Effects of Bedtime Very Low Dose Cyclobenzaprine on Symptoms and Sleep Physiology in Patients with Fibromyalgia Syndrome: A Double-blind Randomized Placebo-controlled Study

Study overview

This was a randomized, double-blind, placebo-controlled, dose-escalating, parallel-design study in patients with FM and disrupted sleep conducted at 2 Canadian sites. Following a prestudy screening period of up to 21 days, patients were randomly assigned to receive either VLD CBP or placebo, in a 1:1 ratio, once daily for 8 weeks, to be taken between dinner and bedtime. The dosage was 1 mg for the first 7 days, after which, if clinically indicated and according to tolerability, the daily dose could be increased up to 4 mg. The dose could be reduced at the discretion of the investigator. Randomized patients were seen by the study physician(s) every 1 or 2 weeks for 8 weeks to evaluate therapeutic benefit, safety, and tolerability of study drug, with a final visit 4 to 10 days after the last on-treatment visit. For patients who discontinued, the efficacy and safety determinations designated for Day 56 were obtained on the last day the patient took study medication or as soon as possible thereafter (within 3 days for efficacy). The protocol and informed consent forms were reviewed and approved by the ethics review board for each study site. Patients were enrolled in 2 Canadian centers, in Toronto, Ontario, and Calgary, Alberta.

Entry criteria

Male and female patients were eligible for inclusion if they were 18 to 65 years of age (inclusive), had documented FM and sleep disturbance, met the American College of Rheumatology (ACR) 2001 criteria for FM²⁸, reported nonrestful sleep more nights than not for at least 3 months before the start of double-blind treatment, demonstrated the α-nonREM EEG sleep anomaly at screening, and provided signed informed consent. Patients were excluded who had inflammatory rheumatic disease or a primary psychiatric disorder, were using psychopharmacologic drugs or nonsteroidal antiinflammatory drugs (NSAID), had a history or presence of sleep apnea or sleep-related periodic involuntary limb movements, or had used CBP within 3 months prior to the study. While patients had a range of prior treatments for FM, the study was conducted prior to the approval of pregabalin, duloxetine, or milnacipran.

Prior medications or treatments that were exclusion criteria included use of (1) electroconvulsive therapy within 3 months (or residual cognitive effects); (2) fluoxetine within 5 weeks; (3) any investigational drug, antipsychotic drug, or steroid (including topical steroids) within 1 month; (4) any antidepressant (including monoamine oxidase inhibitors), anxiolytic, sedative-hypnotic, anticholinergic, muscle relaxant, lithium, or any other psychopharmacologic drug within 2 weeks of the start of double-blind treatment; (5) any nonpsychopharmacologic drug within 2 weeks of the start of the double-blind treatment period unless a stable dose of the drug had been maintained for at least 1 month (3 months for thyroid or hormonal medications) before the start of the double-blind treatment period; and (6) any capsaicin-containing cream, NSAID, or salicylate within 2 weeks of the start of double-blind treatment.

Treatments permitted during the trial included psychotherapy, if well established before the study; acetaminophen for headache; and nonpsychopharmacologic drugs if the patient had been receiving a stable dose of the drug for at least 1 month before the start of double-blind treatment (at least 3 months before double-blind treatment for thyroid medications and hormonal medications).

EEG and other variables assessed during sleep

Patients were evaluated in the sleep laboratory on 5 nights: at screening, baseline, and 3 on-treatment weeks, usually Weeks 2, 4, and 8. Overnight sleep EEG recording, commonly referred to as polysomnogram recordings (PSG), measured total sleep time, total awake time, sleep efficiency, percentage of time spent in each sleep stage, sleep stage changes, α-EEG sleep ratings, and movement arousals. Sleep records were scored blindly in 30-second epochs according to standard criteria. Scoring was centralized at 1 site (Dr. H. Moldofsky, Toronto). The PSG was performed and analyzed according to the Rechtschaffen and Kales criteria²⁹ before the current American Academy of Sleep Medicine rules were introduced.

