Pharmacological responsiveness of periodic limb movements in patients with restless legs syndrome: a systematic review and meta-analysis

Abstract

Study Objectives:

Periodic limb movements during sleep (PLMS) are a frequent finding in restless legs syndrome, but their impact on sleep is still debated, as well the indication for treatment. We systematically reviewed the available literature to describe which drug categories are effective in suppressing PLMS, assessing their efficacy through a meta-analysis, when this was possible.

Methods:

The review protocol was preregistered on PROSPERO (CRD42021175848), and the systematic search was conducted on and EMBASE (last searched on March 2020). We included original human studies, which assessed PLMS modification on drug treatment with a full-night polysomnography, through surface electrodes on each tibialis anterior muscle. When at least 4 studies were available on the same drug or drug category, we performed a random-effect model meta-analysis.

Results:

Dopamine agonists like pramipexole and ropinirole resulted the most effective, followed by l-dopa and other dopamine agonists. Alpha2delta ligands are moderately effective as well opioids, despite available data on these drugs are much more limited than those on dopaminergic agents. Valproate and carbamazepine did not show a significant effect on PLMS. Clonazepam showed contradictory results. Perampanel and dypiridamole showed promising but still insufficient data. The same applies to iron supplementation.

Conclusions:

Dopaminergic agents are the most powerful suppressors of PLMS. However, most therapeutic trials in restless legs syndrome do not report objective polysomnographic findings, there is a lack of uniformity in presenting results on PLMS. Longitudinal polysomnographic interventional studies, using well-defined and unanimous scoring criteria and endpoints on PLMS are needed.

Citation:

Riccardi S, Ferri R, Garbazza C, Miano S, Manconi M. Pharmacological responsiveness of periodic limb movements in patients with restless legs syndrome: a systematic review and meta-analysis. J Clin Sleep Med. 2023;19(4):811–822.

INTRODUCTION

Periodic limb movements (PLM) are stereotyped motor events characterized by dorsiflexion of the foot, sometimes accompanied by a flexion of knee and hip, occurring monolaterally or bilaterally, both during sleep (PLMS) and rest wakefulness.¹ The main feature of PLM is their periodicity, which ranges from 10 to 90 seconds, although typically it is approximately 20–40 seconds. PLM are organized in series of 4 or more consecutive events and are interrupted by intermovement intervals longer than 90 seconds or shorter than 10 seconds or by movements not fulfilling the criteria for duration (0.5–10 seconds, if monolateral, or up to 15 seconds, if bilateral).¹ Initially described by Allison in 1943² and later named “nocturnal myoclonus” by Symonds,³ PLMS were recorded for the first time in 1965 by placing 2 surface electrodes on the tibialis anterior muscles during sleep in a patient affected by restless legs syndrome (RLS).⁴ Despite their occurrence in up to 90% of patients with RLS,⁵ they are a frequent finding also in other sleep and neurological disorders (eg, narcolepsy, sleep apnea, Parkinson disease, etc.).⁶

With the recording of PLMS becoming a standard procedure, it was evident that they were a common “incidental” finding also in the general “healthy” population, especially among older populations.⁷ In a recent Swiss cohort study, the prevalence of individuals older than 35 years with a PLMS index > 15 was 28.6%.⁸

Evidence suggests the D3 dopamine receptor subtype,⁹ located in the spinal cord,¹⁰ as the possible target of dopaminergic agents (DA) for PLM.

Association between PLMS, arousals, and autonomic activations

Usually, PLMS are temporally associated with cortical arousal and autonomic activations,¹¹ but their role in producing sleep fragmentation as well as in being a risk factor for cardiovascular disease is not definitely demonstrated.¹² The acute pharmacological suppression of PLMS with low doses of dopamine agonists does not affect equally electroencephalograpy arousability nor is the acute pharmacological reduction of arousals with benzodiazepines followed by a proportional decrease of PLMS.¹³ The finding of a positive correlation between PLMS index and a poor sleep quality⁸ and between the duration of single leg movements and the duration of single electroencephalograpy arousal,¹⁴ does not imply a causal relationship between the 2 phenomena.

Similar considerations are also valid for the so-called “Periodic Limb Movements Disorder” (PLMD), which, according to the International Classification of Sleep Disorders (ICSD-3),¹⁵ is defined as a symptomatic form of PLMS associated with “clinical sleep disturbance or daytime impairment”, not otherwise explained, and without RLS. However, also for PLMD, systematic investigations on the eventual benefit of PLMS treatment on PLMD-related insomnia or sleepiness are still missing. Therefore, the clinical meaning of PLMS is still not fully understood, and, despite safe and prompt pharmacological suppressors of PLMS are available, there is no common agreement on the need for treatment of PLMS in patients without RLS nor in those with RLS. A possible prognostic issue encouraging the treatment of PLMS is represented by their supposed long-term cardiovascular implications. Limb movements are usually associated with autonomic activations such as repetitive increase in heart rate variability and blood pressure.^16–19 A higher cardiovascular risk associated with PLMS has been reported in specific clinical groups.^20–22 However, in studies conducted on the general population, the correlation was not significant after adjustment for confounding factors,⁸ and in other studies PLMS were not associated with cardiovascular events.²³

Trials on PLMS

Interventional trials with PLMS as a specific outcome measure are very few. Most of the literature data concern RLS, with sensory symptoms or sleep quality used as the main endpoints. Surrogate analyses on the pharmacological response of PLMS in patients with are also scarce, because in most studies a PSG assessment was not performed.

