Several potential mechanisms have been proposed by which WED or PLMS could cause CVD, including insufficient sleep time, mood disturbance, and sympathetic overactivity related to PLMS. Insufficient sleep may be caused by WED either because the waking sensory symptoms delay sleep onset (at the start of the night or after an awakening) or because the associated PLMS result in arousal/awakening from sleep. In the large REST study, a detailed epidemiologic investigation of over 1000 WED sufferers, 75% of people with WED reported sleep disruption (manifesting as difficulty falling asleep, difficulty remaining asleep, disturbed sleep, insufficient sleep, or some combination of these) [2]. Problems with sleep related to WED were rated as the most distressing part of the WED experience by nearly 40% [2]. Sleep deprivation has been consistently associated with features of CVD [3]. There are multiple mechanisms by which sleep deprivation might cause CVD that could be at play in WED patients, including obesity/metabolic dysregulation [4], inflammation [5, 6], hypertension [7, 8], alterations of the stress response [9], and so on. However, it is worth noting that not all studies that have found an association between WED and CVD have shown that it is mediated by sleep time. In particular, in their prospective study, Li et al. found an increase in incident CVD events in chronic WED sufferers that was independent of short sleep time [10].

WED is associated with impaired mood, although not necessarily fully causal to mood symptoms [11, 12]. Supporting the idea that WED may contribute directly to severity of mood symptoms, treatment of WED may improve severity of mood disturbance [13, 14]. Mood disorders are themselves associated with, and may predict incident development of, CVD [15, 16]. Thus WED-induced mood disturbance might be the intermediary by which WED causes CVD [1, 17]. Alternatively, to the extent that WED or PLMS serve as a physiologic stressor, they might in turn cause CVD by this mechanism. Some authors have suggested that sympathetic overactivity related to WED symptoms might impair baroreceptor modulation and thus increase blood pressure, akin to the effects seen in sleep deprivation or sleep apnea [18]. Finally, although small early studies did not demonstrate any relationship between cortisol levels and WED [19, 20], a larger study has recently confirmed that WED patients have increased nocturnal urinary cortisol levels, which might contribute to CVD [21]. At this point, it is unclear whether cortisol abnormalities result in WED symptoms or the reverse, but it is biologically plausible that the distress from WED or associated sleep disruption results in elevations in cortisol [21].

Separate from these possibilities, much of the research assessing a possible causal link between WED/PLMS and CVD has focused on the interplay between WED, PLMS, repetitive nocturnal elevations in blood pressure and heart rate occurring with PLMS, and hypertension (HTN). Results from cross-sectional studies evaluating the association between WED and hypertension have been quite mixed, with positive [18, 22–24], null [25, 26], and inverse [27, 28] associations all reported. A recent meta-analysis reported 10 of 17 studies investigating a potential link between HTN and WED showing a significant positive association [1]. Some of these studies have suggested a dose effect of WED symptom frequency or severity on the likelihood of hypertension [23], such that the distribution of symptom severity across different study populations might account for the discrepant study findings [29]. The relationship might be bidirectional, as suggested by a study of two separate cohorts, one of which demonstrated that hypertension at baseline predicted the subsequent development of WED [30]. Importantly, studies assessing a WED and HTN association have typically not measured PLMS, which might drive the association with hypertension more strongly than the WED symptoms themselves (and, to the extent that asymptomatic PLMS may be present before the onset of WED symptoms, might obscure a temporal relationship between hypertension and WED). WED severity and PLMS are only modestly correlated (Pearson’s correlations r = 0.22–0.46) (52–54), so without direct measurement of PLMS, it may be hard to accurately assess the relationship of WED and hypertension. This is suggested by studies by Billars [31] and Espinar-Sierra [32], which demonstrated dose response relationships between PLMS and HTN. However, considering only those patients with grade I hypertension (i.e., mild HTN) versus controls results in no difference in PLMS between groups [33], suggesting that a certain degree of PLMS or HTN severity may be necessary to drive the association. Among patients with heart failure, PLMS do not appear to be positively associated with prevalent hypertension [34, 35], and may even be negatively associated [36].

