Neuromodulation for Restless Legs Syndrome: A Systematic Review and Mechanistic Considerations for Spinal Cord Stimulation

Restless legs syndrome (RLS) disrupts sleep with patients characteristically experiencing an unrelenting urge to move the affected limb with voluntary limb movements providing only temporary relief. These symptoms show a circadian rhythm, peaking in the evening and during sleep. The disruption of restful sleep causes excessive daytime sleepiness, loss of productivity, depression, and anxiety. This additionally yields increased rates of morbidity and mortality from hypertension, stroke, renal and cardiac disease [1]. With up to 30 million US citizens suffering from RLS and/or its treatments, RLS is the most common sleep disorder [2].

Iron deficiency occurs in 25–40% of patients with RLS, and iron supplementation is the first-line treatment [3]. Standard oral medications include dopaminergic agonists (e.g., ropinirole, pramipexole, rotigotine) and ⍺2δ calcium channel ligands (gabapentin, pregabalin) [4]. Patients with severe or treatment-refractory symptoms often experience augmentation, requiring repeated dose escalations and can experience cognitive and/or behavioral side effects from polypharmacy. This leads most RLS patients to discontinue all standard medication therapies by 10 years, leaving them with chronic, distressing symptoms, an inadequate quality of life (QoL), and a paucity of viable treatment options [5]. Opioids may be utilized in treatment-refractory cases, but this option is associated with social stigma, addiction concerns, and increasing regulatory constraints [2,3]. This leaves many patients with treatment-refractory RLS symptoms that negatively impact their social, occupational, psychological, and medical function.

RLS pathophysiology is unclear, but evidence suggests widespread nervous system hyperexcitability including the basal ganglia, cortex, and spinal cord (SC), and increased sympathetic tone [6]. There are no widely accepted animal model systems for this disorder because animals cannot report urges. This uncontrollable sensation in the legs is highly irritating but typically not painful with patients commonly describing a deep sensation (not on the skin surface) that is temporarily relieved by voluntary leg muscle contractions [7]. Afferents from skeletal muscles and tendons provide proprioceptive information and travel via large, fast-conducting Aα fibers through the dorsal column of the SC [8]. RLS may be associated with SC hyperexcitability given abnormalities noted in the H-reflex [9–16]. Abnormal cortical excitability is suggested by reduced transcranial magnetic stimulation (TMS) motor thresholds, prolonged cutaneous silent periods, and symptom improvement, albeit temporarily, with repetitive TMS (rTMS) [9,17–24].

Neuromodulation for RLS was first investigated in a 1999 case series that utilized TMS [25]. Subsequent studies have examined more invasive modalities including spinal cord stimulation (SCS), deep brain stimulation (DBS), and peripheral neuromuscular stimulation. The gold standard measure of RLS severity is the validated International Restless Legs Syndrome Rating Scale (IRLS Score) where a reduction of ≥ 3 points is considered clinically significant [26, 27]. Evidence from these neuromodulation studies suggests potential effectiveness of circuit-based therapies that avoid lack of systemic side effects from oral medications. In this review, we summarize the history and findings from prior studies of neuromodulation therapies for treatment-refractory RLS and suggest pathophysiological mechanisms. We consider this review necessary to pinpoint treatment methodologies that have shown to be the least and most effective to increase the universal precision of treatment of the disorder (Fig. 1).

We performed a systematic review of the literature on 09/07/2024 in PubMed using PRISMA guidelines included in the supplement for the following search terms:

(“Restless legs syndrome“[tiab] OR “Restless legs syndrome“[MeSH] OR “RLS“[tiab]) AND (“Deep brain stimulation“[tiab] OR “DBS” [tiab] OR “Spinal cord stimulation“[tiab] OR “SCS“[tiab] OR “Neuromodulation“[Title/Abstract] OR “transcranial magnetic stimulation“[Title/Abstract] OR “TMS“[Title/Abstract])

120 articles were exported and screened independently by authors G.T., A.R. and E.W. using Covidence. All conflicts were resolved by M.H. Inclusion criteria were (1) original article that involved (2) neuromodulation of RLS, and that (3) suggested potential mechanisms for RLS. All reviews, commentaries, and other reports on the articles were excluded. We found 54 primary unique articles for the neuromodulation of RLS and summarized their findings in this review in U.S. English (Fig. 2). The review protocol was not registered in a database.

