Review of the Relationship of Restless Legs Syndrome and Periodic Limb Movements in Sleep to Hypertension, Heart Disease, and Stroke

Association of RLS with Hypertension and Heart Disease

In a study by Ulfberg et al.,¹⁵ 4000 men aged 18 to 64 years in central Sweden were mailed questionnaires that adhered to the 4 criteria for RLS developed by the International RLS Study Group. The prevalence of RLS was 5.8% in this sample after adjusting for eligibility, incomplete questionnaires, and nonresponder rates. Statistical adjustments were also made for possible confounding factors such as age, witnessed apnea, smoking, and alcohol consumption. RLS sufferers were more likely to report hypertension (odds ratio [OR] 1.5; 95% confidence interval [CI] 0.9-2.4) and heart problems (OR 2.5; 95% CI 1.4-4.3). Some of the heart problems were presumably secondary to the hypertension, since they were correlated by a factor of 0.4 (P < 0.01). Correlation, however, does not imply causality, as suggested by the larger OR for heart disease, compared with hypertension. Despite the large size of this study, its limitations included the failure to examine the frequency of RLS symptoms and the potential for Type I error, given that the patients were not interviewed or examined to exclude RLS mimics.

In a study by Ohayon and Roth,¹⁶ individuals aged 15 years and older (51.3% women) from 5 European countries (UK, Spain, Germany, Portugal, Italy) were interviewed by telephone about RLS symptoms. The 18,980 individuals represented 80.4% of the population originally selected for interview. The definition of RLS in this study was from an older edition of the International Classification of Sleep Disorders, which was not the most up-to-date definition for RLS from the International RLS Study Group.^1,17–19 Patients were said to meet criteria for RLS if they had: (a) a complaint of unpleasant sensations in the legs at night or difficulties in initiating sleep; (b) disagreeable sensations of “creeping” inside the calves, often associated with general aches and pains in the legs; and (c) discomfort that was relieved by movement of the limbs. The prevalence of RLS was 0.9 % for daily symptoms, 3.1% for symptoms that occurred several times per month, and 6.4% for symptoms that occurred at least 1 night per week. After statistical corrections, the OR for the association of hypertension with RLS was 1.36 (95% CI 1.14-1.61; P < 0.001) and for heart disease with RLS was 1.41 (95% CI 1.06-1.88; P < 0.05).

Phillips et al.²⁰ examined the data from the 2005 National Sleep Foundation Sleep in America, a random-dial telephone poll of 775 women and 731 men with mean age 49 years. They reported that RLS was present for at least a few nights per week in 9.7% of respondents (8% of men and 11% of women). For further analyses, statistical adjustments were made for age, sex, and an existing sleep disorder. Those with a history of hypertension were more likely to endorse RLS symptoms (P < 0.05), and there was a trend for those with heart disease to endorse RLS symptoms (P < 0.10). In this study, patients were considered as affected by RLS if they answered the following 2 questions in the affirmative and had symptoms at least a few days per week: “In the past year, according to your own experiences or what others tell you, how often did you have unpleasant feelings in your legs like creepy, crawly or tingly feelings at night with an urge to move when you lie down to sleep?” and “Would you say these feelings in your legs are worse, about the same as, or better at night or in the evening compared to other times of day?” This study did not include the “improvement by activity” criterion for RLS in the survey, and subjects did not undergo clinical interview or examinations.

Winkelman et al.^21,22 performed 2 separate analyses of data from large databases. An initial analysis of 2821 individuals from the Wisconsin Sleep Cohort noted cardiovascular disease to be more prevalent in individuals with daily RLS, compared with those without RLS (OR = 2.58, 95% CI 1.38-4.84).²¹ A more recent and in-depth cross-sectional survey of 1559 men and 1874 women (mean age of 67.9 years) enrolled in the Sleep Heart Health Study identified putative RLS in 3.3% of men and 6.8% of women.²² RLS was defined by positive responses on a self-administered questionnaire containing the 4 essential RLS criteria established by the International RLS Study Group, with at least moderately distressful symptoms occurring at least 5 times per month.¹ Coronary artery disease was determined by self-report of physician-diagnosed angina, myocardial infarction, or coronary revascularization procedure. Total cardiovascular disease included coronary artery disease or a history of physician-diagnosed stroke or heart failure. The relationship between RLS and coronary artery disease and cardiovascular was examined by logistic regression with adjustment for confounding variables. The adjusted OR for coronary artery disease was 2.22 (95% CI 1.40-3.53) and, for cardiovascular disease, was 2.38 (95% CI 1.55-3.65) for subjects with RLS, compared with those without RLS. The associations of RLS with coronary artery disease and cardiovascular disease were only apparent in those with RLS symptoms at least 16 times per month and were more robust in those with severe, as opposed to moderate, RLS. This study was conducted without a personal patient interview or examination. Unlike the positive relationship found between RLS and hypertension in the Ulberg et al., Ohayon and Roth, and Phillips et al. studies, the 2 studies by Winkelman et al. and a study by Hogl et al. found no relationship between RLS and hypertension, and, in the MEMO study by Rothdach et al., there was even an inverse relationship.^{15,16,20–24}

