Randomized, double-blind, placebo-controlled trial of pergolide in restless legs syndrome

Restless legs syndrome (RLS), or Ekbom's syndrome,¹ affects 2 to 15% of the population.^2-4 A major symptom of the syndrome is uncomfortable sensations in the legs that cause difficulty in getting to sleep.^2,5 Once asleep, the sleep may be disrupted by frequent periodic or semirhythmic movements of the legs referred to as periodic limb movements of sleep (PLMSs).^5,6 As a consequence of these sensory and motor phenomena, sleep is often disturbed and profound daytime sleepiness occurs.²

Although the mechanism underlying RLS is unknown, there is circumstantial evidence to suggest a role for the dopaminergic system in the etiology of RLS.⁷ Levodopa was first demonstrated to be effective for PLMS by Akpinar.⁸ There have been at least five additional controlled studies that have demonstrated the significant benefit of levodopa on RLS and PLMS.^9-13 However, sustained use has been found to produce serious problems, with paradoxical worsening of symptoms referred to as RLS augmentation.¹⁴ A second issue is the short half-life of less than 2 hours for the standard levodopa formulation, which often requires the patient to take a second dose after awakening in the middle of sleep and may cause a rebound of symptoms in the early morning.¹⁵ The dopamine receptor agonist pergolide was demonstrated in an uncontrolled, open-label trial to provide effective treatment of RLS without requiring an early morning dose and with reduced incidence of augmentation.¹⁶ However, pergolide in that clinical series was used only after levodopa failed to maintain clinical efficiency. The current study was designed to evaluate the effects of pergolide in treating RLS using a randomized, double-blinded, placebo-controlled method.

Methods. Patients were identified from our Sleep Disorders Center. To be eligible for the study, all four criteria for RLS as outlined by the International RLS Study Group¹⁷ had to be met: uncomfortable sensations in the legs with an urge to move them, motor restlessness while awake, symptoms worse at rest and relieved with walking, and symptoms worse at night/bedtime compared with other times of day. Because this study was designed to use both subjective (symptom reports) and objective (presence of PLMs and sleep efficiency) outcome measures for reporting treatment effect, we required all participants to demonstrate a minimum of 15 PLMSs per hour on a standard polysomnogram.^9,11 The first 16 consecutive patients who met these criteria, who were not treated previously with pergolide, and who gave informed consent were randomized into the study.

Of the 16 patients, seven were not receiving treatment for RLS. The remaining patients were on some form of medication for their RLS: carbidopa/levodopa (n = 5), propoxyphene (n = 2), clonazepam (n = 1), and alprazolam (n = 1). All current RLS medications were discontinued at least 4 days before baseline assessment. Two patients had severe rebound symptoms when carbidopa/levodopa was discontinued. They were treated for 2 nights with temazepam 30 mg, which was discontinued 48 hours before baseline assessment. Patients were randomized to either placebo or pergolide treatment in a double-blind fashion with eight patients per group (figure 1). The pre-enrollment treatment status for those randomized to pergolide were three subjects with no treatment, three subjects on carbidopa/levodopa, one subject on alprazolam, and one subject on propoxyphene.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Patient randomization and outcome.

Each participant received for each day of the study two separate packets containing the maximum number of capsules to be taken for each of the two daily doses (dinner time and 2 hours before bedtime). The packets were marked with the subject number, date, and the time when the capsules were to be taken. Identical-appearing capsules were used for placebo and pergolide treatments. The dose was increased by the patient over 14 days until the patient felt the benefits from the medication at that dose were adequate and therefore no further increases were made, the maximum dose of 13 capsules was reached, or adverse effects occurred that limited additional increases. The final dose of pergolide reached by an individual could, therefore, range from 0.05 mg twice a day to a maximum of 0.3 mg at dinner and 0.35 mg before bedtime. No further changes in dosage were permitted after day 14. Thus, the dosage remained constant for the 5 days from day 14 through day 18. To avoid any complications from sudden pergolide withdrawal and still not break the double-blind paradigm until all data were analyzed, the same capsules used in the treatment period of the study were provided for each patient in a special package for a clinically managed, tapered withdrawal from treatment. This started at day 19 and could last as long as 5 days.