In addition to DA, other drugs demonstrated to be effective on PLMS with or, more rarely, without RLS. However, despite an objective measure for PLMS and availability of interventional studies on PLMS, a comparison of the efficacy between different pharmacological compounds on PLMS has never been done and still now remains unclear which drug is more effective on PLMS.

The present review and meta-analysis focuses on the pharmacological effect of different drugs on PLMS associated with idiopathic RLS. The results might contribute to the understanding of which neurological circuits and neurotransmitters are implicated in the mechanism of PLMS and guide clinicians in the pharmacological management of PLMS.

METHODS

Review

Search strategies

We performed a systematic literature search through and EMBASE databases, selecting only English-language peer-reviewed articles up to March 2020. The review process followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guideline.²⁴ The protocol for this systematic review was registered on PROSPERO (CRD42021175848).

The initial search was performed crossing the following keywords: “periodic legs movement” or “periodic limb movement” and “restless leg syndrome” with “therapy” or “treatment” or “trial” or “drug”. The search fields were limited to title, abstract, and/or keywords. Afterward, duplicates were removed, and the items retrieved were classified as eligible or not according to selection criteria described below. Reference lists of the selected publications were also examined to assure that no relevant articles were missed (Figure 1).

Figure 1. Flowchart.

Study selection according to systematic review process.

Inclusion and exclusion criteria

For the present analysis we decided to focus on PLMS associated with idiopathic RLS; therefore, we excluded data on PLM during wake and studies reporting PLMS without RLS or occurring within other neurological or medical diseases (symptomatic RLS), such as uremic RLS, RLS associated with neurodegenerative disorders. Iron deficiency per se was not an exclusion criterion.

We selected original studies on human participants, which assessed PLMS modification on drug treatment, including both controlled or open trials and case series. Reviews, meta-analysis, and conference abstracts were excluded.

Further inclusion criteria were sample size of at least 5 patients with primary RLS, participants at least 18 years old, and PLMS index available for baseline and treatment condition. To perform the meta-analysis, mean and standard deviation had to be reported clearly in both conditions, or they should report another parameter from which standard deviation could be derived (such as standard error), or individual data for each participant, so that means and standard deviations could be calculated.

To be included in the analysis, PLMS had to be assessed in a context of a full-night polysomnography (PSG) through bipolar surface electrodes on each tibialis anterior muscle.

Studies with the following features were excluded: sample size smaller than 5 patients, participants younger than 18 years, symptomatic RLS, PLMS in the context of PLMD, and PLM assessed with diagnostic approach different from PSG, ie, actigraphy or actimetric foot movement counter devices. We decided to exclude studies evaluating PLM through actigraphic recordings, because it is not recommended for this scope and does not provide reliable estimates of PLM.²⁵ Moreover, studies comparing the performance of actigraphic scoring of PLM compared to polysomnography are quite few, have different methodologies, and the actigraphs analyzed have different specificity and sensitivity, with difficulty in integrating data from each leg.²⁶

Also, we excluded papers in which PLM were counted per hour in bed or in which the change in PLMS index was reported only for arousal associated PLMS (PLMS arousal index), as this may give only partial information about PLMS and global PLMS index. PLMS index (number of PLMS per hour of sleep), and the difference between baseline and treatment night, was the objective measure we focused on. Studies reporting only a mean change of the PLMS index (delta value resulting from the difference before and after treatment) but not the mean baseline and posttreatment values were excluded.

Meta-analysis

We pooled PLMS index values for different drug treatment categories (at least 4 studies available on the same drug or drug category, see Results) using a random effects model meta-analysis. Means, standard deviations, and sample sizes were used for the computation of effect size and variance, which were then used as input parameters. For all meta-analyses, the DerSimonian-Laird estimator was used because it does not need data that are normally distributed.²⁷ Heterogeneity of parameters was assessed using the Cochran's Q-test, with a P < .10 indicating evidence of heterogeneity.²⁸ Heterogeneity was also quantified by the I² statistics (expressed as a percentage)²⁹; values greater than 50% are considered to indicate substantial heterogeneity.³⁰

The results of the meta-analyses carried out were then used to draw forest plots for each of them.

RESULTS

The initial search produced a total of 816 papers; after removal of duplicate (n = 170) and non-English (n = 32) records, we determined the references of interest through title and abstract and then we went through the full text of the “eligible” and “maybe eligible” studies. We retrieved few additional articles through bibliography review of the selected articles (Figure 1).

A total of 46 articles met the criteria for the review, some of them involved more than 1 compound and, thus, the total of cites listed in Table 1, Table 2, and Table 3 is higher than this total because some of the articles are cited more than once, as needed. All tables report the substance used, the citation of the paper, the dosage and duration of treatment, the number of patients treated, the change in PLMS index expressed as a percentage from baseline or the placebo values (as available) and the scoring criteria used.