To be included for study, patients had to exhibit the α-nonREM sleep anomaly at screening and baseline. The α-nonREM sleep activity was rated in Stages 2, 3, and 4. Total activity (all 3 stages), Stage 2 activity, and Stage 3–4 activity were each scored on a 1–5 scale, according to the percentage of sleep epochs in those sleep stages that contained an α-EEG rhythm³⁰. A patient was determined to have the α-nonREM sleep anomaly at screening if > 40% of sleep epochs contained α-EEG activity (a rating of 3 or higher) or at least 10 episodes of periodic k-α. K-α episodes were defined as periodic (every 20 to 40 s) k-complexes immediately followed by α-EEG activity 0.5 to 5 s in duration. K-α episodes were scored during Stage 2, 3, and 4 sleep^24,30.

Cyclic alternating pattern (CAP)

Subsequent to enrollment and completion, EEG sleep studies in patients with FM were reported that identified increases in the periodic sleep EEG arousal disorder known as CAP in nonREM sleep, a measure of sleep instability related to the symptoms of FM²⁶. Analysis of nocturnal sleep EEG CAP was determined by employing a computerized automatic detection analysis using the Embla PSG system^27,31 that provides the frequency of total CAP and subtypes of CAP_A1, CAP_A2, and CAP_A3^32,33. CAP was scored during Stage 2, 3, and 4 sleep. K-α episodes are a variety of CAP.

Efficacy variables

Pain, tenderness, fatigue, and mood are symptoms of FM consistent with the current Outcome Measures in Rheumatology Clinical Trials (OMERACT) guidelines^1,34.

Safety and tolerability assessments

Adverse events (AE) were recorded at all study visits or as they occurred. Blood chemistry tests (including thyroid-stimulating hormone at screening only), hematology tests, and urinalysis were performed at screening and on Days −1 and 56. A urine drug screen was performed at screening. Serum ß-human chorionic gonadotropin measurement was performed on women of childbearing potential at screening and on Days −1 and 56. Laboratory tests with values that became clinically significantly abnormal after drug administration were repeated until the values returned to normal or the etiology was identified and the sponsor notified. Laboratory determinations were performed by a central laboratory, MDS Pharma Services, Toronto.

Sitting pulse and blood pressure measurements were taken at all visits. Standard 12-lead electrocardiography (ECG) measures were performed at screening and on Days −1 and 56. A comprehensive physical examination, including oral temperature, was performed at screening and on Day 56. Height was measured at screening and weight at all visits.

AE were coded using COSTART. Treatment-emergent AE (TEAE) were summarized by number (%) of patients according to preferred term and system/organ class, and the treatment arms were compared in terms of incidence of AE reported by ≥ 10% of patients in either arm using Fisher’s exact test. AE and serious adverse events (SAE) were summarized by severity and relationship to study treatment.

Laboratory data were summarized, and mean change was presented for each numeric variable. Laboratory data were classified as low, normal, or high and out-of-range values were flagged. Within-group values for numeric variables were summarized. Scatterplots were generated for selected laboratory tests. The treatment arms were compared with respect to mean change using the 2-sample t test. Vital signs and weight were summarized for each study visit, and the groups were compared with respect to mean change using the 2-sample t test. Physical examination data were listed, and any new physical findings were summarized. Selected ECG variables were summarized at baseline and at the end of treatment, and treatments were compared using a 2-sample t test for actual value and change from baseline. The proportions of patients with a normal ECG were compared at baseline and end of treatment by Fisher’s exact test.

Assessment of treatment compliance

A record of all medication dispensed and returned during the study was maintained by the investigators. Treatment compliance was assessed by (1) counting each patient’s remaining capsules at each study visit, (2) subtracting this number from the amount originally in each bottle (yielding the number of capsules taken), and (3) dividing this number into the number of capsules that should have been taken, per protocol, during the period between study visits.

Statistical analyses

Analyses for musculoskeletal pain, fatigue, tenderness, HAD total score, HAD anxiety subscale, HAD depression subscale, sleep α-EEG rating, sleep time, percentage time spent in each sleep stage, percentage REM sleep, latency to Stage 2, REM latency, and sleep efficiency were performed using a paired t test for within-group change with a 2-sided p value. In addition, the VLD CBP and placebo groups were compared using change from baseline at Week 8 using a 2-sample t test with a 2-sided p value. Clinical global impression of change (CGIC) and patient global impression of change (PGIC) scores (each question, e.g., fatigue, sleep disturbance, musculoskeletal pain, and morning stiffness and aching) were analyzed with the Cochran-Mantel-Haenszel test on row means, using modified ridit scores. Since one of the goals of this Phase 2 study was to elicit signals of treatment effect over the many variables studied, no adjustment of p values was made for multiplicity.

