Yes Unclear; on request: yes Unclear; on request: yes Yes Yes 
    Arnold et al. [52Unclear; on request: yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
    Chappell (2008) Yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
    Russel et al. [53Yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
MLN      
    Arnold et al. [16Yes Yes Yes Yes Yes 
    Branco et al. [3Unclear; on request: yes Unclear; on request: yes Unclear; on request: yes Yes Yes 
    Clauw et al. [12Yes Yes Unclear; on request: yes No; complete data provided on request Yes 
    Mease et al. [19Unclear; on request: yes Unclear; on request: yes Unclear; on request: yes No; complete data provided on request Yes 
    Vitton et al. [55Yes Yes Unclear; on request: yes No; complete data provided on request Yes 

aPretends that only data of ITT were reported, but Table 4 includes complete analysis. Adequacy of randomization: a random (unpredictable) assignment sequence. Examples of adequate methods are computer-generated random number table and use of sealed opaque envelopes. Methods of allocation using date of birth, date of admission, hospital numbers or alternation were regarded as appropriate. Adequacy of concealment of treatment allocation: assignment generated by an independent person not responsible for determining the eligibility of the patients. This person has no information about the persons included in the trial and has no influence on the assignment sequence or on the decision about eligibility of the patient (e.g. hospital pharmacy, central independent unit). Adequacy of blinding: patient and care provider and outcome assessor were blinded to the intervention. Complete reporting of outcomes: all outcomes assessed in the study were reported. Intention-to-treat analysis: all randomized patients are reported/analysed in the group they were allocated to by randomization (minus missing values).

Table 3

Effect sizes of AMT, DLX and MLN on outcome variables

Substance Number of studies Number of patients on active treatment SMD (95% CI) Test for overall effect P-value Heterogeneity I2 (%); τ2 
Pain      
    AMT 290 −0.43 (−0.75, −0.11) 0.008 62; 0.14 
    DLX 865 −0.33 (−0.43, −0.3) <0.0001 15; 0 
    MLN 2522 −0.19 (−0.25, −0.14) <0.0001 
Fatigue      
    AMT 176 −0.44 (−0.71, −0.16) 0.002 28; 0.04 
    DLX 629 −0.10 (−0.21, 0.01) 0.06 0; 0 
    MLN 2522 −0.13 (−0.19, −0.07) <0.0001 0; 0 
Sleep      
    AMT 194 −0.56 (−0.78, −0.34) <0.0001 0; 0 
    DLX 230 −0.31 (−0.50, −0.13) 0.0007 0; 0 
    MLN 1985 −0.03 (−0.09, 0.04) 0.43 0; 0 
HRQOL      
    AMT 168 −0.54 (−1.24, 0.17) 0.13 87; 0.44 
    DLX 865 −0.25 (−0.42, −0.08) 0.005 69; 0.04 
    MLN 2522 −0.18 (−0.23, −0.12) <0.0001 0; 0 
30% pain reduction   RR (95% CI)   
    AMT 290 1.60 (1.15, 2.24) 0.006 43; 0.10 
    DLX 865 1.29 (1.12, 1.49) 0.0005 18; 0.01 
    MLN 2522 1.30 (1.17, 1.44) <0.0001 18; 0 
Total drop-out   RR (95% CI)   
    AMT 290 0.73 (0.52, 1.02) 0.07 0; 0 
    DLX 865 0.95 (0.83, 1.08) 0.44 6; 0 
    MLN 2522 0.99 (0.75, 1.31) 0.93 85; 0.08 
Substance Number of studies Number of patients on active treatment SMD (95% CI) Test for overall effect P-value Heterogeneity I2 (%); τ2 
Pain      
    AMT 290 −0.43 (−0.75, −0.11) 0.008 62; 0.14 
    DLX 865 −0.33 (−0.43, −0.3) <0.0001 15; 0 
    MLN 2522 −0.19 (−0.25, −0.14) <0.0001 
Fatigue      
    AMT 176 −0.44 (−0.71, −0.16) 0.002 28; 0.04 
    DLX 629 −0.10 (−0.21, 0.01) 0.06 0; 0 
    MLN 2522 −0.13 (−0.19, −0.07) <0.0001 0; 0 
Sleep      
    AMT 194 −0.56 (−0.78, −0.34) <0.0001 0; 0 
    DLX 230 −0.31 (−0.50, −0.13) 0.0007 0; 0 
    MLN 1985 −0.03 (−0.09, 0.04) 0.43 0; 0 
HRQOL      
    AMT 168 −0.54 (−1.24, 0.17) 0.13 87; 0.44 
    DLX 865 −0.25 (−0.42, −0.08) 0.005 69; 0.04 
    MLN 2522 −0.18 (−0.23, −0.12) <0.0001 0; 0 
30% pain reduction   RR (95% CI)   
    AMT 290 1.60 (1.15, 2.24) 0.006 43; 0.10 
    DLX 865 1.29 (1.12, 1.49) 0.0005 18; 0.01 
    MLN 2522 1.30 (1.17, 1.44) <0.0001 18; 0 
Total drop-out   RR (95% CI)   
    AMT 290 0.73 (0.52, 1.02) 0.07 0; 0 
    DLX 865 0.95 (0.83, 1.08) 0.44 6; 0 
    MLN 2522 0.99 (0.75, 1.31) 0.93 85; 0.08 