Table 1.

Studies on the effects of l-dopa or dopamine agonists other than pramipexole.

Compound	Study	Daily Dosage (duration)	Patients	Criteria	PLMS Index		RLS Severity
					Baseline	Treatment	Baseline	Treatment
l-dopa	Montplaisir et al, 198635	l-dopa, 100–200 mg (5 weeks)	7	Coleman	15.9 ± 5.0	16.3 ± 4.2	2.8 ± 0.1§	0.8 ± 0.4
	Guilleminault et al, 199333	l-dopa/benserazide, 100/25 mg (4-7 weeks)	20	Coleman	101 ± 52	17 ± 22	not reported
	Eisensehr et al, 200432*	l-dopa/benserazide, 100/25 mg (3 weeks)	20	?	43.2 ± 36.9	19.9 ± 23.2	5.59 ± 1.7§	4.4 ± 2.5
	Polo et al, 200734*	l-dopa/carbidopa, 100/25 mg (2 days)	28	ASDA	23.7 ± 17.3	9.5 ± 14.8	not reported
		l-dopa/carbidopa, 50/12.5 mg + entacapone 200 mg (2 days)	28			12.6 ± 12.0
		l-dopa/carbidopa, 100/25 mg + entacapone 200 mg (2 days)	28			6.4 ± 8.6
		l-dopa/carbidopa, 150/37.5 mg + entacapone 200 mg (2 days)	27			3.5 ± 6.0
Pergolide	Earley et al, 199836*	0.1–0.65 mg (18 days)	8	Coleman	48.9 ± 7.8	14.5 ± 5.8	7.2 ± 2.3§	1.8 ± 1.3
Cabergoline	Stiasny et al, 200037	1–4 mg (12 weeks)	9	ASDA	195.8 ± 109.1	26.4 ± 40.2	4.9 ± 4.3§	0.4 ± 0.7
Bromocriptine	Manconi et al, 2011a9*	2.5 mg (single dose)	15	WASM 2006	51.4 ± 39.86	30.9 ± 33.50	6.9 ± 2.20$	3.3 ± 1.72
Lisuride	Benes et al, 200638	0.1–0.4 mg (4 weeks)	10	?	28.3 ± 17.6	10.1 ± 14.5	not reported
Ropirinole	Saletu et al, 200041*	0.5 mg (single dose)	12	ASDA	26.8 ± 25.1	10.3 ± 8.2	not reported
	Happe et al, 200340*	0.25–1.5 mg (4 weeks)	8	?	48.4 ± 24.6	13.2 ± 13.5	15.9 ± 2.4£	8.1 ± 4.9
	Bliwise et al, 200593*	0.2–6 mg (4 + 2 additional weeks)	22	ASDA	19.7 (0–45.6)	19.8 (0–44.4)	22.6 ± 4.6£	8.7 (2.9–14.4)
	Allen et al, 200439*	0.25–4 mg (12 weeks)	29	ASDA	48.5 ± 31.7	11.8	not reported
	Saletu et al, 201042*	0.5 mg (single dose)	40	ASDA	36.0 ± 28.3	12.0 ± 13.6	16.4 ± 8.4£	N/A
Talipexole	Inoue et al, 199969	0.4–0.8 mg (4 weeks)	5	Coleman	≈67% reduction		not reported

*Placebo-controlled trial. ^§Custom scale. ^$Visual analog scale. ^£International RLS Severity Scale. Question marks [?] indicate that scoring criteria were not specified in the text. Data shown as mean ± SD or mean (95% confidence interval). ASDA = American Sleep Disorders Association, NA = not applicable, RLS = restless legs syndrome, WASM = World Association of Sleep Medicine.

Table 2.

Studies on the effects of pramipexole.

