Introduction

Fibromyalgia is a disorder that is found in approximately 2% of the general population and is characterized by chronic widespread pain and tenderness [1]. Additional symptoms that often accompany fibromyalgia are tiredness or fatigue, depressed mood, sleep disturbance, and headache [2, 3]. We will use the term 'tiredness' throughout this report, due to our decision to use the fibromyalgia impact questionnaire (FIQ) outcome measure, unless we refer specifically to studies that have used measures of 'fatigue'.

Duloxetine, a potent reuptake inhibitor of serotonin and norepinephrine, was approved in the USA for the management of fibromyalgia in June 2008. Preclinical models of central sensitization suggest that duloxetine is efficacious in the treatment of patients with persistent/chronic pain. In rodents, duloxetine has demonstrated efficacy in the formalin and capsaicin models of persistent pain, as well as in the partial sciatic nerve ligation and L5/L6 spinal nerve ligation models of neuropathic pain [4, 5]. Duloxetine is efficacious in the treatment of the painful physical symptoms associated with depression and the pain associated with diabetic neuropathy in nondepressed patients [6, 7]. The results of four placebo-controlled clinical trials of duloxetine in the treatment of fibromyalgia have been previously reported [8–11].

Patients with fibromyalgia frequently report moderate to severe fatigue and the prevalence of fatigue in patients with fibromyalgia (76%) is almost double that of patients with osteoarthritis or rheumatoid arthritis (41%) [12, 13]. An evidence-based review of the literature on chronic pain found a highly consistent association between fatigue and pain in multiple studies and concluded that this may represent an etiological relation [14]. This conclusion was based in part on the results of several studies showing that fatigue develops following the onset of chronic pain and that severity of chronic pain is positively associated with the likelihood of subsequent reports of fatigue [15, 16]. However, few studies have assessed whether fatigue or tiredness levels are associated with variations in the efficacy of pharmacologic therapies for chronic pain. This is a particularly important issue in the management of patients with fibromyalgia who often report relatively high levels of fatigue/tiredness in addition to chronic widespread pain. The present study analyzed data from four placebo-controlled trials of duloxetine in patients with fibromyalgia to test the hypothesis that baseline ratings of tiredness would be negatively associated with the efficacy of duloxetine on measures of pain and functional ability in patients with fibromyalgia.

Materials and methods

This study comprised a pooled analysis of four double-blind, placebo-controlled, randomized, multicenter clinical trials of duloxetine in patients with fibromyalgia. These studies were conducted in accordance with the ethical principles that have their origin in the Declaration of Helsinki and that are consistent with good clinical practices and the applicable laws and regulations. Table 1 shows the dosage, administration, and duration of treatment for each of the studies. The details of these studies have been reported previously [8–11] and are summarized here. Male and female patients were entered into studies 1, 3, and 4, whereas study 2 included only female patients. Only three-month data are included in the analyses presented here.

Table 1 Summary of placebo-controlled trials of duloxetine for the treatment of fibromyalgia
Full size table

Included patients were at least 18 years of age and met the criteria for fibromyalgia as defined by the American College of Rheumatology. In study 1, patients were required to indicate a pain score of at least 4 on the FIQ pain item (score range of 0 to 10, with 10 indicating very severe pain) [17]. In studies 2 to 4, patients were required to indicate a pain score of at least 4 on the 24-hour average pain severity item of the brief pain inventory (BPI) (score range of 0 to 10, with 10 indicating pain as bad as you can imagine) [18]. The major exclusion criteria were serious or unstable medical or psychiatric illness, current primary psychiatric diagnosis other than major depressive disorder, a primary diagnosis of anxiety disorder in the past year, pain from traumatic injury, or rheumatologic illness [8–11].

All four studies assessed the effects of duloxetine and placebo on the BPI average pain score, the FIQ (a patient self-reported instrument that assesses the impact of fibromyalgia symptoms and functional impairment; the FIQ total score ranges from 0 (no impact) to 80 (maximum impact)), the patient global impressions of improvement (PGI-Improvement, a patient-rated global assessment of response to treatment with scores ranging from 1 (very much better) to 7 (very much worse)), and the medical outcomes study short form-36 (SF-36, a quality-of-life measure including eight health-status domains that are each scored 0 to 100, with higher scores indicating better health).