Patients were seen weekly to receive new medications, to discuss dose adjustments, and to report any adverse effects. The study coordinator (J.B.Y.) was available on a daily basis to deal with any questions about side effects or dose adjustment. For any reported adverse event, the patient was to rate it as mild, moderate, or severe. Adverse events were reviewed by one of the investigators (C.J.E.) to determine whether the dose of medication should be decreased, maintained at the current dose, or increased according to the schedule. The investigator (R.P.A.) responsible for obtaining subjective reports of outcome and collecting objective sleep data was blinded to any dose adjustments or reports of adverse reactions.

A complete, all-night polysomnogram (PSG) was obtained during the 2 days before initiating treatment (baseline) and during the last 2 days of treatment (days 17 and 18). To reduce "first-night" effect,¹⁸ only the PSG data from the second of the two nights were analyzed. The PSG was obtained from sleep in the patient's home using the Biologic Sleepscan Traveler System (Chicago, IL). The PSG recorded EEG activity (C3-A2, C3-01, 01-A2, 02-A1), bilateral eye movements, respiratory effort (abdominal and thoracic), airflow (oral and nasal), and submental and bilateral anterior tibialis electromyographic activity. Sleep was scored according to the Rechtshaffen and Kales criteria.¹⁹ Sleep efficiency was calculated as the ratio of total sleep time to the total time in bed, multiplied by 100 to give a percentage. The PLMSs were identified and scored using a modification of the Guilleminault criteria advanced by Coleman.²⁰ The PLMSs per hour were calculated as the total number of PLMSs during sleep divided by the total sleep time.

An assessment of RLS symptom duration was obtained during the same 4 days that the PSGs were performed. At the end of each hour while awake, patients were asked to mark on a daily log sheet whether any RLS symptoms had occurred during the preceding hour. They were also asked to mark the time when asleep and the time when lying down awake. The number of hours a patient experienced RLS symptoms for a given day was defined as starting from the first hour with RLS symptoms after 12:00 noon and continuing until the patient reported having at least 1 hour of either sleep or lying still without any RLS symptoms. The data from each 2-day period were averaged to give a final value in number of hours per day. On the morning of day 18, the patients rated their overall improvement (global improvement score) in RLS symptoms compared with baseline using a scale from 0 to 100%. Both of the subjective measures used in this study have been used by us in clinical practice for many years; however, their sensitivity and specificity have never been validated.

Data analyses. The primary outcome measures were the number of PLMSs per hour, percent sleep efficacy, the number of hours per day with RLS symptoms, and the global improvement score. Student's t-test was used for all comparisons except for intergroup comparisons for gender difference in which chi-square analysis was performed. A two-tailed t-test was used for intergroup comparisons on other baseline variables. Because we were specifically hypothesizing improvements in outcome variables with pergolide, a one-tailed t-test was used to test treatment effects. Our previous studies^9,11 comparing placebo with carbidopa/levodopa showed very large treatment effects of 2.0 to 2.5 for these measures. Because the study assumption was the pergolide could be used as a substitute for levodopa, the sample size was set to detect a similar size of treatment effects. With expected effects this large, sample sizes of seven to eight for each group provide power >0.80 for intersubject analyses with α = 0.05. Thus for this first double-blind study, a sample size of 16 (eight per group) was selected as conservative for showing the expected large treatment effects. Larger sample sizes would require a justification other than determining the desired and expected degree of clinical efficacy.

Results. There were no significant differences between treatment and placebo groups for any baseline variables (table). However, the PSG data from days 17 and 18 were lost for one man in the placebo group. Owing to technical problems related to extreme movement during day 17 recording, the PSG was not interpretable. He did complete the log, so the duration of RLS for day 17 was obtained. On the morning of day 18 he gave his subjective estimate of improvement but that night, because of severe RLS symptoms, he started himself back on his previous medication near 1 AM, which invalidated the PSG for day 18. The duration of RLS symptoms was calculated from the day 18 sleep log with the cutoff at 1 AM.

