What treatment works best for restless legs syndrome? Meta-analyses of dopaminergic and non-dopaminergic medications

Abstract

At the time of writing only dopamine agonists are licensed for the treatment of restless legs syndrome (RLS) in various countries, but randomized controlled trials (RCTs) have been performed with other treatments. We performed comprehensive meta-analyses and indirect comparisons of RCTs for all currently recommended treatments of RLS. We searched the Central, Medline, Embase, PsycINFO and CINAHL databases. Outcome measures were the international RLS study group severity scale (IRLS), clinical global impression-improvement, (CGI-I), periodic limb movement index (PLMI), and psychosocial parameters such as quality of life (QoL). We also conducted indirect comparisons by testing for heterogeneity between the substance groups. Placebo (58 trials) and actively (4 trials) controlled RCTs with dopamine agonists (38 trials), levodopa (4 trials), anticonvulsants (13 trials), most of them with α₂δ ligands (11 trials), opioids (1 trial), and iron treatments (6 trials) were included (9596 patients). Although treatment effects showed large variations, changes in the IRLS in the substance groups were comparable (P = 0.78), with a mean reduction in the IRLS of -5.47 points for dopamine agonists, -5.12 points for anticonvulsants (α₂δ ligands and levetiracetam), and -4.59 points for iron treatments. The CGI-I indicated slightly different treatment effects between the substance groups, while PLMI changes during treatment differed (P = 0.002), showing a marked decrease with dopamine agonists (-22.50/h), levodopa (-26.01/h), and oxycodone (-34.46/h) compared with a slight decrease for anticonvulsants (α₂δ ligands and levetiracetam; -8.48/h) and iron treatments (-13.10/h). Quality of sleep and QoL improved moderately in most of the RCTs investigating these parameters (standardized mean difference, SMD) 0.40 and 0.33, respectively). In the few studies evaluating the change of depressive (n = 4) or anxiety symptoms (n = 3), these symptoms improved slightly (SMD -0.24, and -0.21). Adverse effects and dropouts were comparable in number across all substance groups. In meta-regressions, the treatment effect was predicted by the design of the trial (the more sites involved in a trial the lower the effect) and by the duration of action of a medication (the longer the duration of action, expressed as the half-life time of a substance, the greater the improvement), the latter indicating potential superiority of treatments with stable blood concentration. This first meta-analysis of all RCTs for the pharmacological treatment of RLS provides evidence that, besides the well-defined efficacy of dopaminergic treatment, other treatments with different pharmacological principles show efficacy in small samples and may be well-tolerated alternatives for the treatment of RLS. In the group of anticonvulsants, only the trials performed with α₂δ ligands such as gabapentin, gabapentin enacarbil, and pregabalin showed good efficacy. This indicates a specific mechanism of action of these substances in RLS. The group of iron treatments consisted of a few trials with different compounds in oral and intravenous application form, respectively. For a more differentiated evaluation of the efficacy of iron treatments further studies are necessary. The large efficacy of one opioid RCT in RLS has to be confirmed in further studies.