The raw scores for the FIQ tiredness item range from 0 to 10 [17]. In the analyses reported here, the FIQ tiredness item is used to substitute for a formal measure of fatigue (multidimensional fatigue inventory), because the latter was not used in all the four fibromyalgia studies. Based on the FIQ tiredness item score (0 to 10) at study entry, patients were stratified into three subgroups: mild (0 to 3), moderate (4 to 6), or severe (7 to 10) tiredness. Previous studies have also used an 11-point numeric rating scale for measuring fatigue, such as the fatigue severity scale [19] and the brief fatigue inventory where scores within the ranges of 0 to 3, 4 to 6, and 7 to 10 points indicate mild, moderate, and severe fatigue levels, respectively [20].

Treatment-emergent adverse events (TEAEs), collected at every visit by spontaneous reporting from the patient, within the three subgroups were also evaluated. A TEAE was any reported event that first occurred or worsened in severity during treatment, compared with the predefined baseline period.

Statistical methods

All randomized patients (except study 3 patients on the suboptimal dose of 20 mg once daily (n = 79) with at least one nonmissing postbaseline assessment were included in the efficacy analyses.

Changes from baseline to endpoint for continuous efficacy measures were analyzed using an analysis of covariance model, with missing values imputed using the last-observation-carried-forward approach. The model included baseline, treatment, study, tiredness subgroup, and treatment-by-tiredness subgroup interaction.

The baseline characteristics were analyzed using the Cochran-Mantel-Haenszel test controlling for study, or analysis of variance, which included treatment and study.

Categorical outcomes were analyzed using logistic regression (including terms for study, treatment, tiredness subgroup, and treatment-by-tiredness subgroup interaction) or Fisher's exact test. Treatment effects were tested at a two-sided significance level of 0.05. Interaction effects were tested at a two-sided significance level of 0.1. The term 'significant' indicates statistical significance throughout the manuscript.

Results

Demographics

Across all studies, 797 patients received duloxetine and 535 patients received placebo for approximately three months. There were no significant differences between duloxetine-treated and placebo-treated patients in any of the patients' demographic variables. The mean age of patients was 50.2 years (standard deviation = 11.0, range 19 to 83 years); the majority of patients were female (n = 1262, 94.7%) and Caucasian (n = 1169, 87.8%).

Distribution of tiredness subgroups

At baseline, the distribution of tiredness severity in the duloxetine group (based on the FIQ tiredness item score) was 3.64% (n = 29) mild (0 to 3), 16.71% (n = 133) moderate (4 to 6), and 79.65% (n = 634) severe (7 to 10); the distribution in the placebo group was 3.75% (n = 20) mild, 15.57% (n = 83) moderate, and 80.68% (n = 430) severe. Within the duloxetine and placebo groups, the mean baseline FIQ tiredness scores in the mild subgroup were 2.5 and 2.7, respectively, in the moderate subgroup was 5.3 (duloxetine and placebo), and in the severe subgroup were 8.6 and 8.7, respectively.

Efficacy measures by tiredness subgroups

Brief pain inventory

Within the severe subgroup, there was a significant difference between the duloxetine and placebo groups (P < 0.001) in the number of patients with 30% or greater and 50% or greater reductions in the BPI average pain response (Figure 1). For the BPI average pain score, within the severe subgroup there was a significant difference (P < 0.001) in the mean change from baseline to endpoint between the duloxetine and placebo groups (Table 2).

Table 2 Efficacy and functional outcome measures
Full size table
Figure 1
figure 1

Brief pain inventory response. Proportions of patients with (a) 30% and (b) 50% reduction in the average brief pain inventory score. ***P < 0.001 vs. placebo. With both of the outcome measures, between-groups (duloxetine-treated vs. placebo-treated) analysis showed that the group of patients receiving duloxetine exhibited significantly greater numbers of responding patients than among those receiving placebo, but this was only true among the patients in the more severe tiredness group. There was no significant treatment-by-tiredness subgroup interaction between the three subgroups (P > 0.1).