In summary, the results of epidemiologic studies suggest a possible relationship between self-reported RLS symptoms and daytime hypertension and are more consistent in pointing to a relationship between RLS and cardiovascular disease.

Association of PLMS With Heart Disease

Pulse Rate

A number of studies have demonstrated a temporal coincidence between PLMS and pulse rate elevations. A variable portion of these electrocardiographic events are accompanied by cortical arousal. In a study by Winkelman, heart rate was recorded for 10 cardiac cycles before and after PLMS in 8 patients. Leg movements during wakefulness were performed by 4 healthy control subjects as a control condition. A significant rise in heart rate above and beyond the control condition was seen with PLMS. Although heart rate was 10% to 40% higher when electroencephalographic (EEG) arousals accompanied the PLMS, these differences did not reach statistical significance. The authors concluded that pulse rate elevations that accompany PLMS occur whether or not there is an accompanying EEG arousal. Heart rate began to increase 3 cardiac cycles before the onset of PLMS and peaked at 4 cardiac cycles after the onset of PLMS. There was then a decline to below the baseline for cardiac cycles 8 through 10. The average pulse rate was 62 at baseline and 68.2 at peak; the nadir was 59.3 at cardiac cycles 8 through 10.³¹

Sforza et al. found a similar tachycardia associated with PLMS in 10 patients with RLS. PLMS were associated with visible EEG microarousals lasting longer than 3 seconds in one third of all PLMS. Accelerations in heart rate were greatest with a visibly discernible EEG microarousal. This was confirmed by spectral analysis of the EEG, which also revealed brief bursts of theta and delta activity with PLMS in which no visible microarousals were observed. From these analyses, the authors postulated that a hierarchy of arousals exists, starting with autonomic arousal and progressing to cortical arousal, as manifest in delta, then theta, then alpha activity, to full awakening.³²

A key question is whether the sympathetic activation accompanying PLMS is specific to PLMS or is also characteristic of random nonperiodic movements in sleep. Guggisberg et al.³³ answered this question in the affirmative by demonstrating that the sympathetic activation accompanying PLMS is greater than for all movement types, as measured by heart rate variability spectra. Sympathetic activation accompanying PLMS is also evident in the setting of coincident sleep disordered breathing and exceeds that observed with apneic events unaccompanied by somatic activation.³⁴ This line of inquiry emphasizes that PLMS are likely a trait unique to the individual patient rather than a consequence of a coincident alternate phenomenon such as sleep disordered breathing. This view finds further support in analyses of the night-to-night variability in PLMS expressivity within and between subjects³ and the discovery that RLS genes confer their risk by way of PLMS.⁸

Blood Pressure

The above observations on PLMS-related changes in heart rate have been extended to another manifestation of heightened sympathetic discharge, viz., blood pressure. Ali et al. demonstrated that blood pressure elevations on the order of 23% accompanied PLMS in a single patient with narcolepsy. These rises were evident in both systolic and diastolic measures, and there was a coincident acceleration in the pulse rate as well. Ali et al. pointed out that these rises were of a magnitude similar to those observed in obstructive sleep apnea.³⁵

Pennestri et al. from Montplaisir's group confirmed and extended this single observation by documenting blood pressure elevations coincident with PLMS in 10 patients with RLS (systolic blood pressure average 22 mm Hg and diastolic blood pressure average 11 mm Hg over baseline). The observed increases were greatest in patients in whom EEG microarousals accompanied the PLMS. The magnitude of the pressure elevations was greatest in elderly patients and in those with a longer duration of RLS symptoms. Increases in pulse rate also accompanied the PLMS with or without EEG arousal. Periodic limb movements while awake also showed an increase in blood pressure over baseline, but this was not as great as the increases seen in PLMS with or without EEG arousal.³⁶