Efficacy

AMT had a significant effect on pain, fatigue and sleep, but not on HRQOL. Based on Cohen’s categories, the effects of AMT on pain and fatigue were small and on sleep moderate. DLX had a significant effect on pain, sleep and HRQOL, but not on fatigue. Based on Cohen’s categories, the significant effects of DLX were small. MLN had significant effects on pain, fatigue and HRQOL, but not on sleep. Based on Cohen’s categories, the significant effects of MLN were not substantial.

The NNT for a 30% pain reduction for AMT was 3.54 (95% CI 2.74, 5.01), for DLX 8.21 (95% CI 5.91, 13.26) and for MLN 10.96 (95% CI 8.27, 16.26). The RR of drop-out was not significant for all three drugs (Table 3).

Comparative efficacy

In adjusted indirect analyses, AMT was superior to DLX and MLN in reduction of pain, sleep disturbances, fatigue and limitations of HRQOL. DLX was superior to MLN in reducing pain, sleep disturbances and limitations of HRQOL. MLN was superior to DLX in reducing fatigue. AMT was superior to DLX and MLN in 30% pain reduction. There were no significant differences in the acceptability (drop-out rates) between the three drugs (Table 4).

Table 4

Subgroup analyses

Substance Number of studies Total number of patientsa Effect size SMD (95% CI) on pain Test for overall effect P-value Hetero-geneity I2 (%); τ2 
Europe      
    AMT 117 −0.69 (−1.64, 0.26) 0.15 80; 0.56 
    DLX 325 −0.20 (−0.41, 0.02) 0.08 – 
    MLN 876 −0.19 (−0.32, −0.06) 0.005 – 
America      
    AMT 351 −0.34 (−0.66, −0.03) 0.03 49; 0.07 
    DLX 1478 −0.35 (−0.47, −0.24) <0.0001 18; 0 
    MLN 3858 −0.19 (−0.26, −0.13) <0.0001 0; 0 
Industrial sponsoring of the study 
    AMT 273 −0.49 (−1.09, 0.10) 0.10 79; 0.35 
No industrial sponsoring of the study 
    AMT 195 −0.40 (−0.69, −0.12) 0.006 0; 0 
Cross-over design      
    AMT 148 −0.48 (−0.81, −0.15) 0.005 3; 0 
Parallel design      
    AMT 320 −0.45 (−0.93, 0.04) 0.07 74; 0.26 
Substance Number of studies Total number of patientsa Effect size SMD (95% CI) on pain Test for overall effect P-value Hetero-geneity I2 (%); τ2 
Europe      
    AMT 117 −0.69 (−1.64, 0.26) 0.15 80; 0.56 
    DLX 325 −0.20 (−0.41, 0.02) 0.08 – 
    MLN 876 −0.19 (−0.32, −0.06) 0.005 – 
America      
    AMT 351 −0.34 (−0.66, −0.03) 0.03 49; 0.07 
    DLX 1478 −0.35 (−0.47, −0.24) <0.0001 18; 0 
    MLN 3858 −0.19 (−0.26, −0.13) <0.0001 0; 0 
Industrial sponsoring of the study 
    AMT 273 −0.49 (−1.09, 0.10) 0.10 79; 0.35 
No industrial sponsoring of the study 
    AMT 195 −0.40 (−0.69, −0.12) 0.006 0; 0 
Cross-over design      
    AMT 148 −0.48 (−0.81, −0.15) 0.005 3; 0 
Parallel design      
    AMT 320 −0.45 (−0.93, 0.04) 0.07 74; 0.26 