Patient disposition

A total of 36 patients with FM and nonrestorative sleep were enrolled and treated (VLD CBP, n = 18, and placebo, n = 18). Thirty-seven patients satisfied the screening criteria, but prior to treatment 1 patient was discovered to have been randomized in error upon examination of her screening ECG results and was never dosed and was consequently omitted from the intention-to-treat (ITT) and safety populations (Figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

The progress of the study. VLD CBP: very low dose cyclobenzaprine; PSG: polysomnogram; LOCF: last observation carried forward.

Twenty-nine patients completed the study. Similar percentages of VLD CBP and placebo-treated subjects completed the study (89% of the VLD CBP group, 68% of the placebo group). One placebo-treated patient discontinued because of an AE. Three placebo patients were discontinued per protocol due to normalization of an α-nonREM sleep EEG abnormality between screening and baseline (α score < 3). One placebo patient and 2 VLD CBP patients were asked to return after 1 week instead of 2 weeks following Visits 6 and 7 and thus received only 6 weeks of therapy. Three patients considered completers had final efficacy evaluations at Week 7 and 5 patients had final efficacy evaluations after but not during Week 8. The average time in the study was 54 days.

Patient characteristics

Table 1 shows the anthropometric data for patients with FM. There were no meaningful differences between the 2 treatment groups in demographic or other baseline data. All patients were white and all patients but 1 were women, with a mean age of 43 years; 50% of patients in each group had had FM for more than 72 months. The 2 groups were also similar in terms of FM history. The time elapsed since diagnosis of FM was similar in the 2 groups, and all but 1 of the patients had received prior therapies for FM.

Dose titration of CBP

CBP was formulated as 1 mg capsules, and placebo capsules contained no CBP. Investigators were allowed to adjust dosage throughout the study as tolerability allowed. Daily dose ranged from 2 to 4 mg in the VLD CBP group at Week 8, with a group mean of 3.1 mg. Placebo capsules were titrated to a dose corresponding to 2.1 mg.

Tolerability

VLD CBP was well tolerated, with only 1 severe AE (headache), compared to 5 severe AE in the placebo group. One patient in the placebo group discontinued prematurely because of nonsymptomatic arrhythmias. There were no serious AE in this trial. No new or unexpected AE emerged during treatment, nor were there any clinically significant trends in laboratory test results, vital signs, ECG results, or physical examination findings. The most common TEAE, which occurred in > 10% of subjects and in > 5% of VLD CBP-treated subjects, are shown in Table 2. Overall, the types of TEAE observed were relatively consistent with those reported in the Flexeril product label.

VLD CBP effects on FM symptoms

Eighteen VLD CBP-treated and 18 placebo-treated subjects were evaluated using last observation carried forward (LOCF) and compared using a paired t test for within-group change with a 2-sided p value (Table 3). In addition, the VLD CBP and placebo groups were compared using change from baseline at Week 8 using a 2-sample t test with a 2-sided p value (Table 3).

Bedtime musculoskeletal pain and fatigue

The bedtime questionnaire recorded subjects’ assessments of pain and fatigue about 24 h after ingestion of the VLD CBP or placebo. Musculoskeletal pain was rated on a 7-point scale of 0–6 (from “no discomfort” to “worst possible discomfort”) at each of 10 specific body sites. Scores at these sites were then averaged to provide an assessment of overall musculoskeletal pain. For subjects who received VLD CBP, the mean musculoskeletal pain score changed from 2.3 at baseline to 1.7 at Week 8, which was a decrease (or improvement) of 0.6 (26.1%; p = 0.010). In contrast, placebo treatment did not result in statistically significant changes in musculoskeletal pain, which remained at an average score of 2.1 at baseline and at Week 8 (0.0%; p = 1.000). The mean change from baseline of the VLD CBP and placebo groups at Week 8 differed significantly and revealed that VLD CBP treatment significantly improved musculoskeletal pain (p = 0.044; Table 3).