Compound	Study	Daily Dosage (duration)	Patients	Criteria	PLMS Index		RLS Severity
					Baseline	Treatment	Baseline	Treatment
Pramipexole	Montplaisir et al, 199951*	0.375–1.5 mg (4 weeks)	10	Coleman	77.1 ± 47.5	1.7 ± 3.1	12.7 ± 4.0§	1.5 ± 3.8
	Saletu et al, 200254*	0.27 mg (single dose)	11	ASDA	33.7 ± 58.8	8.2 ± 6.1	23.2 ± 6.5£	13.0 ± 10.1
	Stiasny-Kolster and Oertel, 200455	0.125–0.75 mg (1–-11 days)	17	?	73.3 ± 58.6	11.3 ± 17.2	29.8 ± 4.7£	7.3 ± 5.9
	Partinen et al, 200652*	0.125 mg (3 weeks)	20	AASM?	median 42.85	median 22.30	22.9 ± 4.2£	22.4 ± 4.7
		0.25 mg (3 weeks)	22			median 29.40		23.0 ± 3.4
		0.50 mg (3 weeks)	22			median 24.75		23.6 ± 3.7
		0.75 mg (3 weeks)	22			median 29.68		21.7 ± 4.6
	Merlino et al, 200694	0.125 mg (single dose)	13	WASM 2006	26.6 ± 15.9	5.8 ± 9.3	7.8 ± 1.0$	2.0 ± 1.7
	Manconi et al, 200748*	0.25 mg (single dose)	18	WASM 2006	45.8 ± 33.56	9.4 ± 11.40	7.4 ± 1.68$	1.3 ± 1.62
	Manconi et al, 200849*	0.25 mg (single dose)	25	WASM 2006	61.9 ± 58.06	11.1 ± 17.38	27.3 ± 5.2£	N/A
	Inoue et al, 201047*	0.125–0.75 mg (6 weeks)	21–22	?	median 22.15	median 0.0	23.4 ± 6.4$	7.3 ± 8.1£
	Ferri et al, 201045*	0.25 mg (single dose)	19	WASM 2006	40.2 ± 30.20	7.9 ± 10.66	6.8 ± 1.47$	1.3 ± 1.74
	Inoue et al, 201047*	0.125–0.75 mg (6 weeks)	20	ASDA	median 27.75	median 0.85	23.4 ± 6.4$	7.3 ± 8.1
	Manconi et al, 2011a9*	0.25 mg (single dose)	15	WASM 2006	41.6 ± 32.41	7.7 ± 9.15	25.2 ± 4.17$	N/A
	Manconi et al, 2011c50*	0.25 mg (single dose)	13	WASM 2006	30.9 ± 15.35	7.1 ± 10.57	>20$	N/A
	Manconi et al, 201213*	0.25 mg (single dose)	17	WASM 2006	42.7 ± 32.18	6.0 ± 8.31	25.9 ± 4.45$	N/A
	Galbiati et al, 201546	0.25 mg (12 weeks)	18	WASM 2006	53.64 ± 48.96	8.40 ± 16.50	26.20 ± 3.01$	not reported
	Rocchi et al, 201553	0.25 mg (12 weeks)	20	WASM 2006	19.9 ± 16.8	4.5 ± 7	27.3 ± 7.2$	10.5 ± 6.7
	Choi et al, 201744	0.125–1 mg (12–16 weeks)	12	WASM 2006	23.80 ± 26.00	5.29 ± 4.12	29.92 ± 4.98$	13.50 ± 10.94
	Thiedemann et al, 201356	0.18 mg (single dose)	15	WASM 2006	24.8 ± 19.6	5.9 ± 9.9	not reported

*Placebo-controlled trial. ^§Custom scale. ^$Visual analog scale. ^£International RLS Severity Scale. Question marks [?] indicate that scoring criteria were not specified in the text. Data shown as mean ± SD or mean (95% confidence interval). ASDA = American Sleep Disorders Association, N/A = not applicable, RLS = restless legs syndrome, WASM = World Association of Sleep Medicine.

Table 3.

Studies on the effects of α2δ-ligands, antiepileptics, benzodiazepines, opioids, and other drugs.

Compound	Study	Daily Dosage (duration)	Patients	Criteria	PLMS Index		RLS Severity
					Baseline	Treatment	Baseline	Treatment
α2δ-ligands
Gabapentin	Happe et al, 200158	300–1,200 mg (4 weeks)	9	?	38.2 ± 18.8	20.8 ± 23.9	16.7 ± 4.1£	8.2 ± 5.5
	Happe et al, 200340	300–1,200 mg (4 weeks)	8	?	39.2 ± 19.8	22.6 ± 24.9	16.6 ± 4.3£	6.8 ± 3.9
	Saletu et al, 201042*	300 mg (single dosage)	40	ASDA	29.0 ± 40.0	21.4 ± 29.9	16.6 ± 9.0£	N/A
Pregabalin	Garcia-Borreguero et al, 201057*	150–450 mg (12 weeks)	30	?	31.25 ± 24.96	13.79 ± 14.42	19.80 ± 4.16£	6.85 ± 6.87
Antiepileptics
Carbamazepine	Zucconi et al, 198963	3–7 mg/kg/day (30–45 days)	9	Coleman	27.8 ± 17.67	29.2 ± 28.57	4.6 ± 0.73§	2.4 ± 1.59
Valproic acid	Eisensehr et al, 200432*	600 mg (3 weeks)	20	?	43.2 ± 36.9	38.0 ± 32.3	5.5 ± 1.7$	3.8 ± 2.5
Perampanel	Garcia-Borreguero et al, 201762	2–4 mg (8 weeks)	20	AASM	27.8 ± 6.9	4.36 ± 2.0	23.7 ± 4.2£	11.5 ± 5.3
Benzodiazepines
Clonazepam	Manconi et al, 201213*	0.5 mg (1 night)	15	WASM 2006	44.1 ± 39.77	49.9 ± 43.41	21.3 ± 63.94£	N/A
Opioids
Oxycodone	Walters et al, 199364*	5–25 mg (2 weeks)	9	Coleman	38.8 ± 28.7	18.4 ± 26.5	5.1 ± 1.21§	2.5 ± 2.48
Various opioids	Walters et al, 200165	chronic	5	ASDA	29.42 ± 6.30	19.94 ± 11.73	not reported
Other Drugs
Dipyridamole	Garcia-Borreguero et al, 201888	100–400 mg (8 weeks)	13	AASM	26.7 ± 7.2	4.3 ± 1.9	23.4 ± 4.6£	10.7 ± 4.5
Iron sucrose	Earley et al, 200966	1,000 mg (single dose)	11	ASDA	107 ± 93	not reported	30.8 ± 9.2£	not reported
Apomorphine	Haba-Rubio et al, 200368	0.5 mg (single dose)	9	Coleman	58.5 ± 45.5	36.0 ± 36.5	not reported
Istradefylline	Deerce et al, 200767	80 mg (3 weeks)	5	AASM	41.6 ± 18.64	21.2 ± 22.22	29.8 ± 7.29£	24.6 ± 10.76
Clonidine	Wagner et al, 199671	0.5 mg (2 weeks)	10	ASDA?	36.8 ± 40.1	43.2 ± 36.4	2.2 ± 0.8§	0.5 ± 0.6
Bryophyllum pinnatum	Von Manitius et al, 201970	1,400–2,100 mg (8–10 weeks)	5	AASM	41.6 ± 38.30	30.2 ± 27.87	20.6 ± 2.61£	11.6 ± 7.09