aThe total number of patients in the comparisons is higher than the total number of patients in the studies because different dosages of DLX and MLN were compared with the same placebo group in some trials with DLX and MLN.

Subgroup and sensitivity analyses

There were no significant effects on pain for AMT and DLX studies conducted in Europe and in AMT studies with industrial support and parallel design (Table 5). The effect of AMT on pain was not significant in two studies with extracted values from tables [−0.43 (95% CI −0.96, 0.11); I2 = 35%; P = 0.12], but was in the seven studies with reported numerical data [−0.47 (95% CI −0.89, −0.06); I2 = 71%; P = 0.03]. The effect of AMT on pain was not significant in two studies with ITT analysis [−0.91 (95% CI −2.26, 0.43); I2 = 89%; P = 0.28], but was in the seven studies without ITT analysis [−0.32 (95% CI −0.61, −0.04); I2 = 39%; P = 0.03]. The effect of the two AMT studies with ITT on fatigue and sleep was significant (data not shown).

Table 5

Adjusted relative efficacy and acceptability (indirect estimates from placebo-controlled trials with parallel design) of AMT, DLX and MLN in FMS

 Adjusted indirect estimates (ratios)
 
Comparison (effect estimate) AMT vs DLX (95% CI) AMT vs MLN (95% CI) DLX vs MLN (95% CI) 
Pain (SMD) 1.35 (1.10, 1.64) 2.27 (1.86, 2.83) 1.68 (1.52, 1.85) 
Fatigue (SMD) 7.31 (5.87, 9.11) 3.39 (2.66, 4.39) 0.46 (0.41, 0.51) 
Sleep (SMD) 1.80 (1.43, 2.27) 18.54 (15.80, 21.76) 10.28 (8.50, 11.82) 
Quality of life (SMD) 1.93 (1.50, 2.48) 5.42 (4.01, 7.31) 2.80 (2.27, 3.49) 
30% pain reduction (RR) 1.24 (1.02, 1.52) 1.09 (1.01, 1.45) 0.99 (−0.12, 1.10) 
Total drop-out rate (RR) 0.77 (0.48, 1.25) 0.74 (0.47, 1.17) 0.90 (0.76, 1.22) 
 Adjusted indirect estimates (ratios)
 
Comparison (effect estimate) AMT vs DLX (95% CI) AMT vs MLN (95% CI) DLX vs MLN (95% CI) 
Pain (SMD) 1.35 (1.10, 1.64) 2.27 (1.86, 2.83) 1.68 (1.52, 1.85) 
Fatigue (SMD) 7.31 (5.87, 9.11) 3.39 (2.66, 4.39) 0.46 (0.41, 0.51) 
Sleep (SMD) 1.80 (1.43, 2.27) 18.54 (15.80, 21.76) 10.28 (8.50, 11.82) 
Quality of life (SMD) 1.93 (1.50, 2.48) 5.42 (4.01, 7.31) 2.80 (2.27, 3.49) 
30% pain reduction (RR) 1.24 (1.02, 1.52) 1.09 (1.01, 1.45) 0.99 (−0.12, 1.10) 
Total drop-out rate (RR) 0.77 (0.48, 1.25) 0.74 (0.47, 1.17) 0.90 (0.76, 1.22) 

Significant comparisons (95% CIs with lower and upper limit <1 or >1) are marked as bold.