A risk-of-bias assessment was performed using the Methodological Index for Non-Randomized Studies (MINORS) scale (Table 1). Scores were assigned in 12 unique categories on a scale from 0 to 2 with a potential maximum score of 24. A higher resulting score is correlated with a lower risk of bias.

This review summarizes the current literature of neuromodulation of RLS. We compiled case reports, case series, retrospective cohort studies, and prospective trials that utilized neuromodulation modalities, including DBS, SCS, DRGS, TMS, TENS, tDCS, NPNS, and tVNS (Fig. 1). While less invasive methods, such as TMS, tsDCS, NPNS, and others did demonstrate successful symptomatic improvement, these results were transient with a decreased magnitude of effect compared to more invasive methods such as DBS and SCS. Overall, we found that DBS exhibits conflicting results with some studies showing improved symptoms while others demonstrated what seemed to be induced RLS symptoms. It has been suggested that this may be due to variation in the microanatomy within the STN and electrode placement or, alternatively, reduction in Levodopa equivalents after surgery leading to RLS symptom induction [52].

Contrarily, SCS demonstrated more consistent, promising results with greater magnitude of IRLS score improvement compared to the other modalities. In fact, some case reports show durable and complete resolution of RLS symptoms [33]. It is important to note that when reported, IRLS scores all decreased by more than 3 points, which represents the previously defined clinically significant improvement in RLS symptoms (Table 2). Unfortunately, the paucity of studies and heterogeneous reporting prevent meta-analysis of the treatment modalities necessary to provide substantial evidence supporting the use of SCS to treat RLS symptoms suggesting the need for multicenter, prospective trials (Table 3).

The pathophysiological mechanism behind SCS for RLS is unclear, but evidence suggests abnormalities in the dopaminergic, serotonergic, and adrenergic pathways with a resultant central nervous system (basal ganglia, cortex, and SC) hypersensitivity [67]. As patients experience circadian cycling of their symptoms, so too do these neurochemicals and neuroelectrophsyiological circuitries [68]. Specifically, these symptoms seem to be inversely coordinated with dopamine release [69]. Lack of dopamine in the basal ganglia, cortex, and SC results in a hypersensitive central nervous system state [70]. This hypersensitivity creates a dysfunctional sensory-motor integration condition in the SC and cortex (Fig. 3)[71]. This dysfunctional integration may help explain RLS symptoms with the inability to filter the “urge” percept at the SC and cortex. The lack of appropriate sensory-motor integration could additionally help explain the involuntary leg movements displayed by RLS patients. In fact, some studies provide evidence of improper SC network modulation with an abnormal H-reflex, the electrical equivalent of a deep tendon reflex, recorded from RLS patients [16]. If the SC is hypersensitive and unable to adequately filter the incoming signal, an involuntary leg movement may result.

The origin of the “urge” percept in RLS is one of much interest and speculation. Patients often report the “urge” originating from within the limbs, with relief from the visceral feeling only relieved by muscle contraction or getting up and walking. We hypothesize that the urge sensation may be generated by hyperactive Type Ia, Type Ib, and/or Type II fibers that originate in the muscle spindle and Golgi tendon organ [72]. These fibers synapse in the dorsal horn of the SC that directly assist in the stretch reflex (deep tendon reflex), relaxing of the muscle tension, as well as providing proprioceptive information to the central nervous system are important for maintaining posture and coordination [73]. The ascending information is carried through large, myelinated Aα fibers in the dorsal column. Should the dorsal horn of the SC be in a hypersensitive state (evening, low dopamine), it is reasonable to speculate that this information may not be properly processed and result in abnormal, involuntary movements. Additionally, the ascending fibers may carry this abnormal percept to the cortex where it could be interpreted by a hyperexcitable cortex with impaired sensory-motor integration as “urge.”

DBS may directly modulate or improve basal ganglia functional output to decrease cortical hyperexcitability [74]. TMS may increase local GABA in the hyperactive cortex [75]. However, TMS has the disadvantage of requiring repeated, in-clinic treatments due to its transient effects. TMS represents a valuable tool for further investigations into the pathophysiology and biological plausibility of effectiveness given its ability to directly evaluate cortical hypersensitivity. DRGS, TENS, tDCS, and NPNS may increase Aβ fiber activity, although evidence suggests that their effects on RLS symptoms are not as robust.