*Placebo-controlled trial. ^§Custom scale. ^$Visual analog scale. ^£International RLS Severity Scale. Question marks [?] indicate that scoring criteria were not specified in the text. Data shown as mean ± SD or mean (95% confidence interval). AASM = American Academy of Sleep Medicine, ASDA = American Sleep Disorders Association, N/A = not applicable, WASM = World Association of Sleep Medicine.

Tables also list the concomitant reported effect on self-reported symptoms, expressed, when available, as International Restless Legs Syndrome Study Group Rating Scale (IRLS)³¹ score, or in other custom scales or visual analog scale, as specified.

Studies on l-Dopa and dopamine agonists

The most numerous group of studies listed in Table 1 involved l-Dopa and dopamine agonists, except for pramipexole, to which a separate table was dedicated, due to the number of studies published. Four studies reported the effects of l-Dopa in association with benserazide or carbidopa or entacapone. Three of them^32–34 reported an important reduction in PLMS ranging from −46.8% to −85.2%. The oldest study (with the smallest sample size) did not find a reduction in PLMS.³⁵ The duration of the drug therapy varied greatly among the different studies (from 2 days to 7 weeks) and within the same study in 1 case,³³ but it does not seem to be associated with any trend in the magnitude of PLMS decrease (Spearman rho correlation coefficient r_s = −0.8, not significant). Long-term efficacy data of l-dopa on PLMS are lacking.

Figure 2A, top shows the meta-analysis of the 4 papers mentioned above, with 1 of them providing 2 values, 1 for l-Dopa/carbidopa (100/25 mg) and another for l-Dopa/carbidopa + entacapone (100/25 + 200 mg).³⁴ In addition to the significant heterogeneity found (probably due essentially to the discordant single oldest study mentioned above³⁵), this analysis shows a significant overall effect size of l-dopa (−1.05 [confidence interval −1.60, −0.49]).

Figure 2. Meta-analysis of studies.

(A) Top panel: Meta-analysis of studies on l-dopa. Bottom panel: Meta-analysis of studies with different dopamine agonists, other than pramipexole. (B) Meta-analysis of studies on pramipexole. (C) Meta-analysis of studies on α2δ-ligands.

The same Table 1 includes results for dopamine agonists other than pramipexole. An old report on pergolide shows a 70.3% decrease in PLMS index.³⁶ A strong effect was also reported in the single study retrieved on cabergoline.³⁷ Only 1 study that matched the inclusion criteria was found on bromocriptine, which was found to reduce PLMS by approximately 40%.⁹ It was not possible to find any study testing rotigotine satisfying all the inclusion criteria. Lisuride was also used in 1 study, which reported a relatively strong effect on PLMS.³⁸ The effects of ropinirole on PLMS have been reported by 5 studies according to the inclusion criteria of our review,^39–42,93 which show relatively concordant percentage values of PLMS decrease (from 61.6 to 75.5%). Also, in this group of reports the duration of the drug therapy varied greatly among the different studies (from 1 single dose to 12 weeks), but, again, this does not seem to be associated with any trend in the magnitude of PLMS decrease (r_s = −0.190, not significant).

The single study about sumanirole had to be excluded because it mentioned only the mean change from baseline.⁴³

In Figure 2A, the bottom panel shows the meta-analysis performed by including the papers mentioned in the above paragraph that reported data allowing it. Also in this case, a significant heterogeneity was found (probably due essentially to the discordant single study on pergolide³⁶); however, this analysis also shows a large overall effect size of this heterogeneous group of dopamine agonists (−1.49 [confidence interval −2.23, −0.75]).

In Table 2, it is worth noting that pramipexole was used in 16 of the studies retrieved.^{9,13,44–56,94} On average, the decrease in PLMS index was reported to be approximately 75–80% (range 47.6–94.6). The study that evaluated multiple doses indicated a relative independence of the effect from the drug dose.⁵² Also with pramipexole, the duration of the drug therapy varied greatly among the different studies (from 1 single night to 12 weeks) and within the same study in 2 cases.^44,55 Similar to the effects described above, this variability in duration does not seem to be associated with any trend in the magnitude of PLMS decrease (r_s = −0.092, not significant).