Meta-regression

Post hoc we decided to meta-regress outcomes with methodological quality only for AMT trials because the methodological quality of the DLX and MLN trials was, after requests, equivalent. There was no significant correlation of the cumulative Cochrane risk of bias tool and the SMD of AMT on pain (β = 0.06, P = 0.6).

Publication bias

Visual inspection of the funnel plots of the AMT studies was not suggestive of a publication bias. The Egger’s intercept was positive (3.36), but not significant (two-tailed P = 0.10).

Discussion

Summary of evidence

In adjusted indirect comparisons, AMT was superior to DLX and MLN in reducing FMS symptoms. The evidence of superior efficacy of AMT over DLX and MLN is weakened by the methodological problems of some AMT trials. A significant effect on pain was only found in studies with cross-over design, no ITT analysis and non-European patients. However, within a conservative approach of imputing reponse rates for 30% pain reduction of AMT trials, the superior efficacy of AMT over DLX and MLN in pain reduction was confirmed. There were no significant differences between the three drugs in terms of acceptability.

Agreement with other systematic reviews

Our results are in line with a recent systematic qualitative review on AMT in FMS, which concluded that AMT demonstrated a therapeutic response compared with placebo in the domains of pain, sleep and fatigue [8]. In contrast to Nishishinya et al. [8], we performed a quantitative synthesis of the AMT trials because the heterogeneity of most outcomes of AMT was not as high as suggested by the authors. We addressed the methodological pitfalls of some AMT trials, which had been worked out by the authors by subgroup and sensitivity analyses and meta-regression. Adding two large RCTs with MLN in indirect comparison with DLX did not change the results of our previous indirect comparison of the two drugs [14].

Agreement with excluded studies

Three of four excluded studies confirmed the efficacy of AMT in reducing some FMS symptoms. In 7 of 23 N-of-1 trials, AMT 10–50 mg/day over 3–12 weeks was superior to placebo in reducing a symptom score of pain, fatigue, sleep, morning stiffness, headache and bowel problems [35]. In an 8-week trial, AMT 10 mg was superior to laser placebo in reducing morning stiffness [37]. In a 12-week trial, AMT 25 mg was superior to placebo in reducing fatigue, but not pain or sleep disturbance [38]. In a 4-week study, AMT 10–50 mg/day combined with guided imagery was not superior to guided imagery combined with pharmacological placebo in reducing pain [33].

Limitations

The major methodological limitation of the analysis is that the similarity assumption for indirect comparisons was only partially fulfilled. Some patient characteristics differed significantly, but in our opinion that was not clinically relevant between AMT studies and DLX/MLN trials. The methodological quality of the AMT studies was lower than that of the DLX and MLN studies. Therefore, the results of the adjusted indirect comparisons should be interpreted with caution.

We did not succeed in receiving missing outcome data of one AMT trial and the methodological details of all AMT trials. The missing outcomes of pain, sleep and fatigue in the study of Heymann et al. [36] would probably not change the results of our quantitative analysis because AMT had a strong effect on the FM impact questionnaire (FIQ) total score in this study. The FIQ total score comprises the missing outcomes.

Conclusions

Clinical practice

The recommendation of the German FMS guideline that AMT is the first pharmacological treatment option is substantiated by the indirect comparison of AMT with DLX and MLN, and the lack of approval of DLX and MLN for FMS in Europe. In contrast to DLX and MLN, AMT has moderate effects on sleep and small effects on fatigue, thus covering three key symptoms of FMS. However, the dosages of AMT in the FMS trials (10–50 mg/day) did not reduce depressed mood [22]. DLX is approved in most European countries for major depression. Its efficacy in pain reduction had been shown for FMS patients with and without major depression. Therefore, DLX can be regarded to be the first option for FMS patients with major depression. The different profiles of side effects of the three drugs [8, 14] and their relevance for the individual might also influence the choice of drug. It should be noted that a postive effect on FMS symptoms of AMT has only been confirmed for 3 months and of DLX and MLN for up to 6 months in placebo-controlled trials. Therefore, trial of omission can be discussed with responders after these time intervals.