We believe that SCS currently represents the most intriguing, viable neuromodulation therapy for RLS that demands further investigation. We hypothesize that SCS could interrupt this “urge” signal in multiple ways. First, it is possible that the SCS is directly modulating the signal traveling in the Aα neurons of the dorsal column that could block the “urge” percept prior to it arriving at higher brain centers [76]. Additionally, the gate control theory of pain that underlies the current biological principal theory of SCS effectiveness dictates that the dorsal column Aβ fibers that carry light touch information, through their normal orthodromic conduction, synapse upon an inhibitory interneuron in the spinal cord that further synapses within the dorsal horn [77]. Stimulation of Aβ fibers, and, thus, inhibitory spinal interneurons, leads to increased GABA within the dorsal horn and increased inhibition of incoming sensory signals, potentially including hyperactive proprioceptive signals [78]. With its biological plausibility, preliminary effectiveness demonstrated in prior reports, and long safety record, we believe that SCS represents a strong risk to benefit ratio for further prospective investigations. It is also important to note that SCS is less invasive than DBS, not requiring penetration of the dura. It is fair to speculate that patients may be more receptive to this neuromodulation modality even without treatment-refractory symptoms, similar to PD patients undergoing DBS with the goal of medication reduction.

Multiple limitations exist in this review that raise questions regarding the study’s quality. Publication bias exists as it is intrinsic to systematic reviews. This protocol was also not registered in a database before beginning the review, which poses a risk of duplication of study type and topic. Heterogeneity among the studies included in this review is a critical limitation as different protocols and different patient populations produce results that may not be comparable among each other. The statistically significant effects of each neuromodulation modality were unable to be evaluated due to the heterogeneity in follow-up, outcomes reported, and lack of data granularity limiting the article to observational anecdotes and summaries of the results. Despite this, the degree of RLS score recovery seems to positively correlate with the degree of invasiveness of a given procedure among the studies included in this review. It is important to acknowledge that expectation bias may exist within increasingly invasive neuromodulatory treatment; increased invasiveness of a treatment may lead to higher expectations for resolution of symptoms due to increased need for hardware, software, increased procedure time, increased procedure complexity, and more, which can produce a placebo effect. The methodological quality of the studies included in this review also contain risk of bias; While some of the studies were RCTs, many were also case series or case reports which have a higher risk of bias due to their uncontrolled nature whereas RCTs control for as many potential confounding variables as possible [79]. Lastly, many studies investigating neuromodulation to treat RLS symptoms include patients that have RLS concomitant to another pathology with very few patients having RLS as their primary indication for neuromodulation. Many studies included in this review examine the effects of neuromodulation on secondary RLS as opposed to primary RLS. This is important to acknowledge as comorbid RLS differs to primary RLS. This may result in limited applicability or translatability of this review’s conclusions and may result in inaccurate generalizations under the assumption that therapies are directly targeting RLS symptoms rather than resolution of RLS symptoms being a secondary outcome of treating the concomitant disorder that is not RLS. Studied observing effects of neuromodulation in patients with primary RLS should be conducted.

Without rigorous statistical analyses from prospective trials of the multiple neuromodulatory techniques, discerning the most effective form of neuromodulation for an individual patient remains challenging. In treatment-refractory PD, treatment selection is based upon several factors, such as concomitant sequelae, disease burden, and invasiveness of the procedure. In many cases, patients with debilitating PD who have comorbid RLS undergo DBS treatment primarily for their Parkinson’s symptoms. More intriguing, however, is the patient with primary idiopathic RLS for which little data in the more invasive treatments (i.e., SCS, DBS) exist. Comparison across the newer, less invasive trials may be useful for less severe cases of treatment-refractory RLS, or for use supplemental to medical treatment. Based on the literature available, the degree of RLS score recovery correlates with the procedure’s invasiveness, with the highly invasive modalities including DBS and SCS providing the most symptom relief. Furthermore, SCS has been shown to completely reduce subjective intensity of RLS symptoms for up to 2.5 years and, overall, demonstrate more consistent, promising results with greater magnitude of RLS symptoms resolution in comparison to other neuromodulatory methods. We propose utilizing the IRLS, CGI-I, and objective measures of sleep and leg movements, preferably in the ambulatory setting, as key measures in these trials. The sequential trials and analyses must control for treatment modalities, severity of disease, invasiveness of procedure, and anatomical location/optimization of the invasive procedures to develop guidelines and personalized treatment plans. Further prospective, controlled, multi-center, multi-modality trials studying the feasibility, safety, and efficacy of each neuromodulatory modality are necessary due to the heterogeneity of the data and methods of each study, and to provide significant evidence to guide treatment of treatment-refractory RLS using neuromodulation.