Full size image

Fibromyalgia impact questionnaire

For the FIQ pain score, within the severe subgroup there was a significant difference (P < 0.001) in the mean change (standard error (SE)) from baseline to endpoint between the duloxetine (-2.1 (0.1)) and placebo (-1.4 (0.1)) groups (Table 2). Within the severe subgroup, there were also significant differences in the mean change from baseline to endpoint between the duloxetine and placebo groups for the FIQ total score, physical impairment score, interference with work, stiffness, anxious, and depression scores. For the FIQ tiredness score, there was no significant difference in the mean change (SE) from baseline to endpoint between the duloxetine and placebo groups, within any of the tiredness subgroups (mild 2.4 (0.7) (duloxetine), 1.6 (0.8) (placebo), moderate -0.2 (0.2) (duloxetine), 0.1 (0.3) (placebo), and severe -1.8 (0.1) (duloxetine), -1.6 (0.1) (placebo)). Other FIQ measures are also reported in Table 2.

The number of patients with 30% improvement in tiredness at endpoint was 257 (33.9%) in the duloxetine group and 150 (29.0%) in the placebo group (P = 0.067). The number of patients with 50% improvement in tiredness at endpoint was 173 (22.8%) in the duloxetine group and 93 (18.0%) in the placebo group (P = 0.042). No significant differences were observed between the duloxetine and placebo groups in the number of patients with 30% worsening in tiredness at endpoint (duloxetine, n = 62 (8.2%), placebo, n = 41 (7.9%); P = 0.917) or in the number of patients with 50% worsening in tiredness at endpoint (duloxetine, n = 46 (6.1%), placebo, n = 27 (5.2%); P = 0.542).

Patient global impressions of improvement

We found differences between the duloxetine-treated and placebo-treated patients in the frequencies of PGI-Improvement scores of 1 or 2 (patients feeling very much better or much better) across two of the three tiredness subgroups. There were significantly greater numbers of patients in the duloxetine, compared with the control group, who reported global positive improvements in both the severe tiredness subgroup (duloxetine, n = 233 (38.26%); placebo, n = 87 (21.07%); P < 0.001), and moderate tiredness subgroup (duloxetine, n = 45 (35.71%); placebo, n = 18 (21.95%); P = 0.044). This difference was not found in the mild tiredness subgroup (duloxetine, n = 15 (53.57%), placebo, n = 7 (36.84%); P = 0.373).

Medical outcomes study short form-36

Within the severe subgroup, there was a significant difference (P < 0.05) in the mean change of the SF-36 scores from baseline to endpoint between the duloxetine and placebo groups on the health outcome measures of the SF-36 mental component summary; bodily pain, general health, mental health, physical functioning, role-emotional, role-physical, social functioning, and vitality (Table 2).

Treatment-by-tiredness subgroup interaction

The effect of duloxetine on the BPI average pain score was consistent across patients with mild, moderate, or severe tiredness (P > 0.1 for treatment-by-tiredness subgroup interaction). For the FIQ pain score, there was no significant therapy-by-tiredness subgroup interaction (P = 0.74). The effect of duloxetine on the FIQ total score, physical impairment score, interference with work score, pain score, tiredness score, rested score, stiffness score, and depression score was consistent across patients with mild, moderate, or severe tiredness (P > 0.1 for treatment-by-tiredness subgroup interaction). No significant therapy-by-tiredness subgroup interaction in the PGI-Improvement score was observed among the three tiredness subgroups (P = 0.908). The effects of duloxetine on the health outcome measures of the SF-36 mental component summary, physical component summary, bodily pain, general health, mental health, physical functioning, role-emotional, role-physical, social functioning, and vitality were consistent across patients with mild, moderate, or severe tiredness (P > 0.1 for treatment-by-tiredness subgroup interaction).

Treatment-emergent adverse events

A significant (P < 0.1) treatment-by-tiredness interaction in the percentage of patients with one TEAE or more was observed. Significant treatment-by-tiredness interactions were found for the percentage of patients with the following TEAEs: hypoaesthesia, arthralgia, cough, myalgia, bronchitis, influenza, palpitations, restless leg syndrome, and osteoarthritis (Table 3). Nausea was the most common TEAE, with an occurrence of 28% (duloxetine) and 15% (placebo) in the mild subgroup, 29% (duloxetine) and 15% (placebo) in the moderate subgroup, and 30% (duloxetine) and 11% (placebo) in the severe subgroup. However, there was no significant treatment-by-tiredness subgroup interaction (P > 0.1) in the percentage of patients with nausea. Comprehensive pooled safety analyses for these four studies have been performed and previously reported [21].