Research

More stringent RCTs with longer follow-up periods are required to determine the long-term efficacy and safety of AMT [8]. We recommend that the EMEA requires for future approval of new drugs that they provide an additional benefit compared with standard therapies such as AMT. Meanwhile, head-to-head comparisons of the three drugs should be conducted with patient samples representative of clinical practice. These samples should include patients with comorbid anxiety and affective disorders, which are frequently associated with FMS [57]. These studies should be conducted in America as well as in Europe and Asia. Without financial support the required multinational studies will not be possible. Funding by public sources instead of pharmaceutical companies might be preferable.

graphic

Disclosure statement: C.S. has served on scientific advisory boards for Pfizer Inc., Eli Lilly and Company, Grünenthal Inc., Astellas Pharma Inc. and Baxter International Inc.; she has received speaker honoraria from Pfizer Inc., Boehringer Ingelheim and Genzyme Corporation. W.H. has received honoraria for educational lectures from Eli Lilly, Janssen-Cilag and Mundipharma; consulting fees from Eli Lilly and Pfizer, and a congress travel grant from Eli Lilly. F.P. has received honoraria for educational lectures from Pierre Fabre, Grünenthal and Janssen-Cilag; consulting fees from Pfizer, Grünenthal and Eli Lilly; a research grant from Pierre Fabre and a congress travel grant from Eli Lilly and Pfizer. The other author has declared no conflicts of interest.

Supplementary data

Supplementary data are available at Rheumatology Online.

References

1
Wolfe
F
Clauw
DJ
Fitzcharles
MA
, et al.  . 
The American College of Rheumatology preliminary diagnostic criteria for fibromyalgia and measurement of symptom severity
Arthritis Care Res
 , 
2010
, vol. 
62
 (pg. 
600
-
10
)
2
Häuser
W
Zimmer
C
Felde
E
Köllner
V
What are the key symptoms of fibromyalgia syndrome? Results of a survey of the German Fibromyalgia Association
Schmerz
 , 
2008
, vol. 
22
 (pg. 
176
-
83
)
3
Branco
JC
Bannwarth
B
Failde
I
, et al.  . 
Prevalence of fibromyalgia: a survey in five European Countries
Semin Arthritis Rheum
 , 
2010
, vol. 
39
 (pg. 
448
-
55
)
4
Häuser
W
Schmutzer
G
Glaesmer
H
Brähler
E
[Prevalence and predictors of pain in several body regions. Results of a representative German population survey] (in German)
Schmerz
 , 
2009
, vol. 
23
 (pg. 
461
-
70
)
5
Berger
A
Dukes
E
Martin
S
Edelsberg
J
Oster
G
Characteristics and healthcare costs of patients with fibromyalgia syndrome
Int J Clin Pract
 , 
2007
, vol. 
61
 (pg. 
1498
-
508
)
6
Carville
SF
Arendt-Nielsen
S
Bliddal
H
, et al.  . 
EULAR evidence based recommendations for the management of fibromyalgia syndrome
Ann Rheum Dis
 , 
2008
, vol. 
67
 (pg. 
536
-
41
)
7
Häuser
W
Eich
W
Herrmann
M
Nutzinger
DO
Schiltenwolf
M
Henningsen
P
Fibromyalgia syndrome: classification, diagnosis, and treatment
Dtsch Arztebl Int
 , 
2009
, vol. 
106
 (pg. 
383
-
91
)
8
Nishishinya
B
Urrútia
G
Walitt
B
, et al.  . 
Amitriptyline in the treatment of fibromyalgia: a systematic review of its efficacy
Rheumatology
 , 
2008
, vol. 
47
 (pg. 
1741
-
6
)
9
European Medicines Agency
 