Fatigue was rated on a 7-point scale of 1 to 7 (where 1 = “full of energy” and 7 = “totally physically exhausted”). For subjects who received VLD CBP, the mean fatigue score changed from 5.0 at baseline to 4.3 at Week 8, which was a decrease (or improvement) of 0.7 (14.0%; p = 0.039). In contrast, placebo treatment did not result in statistically significant changes in fatigue, from 4.7 at baseline to 4.9 at Week 8, which was an increase of 0.2 (4.3%; p = 0.647). However, the mean change from baseline of the VLD CBP and placebo groups at Week 8 did not differ significantly (Table 3).

Tenderness

In patients with FM, the application of pressure (4 kg/cm²) to specific body regions (9 bilateral “tender points,” 18 total “tender points”) elicits a more painful reaction than normal; this may be assessed by using a dolorimeter to determine the specific amount of pressure (0–6 kg/cm²) required to elicit pain at each of 6 tender points and totaling this score (0–36 kg/cm²). For subjects who received VLD CBP, the dolorimetry score changed from 14.3 kg/cm² at baseline to 18.6 kg/cm² at Week 8, which was an increase (or improvement) of 4.3 kg/cm² (30.1%; p = 0.006). In contrast, placebo treatment did not result in statistically significant changes in dolorimetry score, from 15.6 kg/cm² at baseline to 16.1 kg/cm² at Week 8, a change of 0.5 kg/cm² (3.2%; p = 0.614). The mean change from baseline of the VLD CBP and placebo groups at Week 8 revealed that VLD CBP treatment improved the dolorimetry score significantly (p = 0.029; Table 3).

Hospital Anxiety and Depression Scale (HAD)

The HAD is a widely used patient self-rated scale with 14 questions (7 anxiety and 7 depression questions) ranging from 0 to 42. For subjects who received VLD CBP, the HAD score changed from 13.7 at baseline to 10.4 at Week 8, which was a decrease (or improvement) of 3.3 (24.1%; p = 0.012). In contrast, placebo treatment did not result in statistically significant changes in HAD scale, 15.7 at baseline and 15.1 at Week 8 (3.8%; p = 0.459). Comparison of the change from baseline between the VLD CBP and placebo groups at Week 8 did not reveal a significant effect of VLD CBP treatment on the HAD scale (Table 3).

The HAD depression subscale score (range 0–21) was also analyzed. For subjects who received VLD CBP, the HAD depression subscale changed from 6.3 at baseline to 4.9 at Week 8, a decrease (or improvement) of 1.4 (22.2%; p = 0.017). In contrast, placebo treatment did not result in statistically significant changes in intragroup HAD depression subscale scores, from 6.7 at baseline to 7.4 at Week 8, an increase of 0.7 (10.4%; p = 0.319). Comparison of the change from baseline between the VLD CBP and placebo groups at Week 8 revealed that VLD CBP treatment was associated with a significant improvement in the HAD depression subscale score (p = 0.023; Table 3).

Clinician-rated and patient-rated improvement scales for fatigue

At each post-baseline visit, the CGIC and PGIC scores were rated separately by the clinician and by the patient, respectively. Fatigue was evaluated using a scale of 5 (“worse”) to 1 (“marked improvement”). VLD CBP treatment was associated with a statistically significant improvement in patient-rated and clinician-rated change in fatigue (p = 0.001 and p = 0.004, respectively). In contrast, placebo treatment did not result in statistically significant changes in any of these measures. Relative to placebo, VLD CBP treatment did not significantly improve patient-rated or clinician-rated improvement in fatigue (p = 0.133, p = 0.237, respectively; Table 3).

Effect of VLD CBP on macrostructure of sleep EEG

VLD CBP treatment at bedtime was studied for its effects on sleep EEG. PSG was performed at prestudy screening, at baseline, and typically at Weeks 2, 4, and 8. In the VLD CBP-treated group from baseline to Week 8, total time awake decreased from 1.3 h to 0.8 h, which was an improvement of 38.5% (p = 0.011), while total sleep time increased 5.7 h to 6.4 h, an improvement of 12.3% (p = 0.005) and sleep efficiency increased 73.6% to 85.1%, an improvement of 15.6% (p = 0.023). In contrast, placebo treatment did not result in statistically significant changes in any of these measures. Compared to placebo, VLD CBP treatment did not significantly change total time awake, total sleep time, or sleep efficiency (Table 4).