Significant treatment effects (p < 0.05) were seen with pergolide for all outcome measures (figures 2 and 3). The patients on pergolide showed an improvement in sleep efficiency from a mean (± SEM) of 61 ± 5% to 79 ± 2%, a reduction in leg movements from a mean of 48.9 ± 7.8 PLMSs per hour to 14.5 ± 5.8 PLMSs per hour, and a decrease in the duration of RLS symptoms from a mean of 7.0 ± 2.3 hours/day to 1.8 ± 1.3 hours/day. The average global improvement score was 61 ± 14% for pergolide compared with 19 ± 11% for placebo treatment (p = 0.009). Five patients reported almost no RLS symptoms on pergolide; however, one patient on pergolide reported no overall improvement despite a reduction in leg movements from 66 to 24 PLMSs per hour and an improvement in sleep efficiency from 65 to 75%. The placebo group showed no statistically significant change from baseline for any of the four outcome variables. However, one patient from the placebo group reported a 90% improvement in symptoms despite minimal changes in the objective measures: PLMSs of 39 versus 36 per hour and sleep efficacy of 77% versus 73%.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Sleep variables at baseline (black bars) and on treatment (shaded bars) for each medication group (mean ± SEM). *Baseline versus treatment.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Mean (±SEM) hours per day of restless legs syndrome symptoms at baseline (black bars) and on treatment (shaded bars). *Baseline versus treatment.

The patients on pergolide reached a median final daily dose of 0.35 mg pergolide. Two patients responded to as little as 0.1 mg total dose, whereas one patient reached the maximum daily dose of 0.65 mg pergolide. This latter patient was the one who reported having no benefit from the medication. In contrast, the patients on placebo reached a median final capsule usage of 13, with only one person not reaching the maximum daily capsule use of 13. This patient's final capsule usage was two and is the only placebo recipient who reported complete symptomatic relief.

Adverse effects were commonly reported by both placebo and pergolide recipients and were, for the most part, mild. During the 3-week period, side effects were reported as severe by four patients who took pergolide (stomach pain, constipation, and increased dreaming) and by two patients on placebo (terrible taste and itchy eyes). There were no symptoms to suggest RLS rebound or augmentation. None of the patients discontinued treatment or limited the dose they wished to use because of adverse effects.

Discussion. All the patients receiving pergolide had improvement in sleep efficiency and PLMSs per hour, and all but one patient showed improvement in the subjective measures of RLS symptoms. The one patient who reported no improvement in symptoms despite having moderate improvements in sleep efficacy and leg movements had reached the maximum dose of 0.65 mg. Additional increases in the pergolide may have led to more subjective relief of this patient's symptoms. However, we had to limit the maximum dose used so as not to prolong the placebo period. As anyone with moderate to severe RLS can attest, going one night, let alone 3 weeks, without treatment is almost unthinkable. Therefore, the single biggest contention in doing good studies with RLS is the time spent in the placebo period. The one patient on placebo who resumed his medication on day 18 is an example of how desperate patients can feel and also shows how difficult it is to maintain the integrity of the experimental design in the face of such potential attrition.

The subjective improvement in symptoms by four placebo recipients underscores the importance of placebo-controlled, double-blinded studies and the use of objective measurements in evaluating any treatment of RLS. A large placebo effect was also reported in the Scandinavian study of carbamazepine treatment of RLS, which relied solely on subjective reports of outcome.²¹ In that study, 29% of the patients reported complete remission of RLS "attacks" and another 31% reported few (one or two) "attacks" for the fifth week on placebo. In our current study, the placebo response rate was about the same as in the Scandinavian study. One patient (12.5% of the sample) reported virtually complete remission of symptoms and another three patients (37.5% of the sample) reported some reduction in symptoms. In contrast, the objective measures used in this study showed that none of the patients on placebo had significant improvements.

Overall, these results strongly support the effectiveness of pergolide in the management of RLS. Although unblinded, clinical series suggest a role for pergolide in the long-term management of this syndrome.⁶ Larger, multicenter trials with longer periods of follow-up are required to determine better how effective long-term use of this drug might be.

Acknowledgment

Martin Kohler, MD, of Athena Neurosciences kindly provided valuable consultation for this study; Carol Sterling-Ward of Johns Hopkins Bayview Medical Center provided secretarial support in preparing this manuscript; and Thomas Lehr, R.EEGT, of Johns Hopkins Bayview Medical Center provided technical assistance for the polysomnograms.