10
European Medicines Agency
 
11
Briley
M
Drugs to treat fibromyalgia – the transatlantic difference
Curr Opin Investig Drugs
 , 
2010
, vol. 
11
 (pg. 
16
-
18
)
12
Clauw
DJ
Mease
P
Palmer
RH
Gendreau
RM
Wang
Y
Milnacipran for the treatment of fibromyalgia in adults: a 15-week, multicenter, randomized, double- blind, placebo-controlled, multiple-dose clinical trial
Clin Ther
 , 
2008
, vol. 
30
 (pg. 
1988
-
2004
)
13
Garber
AM
Tunis
SR
Does comparative-effectiveness research threaten personalized medicine?
N Engl J Med
 , 
2009
, vol. 
360
 (pg. 
1925
-
7
)
14
Häuser
W
Petzke
F
Sommer
C
Comparative efficacy and harms of duloxetine, milnacipran, and pregabalin in fibromyalgia syndrome
J Pain
 , 
2010
, vol. 
11
 (pg. 
505
-
21
)
15
Branco
JC
Zachrisson
O
Perrot
S
Mainguy
Y
Multinational Coordinator Study Group
A European multicenter randomized double-blind placebo-controlled monotherapy clinical trial of milnacipran in treatment of fibromyalgia
J Rheumatol
 , 
2010
, vol. 
37
 (pg. 
851
-
9
)
16
Arnold
LM
Gendreau
RM
Palmer
RH
Gendreau
JF
Wang
Y
Efficacy and safety of milnacipran 100 mg/day in patients with fibromyalgia: results of a randomized, double-blind, placebo-controlled trial
Arthritis Rheum
 , 
2010
, vol. 
62
 (pg. 
2745
-
56
)
17
Moher
D
Liberati
A
Teztlaff
J
Altman
G
PRISMA Group
Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement
Ann Int Med
 , 
2009
, vol. 
51
 (pg. 
1
-
7
)
18
Higgins
JPT
Green
S
Cochrane Handbook for systematic reviews of intervention
 
19
Mease
P
Arnold
LM
Choy
EH
, et al.  . 
Fibromyalgia syndrome module at OMERACT 9: domain construct
J Rheumatol
 , 
2009
, vol. 
36
 (pg. 
2318
-
29
)
20
Cipriani
A
Furukawa
TA
Salanti
G
, et al.  . 
Comparative efficacy and acceptability of 12 new-generation antidepressants: a multiple-treatments meta-analysis
Lancet
 , 
2009
, vol. 
373
 (pg. 
746
-
58
)
21
Häuser
W
Bernardy
K
Üçeyler
N
Sommer
C
Treatment of fibromyalgia syndrome with antidepressants: a meta-analysis
JAMA
 , 
2009
, vol. 
30
 (pg. 
198
-
209
)
22
National Institutes for Health
 
http://clinicaltrials.gov/ct2/results?term=duloxetine (or amitriptyline or milnacipran)+and+fibromyalgia (31 May 2010, date last accessed)
23
Pharmaceutical Research and Manufacturers of America (PhRMA)
 
http://www.clinicalstudyresults.org/home/. (31 March 2010, date last accessed)
24
The Nordic Cochrane Centre. Review Manager (RevMan) [Computer program]
 
Version 5.0 for Windows. Copenhagen: Cochrane Collaboration, 2009
25
Biostat
Comprehensive metaanalysis software
 
Version 2.0. Englewood. www.metaanalysis.com
26
Laird
NM
Mostellier
F
Some statistical methods for combining experimental results
Int J Technol Assess Health Care
 , 
1990
, vol. 
6
 (pg. 
5
-
30
)
27
Cohen
J
Statistical power analysis for the behavoral sciences
 , 
1988
Hillsdale
Lawrence Erlbaum Associates
28
Furukawa
TA
Cipriani
A
Barbui
C
Brambilla
P
Watanabe
N
Imputing response rates from means and standard deviations in meta-analyses
Int Clin Psychopharmacol
 , 
2005
, vol. 
20
 (pg. 
49
-
52
)
29
Bucher
HC
Guyatt
GH
Griffith
LE
Walter
SD
The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials
J Clin Epidemiol
 , 
1997
, vol. 
50
 (pg. 
683
-
91
)
30
Bland
JM
Altman
DG
Statistics notes
The odds ratio. Br Med J
 , 
2000
, vol. 
320
 pg. 
1468
 