Table 3 Treatment-emergent adverse events (%) with significant treatment-by-tiredness subgroup interaction
Full size table

Discussion

In a previously conducted pooled analysis of these same four randomized, double-blind, placebo-controlled trials in patients with fibromyalgia, almost half (47.7%) of the patients treated with duloxetine had a 30% or greater reduction in BPI average pain, and 35.3% of patients had a 50% or greater reduction in pain [22]. Among placebo-treated patients, 32.1% had a 30% or greater reduction in BPI average pain, and 22.2% had a 50% or greater reduction [22]. A previous pooled analysis of these four clinical trials [22] demonstrated that a significantly greater proportion of patients treated with duloxetine versus placebo reported feeling very much better or much better. More than one third (38.4%) of duloxetine-treated patients reported feeling much better and less than one quarter (21.7%) of the placebo-treated patients reported feeling much better, based on PGI-Improvement scores.

Several double-blind, placebo-controlled studies have shown that duloxetine, compared with placebo, more frequently produces clinically-significant improvements in pain and functional ability [8–11]. A pooled analysis of these studies showed that duloxetine was statistically superior to placebo with respect to improvement on FIQ total scores and all SF-36 domains [22]. Furthermore, the pooled data indicated that those patients who achieve 50% or greater improvement in tiredness were more likely to have received duloxetine than placebo [22].

In the pooled analysis of four placebo-controlled trials in patients with fibromyalgia reported here, the efficacy of duloxetine did not vary as a function of baseline ratings of tiredness, as defined by the FIQ tiredness rating.

At baseline, the distribution of tiredness severity in the duloxetine and placebo groups were less than 4% of patients in the mild subgroup and more than 90% of total patients in the moderate to severe subgroups. This is not surprising, as a previous study has reported that up to 90% of patients with fibromyalgia report moderate to severe fatigue levels [12].

The rates of clinically significant (≥30% and ≥50%) improvement in BPI average pain were similar across the three tiredness subgroups, demonstrating that the efficacy of duloxetine in the treatment of pain in patients with fibromyalgia is not associated with baseline tiredness levels.

Analyses of the PGI-Improvement produced findings similar to those for BPI average pain ratings. As with the other efficacy measures reported here, there was no significant therapy-by-tiredness subgroup interaction on the measure of PGI-Improvement. Baseline tiredness levels were not significantly associated with the effects of duloxetine on the patients' reports of functional ability using the FIQ total score and the FIQ measures of physical impairment, interference with work, pain, stiffness, depression and the SF-36 mental and physical component summary scores, bodily pain, general health, mental health, physical functioning, role-emotional, role-physical, social functioning, and vitality.

Some limitations of this work should be considered. The results are based on four acute treatment trials of 12 weeks and the results may not generalize to a longer duration of treatment. In addition, sleep and fatigue/tiredness may have a correlation. Future studies should measure sleep quality and its effect on fatigue/tiredness in the fibromyalgia population. The results of these studies may not generalize to individuals with fibromyalgia symptoms diagnosed with Diagnostic and Statistical Manual of Mental Disorders-IV axis-one psychiatric disorder, rheumatic disease, or treatment-resistant patients, because such patients were excluded from the studies. Most of the patients included in these studies were middle-aged Caucasian women, which may limit generalization of these results to men and other individuals with fibromyalgia. The FIQ tiredness item is used to substitute the formal measure of fatigue, because the formal fatigue measure was not used in all four of the fibromyalgia studies. The FIQ tiredness item may not capture all aspects of fatigue as a formal fatigue scale would. It is worth noting that question number 20 of the Beck depression inventory refers to fatigue and tiredness interchangeably. Lastly, the distribution of patients in the tiredness subgroups was skewed due to the large number of patients in the severe subgroup (up to 81%) compared with the moderate (up to 17%) and the mild (<4%) subgroups; therefore, the information on patients in the mild and moderate subgroups may not be as robust as that in the severe subgroup.

Conclusions

This pooled analysis of four randomized placebo-controlled studies provides evidence that the efficacy of duloxetine on pain does not vary as a function of patients' baseline ratings of tiredness, thus implying that baseline tiredness may not need to be factored into models of efficacy regarding pain.