Figure 2B displays the meta-analysis performed by including papers mentioned above that reported data allowing it. No significant heterogeneity was found along with a high and large overall effect size of pramipexole (−1.35 [confidence interval −1.56, −1.14]).

Studies on the effects of α2δ-ligands

Four studies were found to meet the inclusion criteria that reported the effects of α2δ-ligands on PLMS (Table 3) in patients with RLS.^40,42,57,58 Two studies about gabapentin enacarbil had to be removed because they displayed just the mean change in comparison with placebo.^59,60 The decrease in PLMS index was reported to range between 27.6% and 55.8% (on average, approximately 38%). A possible significant effect of therapy duration is indicated in this case by the Spearman rho correlation coefficient (r_s = −0.872, P = .054), with longer treatments accompanied by a more evident reduction in PLMS index. However, the study about gabapentin single dosage⁴² evaluated a low dose of this drug (300 mg) compared to the 4-week studies using up to gabapentin 1,200 mg. Interestingly, in only 1 study out of 5,⁴² the criteria that were used for PLMS scoring are clearly stated.

Figure 2C depicts the meta-analysis of the 4 papers reported in Table 3 on the effect of α2δ-ligands. The heterogeneity test was nonsignificant; however, I² almost reached the value of 50%. This analysis shows a medium-to-large overall effect size of these compounds (−0.67 [confidence interval −1.10, −0.23]) that is clearly lower than that found above for l-dopa and the other dopamine agonists.

Studies on the effects of benzodiazepines

Table 3 also includes the only 2 studies retrieved that analyzed the effects of benzodiazepines, with both of them using clonazepam as the compound of choice.^13,61 The approximate average change reported by these studies in PLMS index is −20%, with a wide range from −43.3% to +10.8, with 1 study reporting a nonsignificant increase.¹³ The older study by Peled and Lavie⁶¹ had to be removed because, among the patients complaining of poor nocturnal sleep, only 3 had RLS. The duration of the therapy with benzodiazepines also varied importantly (from 1 single night dose to 4 weeks); also in this case, the variable duration does not seem to be associated with any trend in the magnitude of PLMS change (r_s = −0.147, not significant).

As only 2 papers were available, a meta-analysis could not be performed.

Studies on the effects of antiepileptics

Three studies used different antiepileptics in patients with RLS (Table 3).^32,62,63 Only perampanel⁶² was found to induce a significant decrease in PLMS, while carbamazepine⁶³ and valproic acid³² were found to be ineffective on PLMS. No meta-analysis was possible on these studies.

Studies on the effects of opioids

Only 2 studies on the effects of opioids involving patients with RLS were eligible for our review (Table 3).^64,65 Different types of opioids and different dosages were used in small groups of patients. In case series, the PLMS index was reported to decrease by approximately 45% (range 32.3–52.6%). The duration of the treatment with opioids was very variable (from 2 weeks to 7 years); however, a significant tendency to attenuate with increasing duration was found for the decrease in PLMS (r_s = 0.975, P < .005).

As for benzodiazepines, and also in the case of opioids, the number of studies retrieved was not enough to perform a meta-analysis.

Studies on the effects of iron

Only 1 study on the effects of iron was eligible for our review, using a single dose of iron sucrose,⁶⁶ which reported a 22.3% decrease in PLMS index in 11 patients with RLS, evaluated 2 weeks after the infusion of a single dose of 1,000 mg split into 2 doses of 500 mg each and administered apart on 2 subsequent day.

Studies on the effects of various pharmacological treatments

Finally, Table 3 also lists studies on the effects of various pharmacological treatments on PLMS. Most of these studies show a decrease in PLMS that is reported to be significant in studies in which the sample size was sufficient for a reliable statistical analysis,^67–70 with the exception of clonidine, for which no change in PLMS index was demonstrated.⁷¹ It was not possible perform any meta-analysis with these available studies.

Commentary on the effect of antidepressants and antipsychotic agents

The aim of the review was to evaluate the effect of pharmacological treatment on PLMS, considering both the possibility to diminish or to increase them. However, the studies that evaluated drugs that could increase PLMS, mainly antidepressant and antipsychotic drugs, failed to meet the inclusion criteria, mostly because they involved psychiatric patients without a diagnosis of RLS.

Commentary on the effect on PLMS not associated with RLS

The present review does not consider the pharmacological response of PLMS alone not associated with RLS; therefore, is still unclear if asymptomatic PLMS or PLMS in the context of the so-called PLMD is sensitive to the same molecules active in RLS-related PLMS. Whether PLMD constitutes a distinct pathological entity is still a matter of discussion. Furthermore only few small sample size studies with a structured design tested the effect of drugs on PLMD. Bonnet and Arand et al⁷² did not find a significant reduction of PLM in patients with PLMD with low doses of triazolam compared to placebo. However, both night sleep and daytime sleepiness improved in patients treated with triazolam. Kaplan et al⁷³ confirmed the efficacy of l-dopa-carbidopa in suppressing PLM in 6 patients with PLMD. The same authors could not confirm the same efficacy of propoxyphene. Kunz et al⁷⁴ observed a positive effect of 3 mg of melatonin at bedtime on PLMS in 9 patients with PLMD in an open-label design. Through a retrospective open-label study on 31 patients with PLMD, Grewal et al⁷⁵ found a significant reduction of PLMS with selegeline in the dose range of 5 mg to 15 mg/day. The good response of PLM was not accompanied by a similar improvement of sleep parameters.