Individual leg movements during sleep are accompanied by transient increases in heart rate and blood pressure in patients with WED as well as in those with PLMS but without any sleep symptoms [37–43]. Whether this cardiovascular activation is a result of PLMS themselves is an unresolved question. Most, although not all, studies assessing heart rate changes associated with limb movements have shown a change in heart rate that begins before the onset of movement as measured by surface EMG of the anterior tibialis muscle [39–43]. Although this suggests that the leg movements must not be causing the autonomic activation, there are several caveats of importance. First, although the default muscle with which to study PLMS is the anterior tibialis, only 32–53% of PLMS in WED patients begin in the anterior tibialis muscle [44, 45], so a careful evaluation of the relative onset of movement and cardiovascular activation will require monitoring of multiple muscle groups. Second, other features of limb movements do appear to affect the amplitude of autonomic activation, suggesting that the movement at a minimum modifies the autonomic response. Specifically, movements that are bilateral, have a shorter intermovement interval or are more frequent, or have a longer movement duration are associated with greater autonomic activation [42, 46, 47]. Changes in heart rate and blood pressure are also known to occur in association with apneas, and the magnitude of these changes is increased in those apneas associated with a leg movement relative to those apneas without leg movement [48], suggesting an additive effect from the leg movement.

Several alternate explanations for the coexistence of movement and autonomic activation may be hypothesized. First, the increase in heart rate and blood pressure might be a physical consequence of the movement itself, for example, due to increased venous return caused by movement [49]. However, voluntary leg movements during wakefulness that simulate PLMS are not associated with the same pattern of autonomic activation (either heart rate or blood pressure), suggesting this explanation does not underlie the observed relationship [39, 49].

Alternatively, there may be a common pacemaker that drives both the autonomic activation and leg movements. Most studies have suggested that the presence of an electrographic arousal, which may also begin before the measured onset of the leg movement, increases the degree of autonomic activation [39, 41, 50, 51], suggesting the influence of a common pacemaker that may influence movement, cerebral arousal, and autonomic arousal [40]. As experimentally induced arousals do not trigger leg movements, it does not appear that either phenomenon (PLMS or arousal) is necessarily causal to the other [52, 53]. Furthermore, treatment effects show a clear separation between cortical arousal and leg movement, such that treatment of PLMS does not remove repetitive arousals and treatment of arousals does not remove PLMS [37, 54–57]. The fact that the magnitude of autonomic activation associated with leg movements depends on whether there is comorbid WED [37, 43] suggests that such a pacemaker may also be affected by underlying neurologic substrate. Arguing against the pacemaker argument is the fact that, while limb movements and autonomic activity are tightly associated, the presence or absence of a cortical arousal with a limb movement is much more variable; a common pacemaker would be expected to cause all three phenomena consistently [47].

Finally, increased sympathetic activation might cause PLMS [47]. This is particularly suggested by the temporal association of autonomic activation and PLMS [47], and would explain the association between conditions of increased sympathethic activation, such as heart failure, and increased PLMS [36, 58, 59]. However, the presence of normal autonomic tone, at least as measured by heart rate variability metrics, between individual PLMS and during wakefulness in otherwise healthy subjects with PLMS argues somewhat against this possibility [43, 53]. Further, if increased autonomic activation were causal to PLMS, patients with autonomic failure would be expected to have fewer PLMS than patients with normal functioning autonomic nervous systems; a small study of patients with multiple systems atrophy or primary autonomic failure compared to healthy controls did not support this hypothesis, as PLM indices were no different [60]. As has been pointed out elsewhere, it is important to keep in mind that these directions of causality are not necessarily mutually exclusive [53], and multiple directions of relationship may exist.

Regardless of direction(s) of causality, PLMS associated with frequent blood pressure elevations will presumably result in a nondipping blood pressure pattern, as frequent blood pressure elevations will result in a higher average nocturnal blood pressure. Indeed, patients with WED (the vast majority of whom will also have PLMS [61]) have been shown to be more likely to demonstrate a nondipping pattern (OR 1.96) [18], and nocturnal dipping is reduced in those WED patients with frequent PLMS compared to WED patients with fewer PLMS [62]. Nondipping itself is associated with CVD [63]. The increase in CVD risk seen with nondipping may be related to the development of structural heart disease, thought to be related to increased hemodynamic stress from the lack of dipping or increased sympathetic activity, or from carotid intima-media thickening [35, 62, 64, 65]. Supporting this, several studies have found an association between severe PLMS and LVH. Patients undergoing polysomnography who have PLMS > 35/h, regardless of WED status, have an adjusted OR for LVH of 2.45 (1.67–3.59) [35]. WED patients with end-stage renal disease on hemodialysis with frequent PLMS (>25/h) have a higher left ventricular mass and left ventricular diameter during diastole, which in turn predicts mortality in these patients [62].