These results were independently confirmed by Siddiqui et al. in a study of 8 patients with RLS. To control for the possibility that movement alone could result in the accompanying changes in pulse and blood pressure seen with genuine PLMS, prior to bedtime patients performed simulated PLMS. Although none of the patients had a history suggestive of sleep apnea, in some patients some apneas were encountered for analysis. Care was taken to assess pulse rates and blood pressure that accompanied any leg movements occurring at the termination of an apnea. The magnitude and nature of diastolic and systolic blood pressure elevations and the rank order of autonomic activation were similar to those observed by Pennestri et al. For systolic blood pressure, the results were as follows: the highest blood pressure was seen with respiratory-related limb movements (mean, 18.9 mm Hg above baseline), the next highest was with PLMS with microarousals (mean, 16.7), and was similar between PLMS without microarousal (mean, 11.2 mm Hg) and periodic limb movements while awake (mean, 11.7 mm Hg), which were both higher than with the simulated PLMS (mean, 3.2 mm Hg).³⁷

In summary, autonomic arousals accompany PLMS and manifest as large rises in pulse rate and blood pressure.

Systemic Hypertension and PLMS

In a study by Espinar-Sierra et al.,³⁸ 91 subjects with essential hypertension were studied polysomnographically for the presence of PLMS. Eighteen percent had PLMS, and the prevalence of PLMS was proportional to the severity of the hypertension. Patients with grade III hypertension had a 36.4% prevalence of PLMS, and patients with grade I and II hypertension had a 13% prevalence of PLMS. This relationship was independent of age, sex, obesity, smoking, alcohol consumption, apnea severity, and use of antihypertensive medications.

Billars et al. investigated the association between hypertension and PLMS in a much larger study of 861 patients with self-reported RLS symptoms. The likelihood of hypertension increased with PLMS severity, being present in slightly more than 60% of subjects with more than 50 PLMS per hour of sleep. The risk for having hypertension was twice as high for those with a PLMS index greater than 30 (OR = 2.26; 95% CI, 1.28-3.99) after controlling for known contributors to hypertension, such as body mass index and age.³⁹ These results could reflect decrements in total sleep time that one would naturally assume to accompany RLS, although, when included in the multivariate logistic regression, self-reported hours of sleep and category of antihypertensive medications did not take away from the robustness of the association between PLMS and hypertension status (personal observation). Because this study employed quantitative assessments of PLMS as the independent variable of interest, it infers that PLMS are causally related to medically diagnosed hypertension and not the reverse. It is also important to emphasize that the OR for hypertension with more than 30 PLMS per hour is of the same order of magnitude as that most often cited for hypertension and sleep disordered breathing (eg, OR = 2.89 for an apnea-hypopnea index > 15).⁴⁰

In summary, PLMS are associated with greater daytime hypertension and the reverse is also true, ie, daytime hypertension is also associated with a greater number of PLMS.

Sympathetic Overactivity in RLS and PLMS Leading to Heart Disease and Stroke Through the Mechanism of Hypertension

Although there are some inconsistencies, the literature summarized above argues that patients with RLS, PLMS, or both RLS and PLMS are at increased risk of developing hypertension, heart disease, and stroke. One of the leading hypotheses for the pathogenesis of RLS and PLMS is that there is a dopaminergic deficit in these 2 related conditions. The strongest evidence for this hypothesis is the efficacy of dopaminergic therapy for both conditions.^45,46 There are several dopaminergic pathways in the brain, including the striatonigral, mesolimbic, mesocortical, and tubero-infundibular. The A11 dopaminergic diencephalospinal pathways innervate preganglionic sympathetic neurons and the dorsal horn in the spinal cord. One hypothesis gaining substantial support suggests that hypofunction of the A11 diencephalospinal pathway leads to enhanced sympathetic outflow to the periphery, inclusive of somatic muscle fibers. As a result of this increased sympathetic activity, altered activity in afferents from the muscle fibers back to the spinal cord in turn may heighten spinal sensory signaling and thereby trigger RLS symptoms. This afferent input is also normally dampened by A₁₁ innervation at the level of the dorsal horn. Thus, spinal dopaminergic hypofunctioning theoretically gives rise to a positive feedback loop whereby disinhibition manifests as both heightened sympathetic efferent drive leading to altered sensation as well as insufficient suppression of increased sensory afferent drive (ie, aberrant sensory equivalents of RLS complaints) (see Figure 1). The A₁₁ diencephalospinal pathway is also the only dopaminergic pathway in which a hypofunction might be reasonably expected to lead to sympathetic hyperactivity and hypertension with a subsequent increased risk for developing heart disease and stroke.⁴⁷ Interestingly, vasodilators, one of the original therapies for RLS shown to be effective by Ekbom, also act at the dorsal horn in addition to the periphery.⁴⁸ The A₁₁ hypodopaminergic theory is further strengthened by demonstration that lesioning of this pathway leads to restlessness in rats that is reversible with the dopamine agonist pramipexole.⁴⁹ It is well documented by cerebrospinal fluid analyses, magnetic resonance imaging, and autopsy studies that iron deficiency in the central nervous system plays an important role in modifying expression of RLS. Iron deficiency contributes to impairments in dopamine signaling, and it may be that the negative impact of iron deficiency upon RLS symptoms plays out at the spinal level (see Figure 2).⁵⁰