DISCUSSION

With this paper we reviewed for the first time the literature on the pharmacological responsiveness of PLMS in patients with idiopathic RLS. Most of the data come from self-initiated small and short-term trials, where the PLMS index usually represents a secondary endpoint besides the first outcome of sensory RLS symptoms. It is certainly noteworthy that most scientific therapeutic trials in RLS do not report objective PSG findings. This is probably due to high costs of PSG and to the uncertain pathologic value of PLMS. A further element that may detract interest in PLMS assessment might be the absence of a clear correlation between the PLMS index and the RLS severity.⁷⁶ On the other side, it should be considered that PLMS are free from placebo effect, which is high in RLS,⁷⁷ and that increasing literature on the cardiovascular impact of PLMS has become available in recent years.⁷⁸

Moreover, even when PSG with standard tibialis anterior EMG recording is performed, we noted a lack of uniformity in presenting the results on PLM, leading to a significant difficulty in interpreting and comparing studies. For instance, some studies were excluded by the analysis because reported PLM per hour of time in bed instead of the PLM per hour of sleep or because they focused only on PLMS associated with cortical arousal without mentioning the total PLMS index or do not even mention the PLMS index at all.

Also, among the studies included in the analysis, some other factors make their comparison difficult and increase the heterogeneity of the results. The most important one concerns the different scoring criteria used to define PLMS. This is due to the fact that the available literature covers many years, with the earliest trials using the Rechtschaffen and Kales or Coleman criteria⁷⁹ and the latest ones using different versions of American Sleep Disorders Association/American Academy of Sleep Medicine or World Association of Sleep Medicine criteria.^80–82 Differences in dose and in duration of the trials also account for additional contributors to heterogeneity.

It is worth notice that there is also lack of uniformity in reporting modification in RLS symptoms. Apart from the studies published before 2003, year of IRLS validation and also more recent ones differ, as some authors report IRLS pretreatment and posttreatment and others report baseline IRLS, but modification after treatment is expressed only as the change from baseline and other studies report baseline IRLS but treatment effect is described through a visual analog scale.

At a first glance, the results clearly indicate that DAs are the most powerful suppressors of PLMS. Pramipexole appears to be the most studied and the one with less heterogeneity in terms of results. On average, pramipexole reduces of about 80% PLMS in patients with RLS. Despite the available trials with pramipexole last no more than 12 weeks, the efficacy of the drug on PLMS does not seem to depend on the treatment duration. Similarly, at least in short trails, the efficacy on PLMS seems to be independent from the dose used, being equally effective up to 3 weeks also when the minimum dose of 0.125 mg was used.^52,55 Despite being less investigated, ropinirole shows a high efficacy in suppressing PLMS, although the reduction of PLMS obtained was about 70%, only a little less than pramipexole.⁸³ The dose equivalence ratio for pramipexole and ropinirole is calculated to be 1:4 (pramipexole 0.25 mg = ropinirole 1 mg).⁸⁴ l-Dopa, combined with benserazide or carbidopa, is able to reduce PLMS by about 50% in short-term trials.^32–35 The lower efficacy of l-dopa in comparison to pramipexole and ropinirole is probably due to its short half-life, which accounts for PLMS suppression mainly in the first half of the night, followed by a decreased efficacy or even a rebound effect in the second half of the night. This intranight rebound effect of l-dopa on RLS symptoms is well known in clinical practice and is well described on PLMS by Montplaisir et al³⁵ who did not find any efficacy of l-dopa on PLMS. Instead, Bassetti et al⁸⁵ showed that dual release l-dopa has an efficacy similar to pramipexole on PLMS; however, this study was excluded by the present review because of the outcome measure (PLM per hour in bed) and PLM measure through actigraphic monitoring. Ergoline derivatives cabergoline and pergolide, each in 1 single study, showed a good efficacy on PLMS, but for safety reasons these drugs are no longer used.⁸⁶ Only 1 study is available on bromocriptine, which has the lowest selectivity ratio on D3 vs D2 dopamine receptors subtype. The authors compared pramipexole (high affinity on D3 receptors subtype) to equivalent doses of bromocriptine, showing a significant difference in favor of pramipexole.⁹ This suggests D3 receptors as a possible main target for the suppressive effects of DA on PLMS. The location of the anatomical site of action of DA within the nervous system is still unknown; however, the preserved effect of DA on PLMS even in 1 participant with complete transverse cervical spinal lesion suggests the spinal cord as the possible site of action.¹⁰