More broadly, a number of studies have evaluated for the effect of PLMS on future development of CVD, cardiovascular mortality, and all-cause mortality. Studies of PLMS have demonstrated that among non-hypertensive elderly men, frequent PLMS predict incident vascular disease, although this relationship was not significant for hypertensive men [66]. Among patients with renal disease, frequent PLMS consistently predict increased mortality [67–69], and the same is true of patients with heart failure [36]. These studies do not directly answer the question of causality between sympathetic activity and PLMS, but do underscore the substantial clinical relevance of this question. If the primary driver for these events (leg movements, autonomic activation, and cortical activation) can be identified and modified, there is real potential for improved clinical outcomes.

It is certainly possible that WED or PLMS are caused by CVD itself. One of two cohorts evaluating this question demonstrated a significant association between myocardial infarction and the subsequent development of WED (OR 2.04); the other cohort found no significant relationship between the two, with an OR of 0.80 [30]. It is unclear how CVD per se might cause WED or PLMS, but it is plausible that associated vascular diseases might do so. That is, either cerebrovascular disease or peripheral arterial disease, both of which are tend to associate with CVD, might cause WED symptoms or PLMS.

Among patients evaluated for WED following an ischemic stroke, anatomical localization is a strong predictor of whether or not WED will be present. That is, patients with subcortical or brainstem (especially basal ganglia, corona radiata, and pons) strokes are much more likely to have WED than patients with cortical strokes [116]. In addition to providing insight into the anatomy of WED itself, this observation argues that cerebrovascular disease, at least when it affects particular brain regions, may cause WED. However, another study evaluating stroke and WED did not find an association between stroke location and WED [117]. Patients with a history of stroke are more likely to have PLMI > 5 than those without a history of stroke (out of 80 total subjects) [118].

Historically, WED has been conceptualized as a disorder of vascular function related to impaired peripheral blood flow or sympathetic dysfunction [74, 119], including during Karl Ekbom’s early descriptions of the disorder [120, 121]. Two small controlled trials of clonidine showed a benefit on WED symptoms and clonidine is considered as an option for the treatment of WED [122, 123]. Ware showed that PLMS patients have blunted peripheral pulses in response to certain maneuvers suggestive of sympathetic overactivity, and vasodilatory agents decreased number of PLMS [124]. This has been confirmed in patients with heart failure; those on vasodilators are less likely to have PLMI > 10 than those not on vasodilators [125]. Thus, patients with peripheral artery disease might be more likely to develop WED. Speaking somewhat against this direction of causality is the young average age of onset of WED symptoms (with bimodal peaks in mid-20s and mid-40s) [126]. However, to the extent that CVD is a disorder marked by years or decades of subclinical damage prior to first clinical manifestations [127], temporal relationships may be somewhat obscured.

Depending on the direction of causality, treatment of WED/PLMS might result in prevention of CVD events. That is, if WED symptoms or PLMS are causal to CVD, their treatment might represent a novel opportunity to intervene to prevent the development or progression of CVD. At present, this is speculative, as scant data are available assessing this question.

In the single prospective study to date designed specifically to address the question of whether WED/PLMS treatment affects CVD risk parameters, Manconi et al. performed a night of polysomnography before and after beginning treatment of WED with pramipexole [43]. As expected, pramipexole decreased the frequency of PLMS. Importantly, pramipexole also decreased the magnitude of the heart rate variability increase accompanying the remaining leg movements [43]. Although this clearly cannot be extrapolated to suggest that treatment of PLMS will have beneficial effects on CVD events, it at least suggests a mechanism by which such benefit might occur. More recently, a retrospective chart review has suggested that in those patients with PLMI > 35/h (but not those with less frequent PLMS), use of dopaminergic treatment was associated with lower progression rate of atrial fibrillation (11.6% progression in treated group vs. 32% in the untreated group) [34]. Somewhat tempering enthusiasm for use of dopamine agonists in patients with WED or PLMS for cardioprotection, at least in the absence of additional data, is the United States Food and Drug Administration’s 2012 safety announcement regarding a possible increase in heart failure risk with pramipexole, requiring further evaluation to establish or refute [128].

Gabapentin enacarbil and other alpha-2-delta ligands are commonly used for WED, but we are not aware of any data specifically addressing the question of CVD risk and WED treatment for this medication class. Less commonly, benzodiazepines may be used for WED. A single case report, notable for its role as the first report of elevations in blood pressure associated with individual periodic leg movements during sleep, demonstrated reduction in arousals associated with PLMS after administration of temazepam. However, elevations in systolic blood pressure were persistent despite temazepam (21.8 mmHg before vs. 22.7 mmHg after temazepam) [56].