31
Song
F
Loke
YK
Walsh
T
Glenny
AM
Eastwood
AJ
Altman
DG
Methodological problems in the use of indirect comparisons for evaluating healthcare interventions: survey of published systematic reviews
Br Med J
 , 
2009
, vol. 
338
 pg. 
b1147
 
32
Egger
M
Smith
GD
Schneider
M
Minder
C
Bias in meta-analysis detected by a simple, graphical test
Br Med J
 , 
1997
, vol. 
315
 (pg. 
629
-
34
)
33
Fors
EA
Sexton
H
Gotestam
KG
The effect of guided imagery and amitriptyline on daily fibromyalgia pain: a prospective, randomized, controlled trial
J Psychiatr Res
 , 
2002
, vol. 
36
 (pg. 
179
-
87
)
34
Isomeri
R
Mikkelsson
M
Latikka
P
Kammonen
K
Effects of amitriptyline and cardiovascular fitness training on pain in patients with primary fibromyalgia
J Muscul Pain
 , 
1993
, vol. 
1
 (pg. 
253
-
60
)
35
Jaeschke
R
Adachi
J
Guyatt
G
Keller
J
Wong
B
Clinical usefulness of amitriptyline in fibromyalgia: the results of 23 N-of-1 randomized controlled trials
J Rheumatol
 , 
1991
, vol. 
18
 (pg. 
447
-
51
)
36
Heymann
RE
Quaresma
M
Helfenstein
M
Feldman
D
A double-blinded, randomized, controlled study between amitriptyline, nortriptyline and placebo in patients with fibromyalgia: analysis of the outcome measures
Rev Bras Reumatol
 , 
1998
, vol. 
38
 (pg. 
119
-
27
)
37
Gur
A
Karakoc
M
Nas
K
Cevik
R
Sarac
J
Ataoglu
S
Effects of low power laser and low dose amitriptyline therapy on clinical symptoms and quality of life in fibromyalgia: a single-blind, placebo-controlled trial
Rheumatol Int
 , 
2002
, vol. 
22
 (pg. 
188
-
93
)
38
Hench PKm Cohen
R
Mitler
MM
Fibromyalgia: effects of tetrazepam, amitriptyline and placebo on pain and sleep. (abstract)
Arthritis Rheum
 , 
1989
, vol. 
32
 
Suppl.
pg. 
S47
 
39
Eli Lilly
Flexible dosed duloxetine vs placebo in the treatment of fibromyalgia
 
NCT00673452. www.clinicaltrials.gov (30 May 2010, date last accessed)
40
Forrest Laboratories
A multicenter, randomized, double-blind, placebo-controlled discontinuation study of the durability of effect of milnacipran for the treatment of fibromyalgia in patients receiving long-term milnacipran treatment
 