Within the VLD CBP group from baseline to Week 8, VLD CBP treatment did not significantly change the percentages of Stage 1, 2, 3, or 4 or REM sleep (Table 4). However, in a change from baseline analysis, compared to placebo, VLD CBP was associated with an 11.1% increase in Stage 2 sleep (56.5% vs 50.6% for placebo; p = 0.021), a 55.1% decrease in Stage 4 sleep (7.5% vs 13.4% for placebo; p = 0.029), and a 17.3% decrease in REM sleep (15.9% vs 19.1% for placebo; p = 0.007; Table 4).

VLD CBP effects on CAP_A2+CAP_A3

Subsequent to enrollment and completion, EEG sleep studies in FM patients were reported that identified increases in the periodic sleep EEG arousal disorder known as CAP in nonREM sleep^26,27. Therefore, an analysis of sleep EEG CAP was performed that measured subtypes CAP_A1, CAP_A2, and CAP_A3 and Total CAP (or CAP_A1 + CAP_A2 + CAP_A3). Subtype CAP_A1 is associated with sleep maintenance or least sleep instability, and subtypes CAP_A2 and CAP_A3 are associated with moderate to prominent increases in sleep instability.

Because CAP_A2 and CAP_A3 are most closely associated with sleep instability^26,27, the sum of CAP_A2 + CAP_A3 rates (CAP_A2+A3) was used as an indicator of disordered sleep. CAP_A2+A3 was normalized (CAP_A2+A3(Norm)) by dividing CAP_A2+A3 by the total CAP rate (CAP_total = CAP_A1 + CAP_A2 + CAP_A3 rates = CAP_A1+A2+A3) and expressed as a percentage. Therefore, CAP_A2+A3(Norm) = 100 × CAP_A2+A3/CAP_A1+A2+A3 and this yields the percentage of total CAP that is associated with sleep instability.

To determine whether patients experienced nights with a potential CAP response to treatment, it was necessary to determine an empirical threshold below which CAP_A2+A3(Norm) values reflect a night of relatively stable sleep for this population. To determine a threshold for CAP_A2+A3(Norm) that could be informative for a potential treatment response, the study CAP data were then evaluated by considering a range of cutoff values for CAP_A2+A3(Norm) from ≤ 10% to ≤ 50%. Testing various CAP_A2+A3(Norm) values revealed that defining a threshold for response CAP_A2+A3(Norm) ≤ 33% distinguished VLD CBP-treated subjects from placebo-treated subjects (Table 5), at which threshold the percentage of patients with increased nights of CAP response while on treatment (ITT, LOCF) was 72% with VLD CBP compared to 33% with placebo (p = 0.019). The finding that CAP_A2+A3(Norm) ≤ 33% may be a relatively healthy CAP_A2+A3(Norm) value for patients with FM is consistent with our analysis of previously published normative data²⁶.

Correlation of CAP_A2+A3(Norm) with FM symptoms

To evaluate whether increased nights with CAP_A2+A3(Norm) ≤ 33% was correlated with clinical improvement measures in pain, fatigue, tenderness, HAD, and HAD depression over the course of the study (LOCF Week 8), Spearman’s rank correlation was investigated separately for each treatment. Data were coded such that improvements were positive. Within the VLD CBP-treated patients, increased nights with CAP_A2+A3(Norm) ≤ 33% was found to correlate positively to decreases in fatigue (rho = 0.62, p = 0.006), HAD total score (rho = 0.505, p = 0.033), HAD depression subscale (rho = 0.556, p = 0.017), patient-rated fatigue (rho = 0.614, p = 0.007), and clinician-rated fatigue (rho = 0.582, p = 0.0112; Table 6). In contrast, improved CAP response was not correlated with either musculoskeletal pain or dolorimetry results (Table 6). Within the placebo-treated subjects, none of these FM symptoms or sleep EEG measures was significantly correlated to increased number of nights with CAP_A2+A3(Norm) ≤ 33% (Table 6). In the placebo group, increased nights of CAP response correlated with measures of improved sleep, that is, a positive correlation with sleep efficiency and a negative correlation with total time awake. Together, these findings suggest that nights with CAP_A2+A3(Norm) ≤ 33% may reflect relatively healthy or restorative sleep for patients with FM as symptoms vary naturally over the course of the condition, as well as providing a potential biomarker for treatment effects.