Considering α2δ-ligands, there are many fewer studies available, and we performed the meta-analysis pooling together gabapentin and pregabalin. The studies evaluating the prodrug gabapentin enacarbil did not match the inclusion criteria for our review. These compounds resulted significantly effective on PLMS with an acceptable heterogeneity, despite a lower effectiveness compared to DA. Unlike dopamine-agonists and l-dopa, it seems that the treatment duration and the dose used might influence the outcome on PLMS, so that the absence of studies longer than 12 weeks for pregabalin⁵⁷ and longer than 4 weeks for gabapentin^40,58 should be taken into account. A further element to be considered is the relatively high dosages used of α2δ-ligands that might limit their use in the clinical practice because of possible side effects such as dizziness, gait problems, somnolence, and weight gain. Of course, DA can also have significant side effects; however, the dosages needed for the treatment of RLS are usually at the lower dosage range, at which they are usually quite well tolerated. It is difficult to establish the dose equivalence between DA and α2δ-ligands. In the unique, non-PSG study conducted so far, 300 mg of pregabalin (split in 2 doses of 150 mg each) was compared to 0.25 and 0.5 mg of pramipexole, with similar results in terms of efficacy on RLS symptoms, but with more frequent side effects and higher dropouts in the group treated with pregabalin.⁸⁷ In the same study, sleep quality improved more with pregabalin than pramipexole, confirming that DA are effective on sensory and motor symptoms but less on sleep structure.

Data about benzodiazepines are scanty and contradictory, with only 2 studies using clonazepam, which was ineffective on PLMS when evaluated in an acute setting at a dose of 0.5 mg¹³ and moderately effective (40% reduction of PLMS) when administered for 1 month and up to 2 mg/day.⁶¹ Antiepileptics like valproate and carbamazepine did not show a consistent effect on PLMS.^32,63 Very promising effects on PLMS have been shown with the antiglutamatergic perampanel in 1 short-term (8 weeks) open study, conducted with few patients, where small dosages of the drug suppressed more than 80% of PLMS. If these results are confirmed in larger controlled studies, the efficacy would be comparable with that of DA.⁶² Similar considerations can be adopted for the adenosinergic nonselective equilibrative nucleoside transporter 1/2 inhibitor dipyridamole, which needs further investigations.⁸⁸

Available data about PLMS responsiveness to opioids are very few and do not allow any meta-analysis.^64,65 However, opioids managed to reduce PLMS, with a magnitude that was similar to α2δ-ligands. Once more, treatment duration ranged from few days to 7 years. in this case, and treatment duration seems to influence PLMS reduction; this should again encourage and guide possible future studies also to take into account this aspect in evaluating treatment effect.

Data about other pharmacological treatments derive from only single studies, and the effect in modifying PLMS is less evident, such as in the case of iron sucrose.

The prominent effect of l-dopa and dopaminergic agents on PLMS seems to support the role of this pathway in generating both RLS symptoms and motor manifestations as PLMS.

Alpha2delta ligands act through glutamatergic modulation, which seems to be increased, at least at the thalamic site, in patients with RLS⁸⁹; furthermore, glutamatergic and dopaminergic networks interact between each other.⁹⁰ Perampanel is an α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor antagonist but seems to determine a greater decrease in PLMS than α2δ-ligands; thus, a specific role for this receptor pathway can be speculated, but studies and data are very scarce. Endogenous opioids might be deficient in the thalamus, and this was postulated to be a possible mechanism involved in sensory modulation in patients with RLS⁹¹; additionally, opioids interact with dopaminergic pathways in the spinal cord.⁹²

New studies are needed, with the specific aim of evaluating modification of PLMS with and without RLS. At the same time, it is urgent to establish a consensus on study methods and outcome measures to assess PLMS in future trials.

Desirable methodological features to study PLMS are bilateral tibialis anterior electromyographic recording during polysomnography, common standard scoring criteria, and longitudinal and long-term studies. It is also important to define clear endpoints, such as PLMS index at baseline and after intervention, PLMS arousal index, periodicity index, and the distribution of intermovement intervals. Other endpoints like PLM per hour of bed or PLM change index should be avoided or not provided at all without the previous ones. Comparative cross-over or parallel trials testing different medications should be also encouraged.

Prospective observational and interventional randomized controlled studies with PLMS as a specific outcome measure are warranted and will shed light on the crucial issue of the causative role of PLMS in sleep fragmentation and cardiovascular diseases.

We are aware that the inclusion criteria we chose are quite strict; however, it would be helpful if more clear-cut and rigorous measures are defined in clinical studies.

In conclusion, available data stand for DA, especially D3 selective agonists, as the strongest suppressor of PLMS in idiopathic RLS, followed by l-dopa and then by α2δ-ligands. Valproate and carbamazepine seem to be ineffective; opioids and benzodiazepine need further studies to confirm a moderate effect, while new compounds, such as dipyridamole and perampanel, seem to be promising but are still underinvestigated.

DISCLOSURE STATEMENT

All authors have seen and approved this manuscript. The authors report no conflicts of interest.

ABBREVIATIONS

DA

dopaminergic agents

IRLS

International Restless Legs Syndrome Study Group Rating Scale

PLMD

periodic limb movement disorder

PLMS

periodic limb movements during sleep

PSG

polysomnography

RLS

restless legs syndrome