NCT01014585. www.clinicaltrials.gov (30 May 2010, date last accessed)
41
Carette
S
McCain
GA
Bell
DA
Fam
AG
Evaluation of amitriptyline in primary fibrositis. A double-blind, placebo-controlled study
Arthritis Rheum
 , 
1986
, vol. 
29
 (pg. 
655
-
9
)
42
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
, vol. 
37
 (pg. 
32
-
40
)
43
Carette
S
Oakson
G
Guimont
C
Steriade
M
Sleep electroencephalography and the clinical response to amitriptyline in patients with fibromyalgia
Arthritis Rheum
 , 
1995
, vol. 
38
 (pg. 
1211
-
7
)
44
Ginsberg
F
Mancaux
A
Joos
E
, et al.  . 
A randomized placebo-controlled trial of sustained-release amitriptyline in primary fibromyalgia
J Musculoskel Pain
 , 
1996
, vol. 
4
 (pg. 
37
-
47
)
45
Goldenberg
DL
Felson
DT
Dinerman
H
A randomized, controlled trial of amitriptyline and naproxen in the treatment of patients with fibromyalgia
Arthritis Rheum
 , 
1986
, vol. 
29
 (pg. 
1371
-
7
)
46
Goldenberg
D
Mayskiy
M
Mossey
C
Ruthazer
R
Schmid
C
A randomized, double-blind crossover trial of fluoxetine and amitriptyline in the treatment of fibromyalgia
Arthritis Rheum
 , 
1996
, vol. 
39
 (pg. 
1852
-
9
)
47
Hannonen
P
Malminiemi
K
Yli-Kerttula
U
Isomeri
R
Roponen
P
A randomized, double-blind, placebo-controlled study of moclobemide and amitriptyline in the treatment of fibromyalgia in females without psychiatric disorder
Br J Rheumtol
 , 
1998
, vol. 
37
 (pg. 
1279
-
86
)
48
Heymann
RE
Helfenstein
M
Feldman
D
A double-blind, randomized, controlled study of amitriptyline, nortriptyline and placebo in patients with fibromyalgia. An analysis of outcome measures
Clin Exp Rheumatol
 , 
2001
, vol. 
19
 (pg. 
697
-
702
)
49
Kempenaers
C
Simenon
G
Vander Elst
M
, et al.  . 
Effect of an antidiencephalon immune serum on pain and sleep in primary fibromyalgia
Neuropsychobiology
 , 
1994
, vol. 
30
 (pg. 
66
-
72
)
50
Scudds
RA
McCain
GA
Rollman
GB
Harth
M
Improvements in pain responsiveness in patients with fibrositis after successful treatment with amitriptyline
J Rheumatol
 , 
1989
, vol. 
19
 
Suppl. 1
(pg. 
98
-
103
)
51
Arnold
LM
Lu
Y
Crofford
LJ
, et al.  . 
A double-blind, multicenter trial comparing duloxetine with placebo in the treatment of fibromyalgia patients with or without major depressive disorder
Arthritis Rheum
 , 
2004
, vol. 
50
 (pg. 
2974
-
84
)
52
Arnold
LM
Rosen
A
Pritchett
YL
, et al.  . 
A randomized, double-blind, placebo-controlled trial of duloxetine in the treatment of women with fibromyalgia with or with out major depressive disorder
Pain
 , 
2005
, vol. 
119
 (pg. 
5
-
15
)
53
Russell
IJ
Mease
PJ
Smith
TR
, et al.  . 
Efficacy and safety of duloxetine for treatment of fibromyalgia in patients with or without major depressive disorder: results from a 6-month, randomized, double-blind, placebo-controlled, fixed-dose trial
Pain
 , 
2008
, vol. 
136
 (pg. 
432
-
44
)
54
Chappell
AS
Bradley
LA
Wiltse
C
Detke
MJ
D’Souza
DN
Spaeth
M
A six-month double-blind, placebo-controlled, randomized clinical trial of duloxetine for the treatment of fibromyalgia
Int J Gen Med
 , 
2009
, vol. 
1
 (pg. 
91
-
102
)
55
Vitton
O
Gendreau
M
Gendreau
J
Kranzler
J
Rao
SG
A double-blind placebo-controlled trial of milnacipran in the treatment of fibromyalgia
Hum Psychopharmacol
 , 
2004
, vol. 
19
 
Suppl. 1
(pg. 
S27
-
35
)
56
Mease
PJ
Clauw
DJ
Gendreau
RM
, et al.  . 
The efficacy and safety of milnacipran for treatment of fibromyalgia. A randomized, double-blind, placebo-controlled trial
J Rheumatol
 , 
2009
, vol. 
36
 (pg. 
398
-
409
)
57
Fietta
P
Fietta
P
Manganelli
P
Fibromyalgia and psychiatric disorders
Acta Biomed
 , 
2007
, vol. 
78
 (pg. 
88
-
95
)

Supplementary data

Comments

0 Comments