As presented in detail above, PLMS accompanied by EEG arousals are associated with significant transient rises in pulse rates and blood pressure.^36,37 These elevations in pulse rates and blood pressure are presumably sympathetically mediated, although the possibility that reduced parasympathetic activity may play a role has not been excluded. It is attractive to postulate that these regular, repetitive blood pressure elevations at night “prime the pump” for the development of daytime hypertension and subsequent stroke and cardiovascular disease. It will be important to differentiate whether it is the PLMS themselves or the arousals associated with them that are more intimately linked to elevations in blood pressure and pulse rate. Arousals from sleep in normal individuals without PLMS are also associated with brief coincident elevations in blood pressure and pulse rate.⁵¹ The number of spontaneous arousals during sleep in normal individuals without PLMS is also a predictor of daytime blood pressure and pulse rate. The higher the arousal rate, the higher the daytime pulse rate and blood pressure.⁵² It could therefore be that the arousals associated with the PLMS, rather than PLMS themselves, are responsible for the priming of daytime hypertension. On the other hand, blood pressure elevations that are nearly as high as those seen with PLMS accompanied by arousals are present with PLMS unaccompanied by EEG arousals.^36,37 It then follows that both the PLMS and the arousals associated with them are intimately connected with the accompanying changes in pulse and blood pressure. Further work needs to be done to determine if autonomic priming of daytime hypertension truly occurs in RLS, and whether it, in turn, is causal to cardiovascular disease and stroke.

Sympathetic Overactivity Associated With RLS/PLMS Leading to Heart Disease and Stroke Without Daytime Hypertension as an Intermediary

One of the inconsistencies with the aforementioned hypothesis is that the 2 studies by Winkelman et al. demonstrate a relationship between RLS and cardiovascular disease but not between RLS and hypertension.^21,22 The study by Högl et al. found no relationships between RLS and hypertension,²³ and the study by Rothdach et al.²⁴ even suggests an inverse relationship. This is in contradistinction with the results of studies by Ulfberg et al.,¹⁵ Ohayon and Roth,¹⁶ and Phillips et al.,²⁰ in which positive associations were noted between RLS and hypertension. Thus, a relationship between RLS and hypertension is not firmly established. This could simply be the result of a false negative in the Winkelman et al., Hogl et al., and Rothdach et al. studies, since the entry point for the studies was a diagnosis of RLS by questionnaire and not PLMS by objective polysomnography. Periodic rises in blood pressure and daytime hypertension are intimately related to PLMS. Because only 80% to 90% of adult patients with RLS have PLMS, the relationship of the RLS/PLMS complex to hypertension could be diluted by a lack of association between hypertension and RLS without PLMS. If cardiovascular disease and stroke are not caused by hypertension in RLS, what potential alternate explanations could there be for this relationship? Winkelman et al. suggest that the transient elevations in blood pressure associated with PLMS could cause cardiovascular disease by increasing the 24-hour blood pressure profile without producing changes in daytime blood pressure. It is known that hypertension is a risk factor for atherosclerotic plaque formation and plaque rupture.^53,54 Winkelman et al. posit that rises in pulse rate and blood pressure associated with PLMS also could cause cardiovascular disease by way of an increased risk of atherosclerotic plaque formation and rupture.²²

Comorbidities Frequently Associated With RLS/PLMS Leading to Cardiovascular Disease and Stroke

Winkelman et al.²² suggest that confounding factors associated with RLS/PLMS, eg, anemia and renal failure, could account for increased cardiovascular and cerebrovascular morbidity. Sleep apnea may also be 1 of these confounding factors. The literature we have quoted previously indicates that PLMS are very common in patients with CHF, and many of the patients with CHF in these studies also had sleep apnea.^25,27,29,30 It is well known that sleep apnea commonly accompanies CHF and that sleep apnea is a risk factor for cardiovascular morbidity and stroke.^27,55 In support of the possibility that RLS/PLMS may predispose to a person to having sleep apnea, we mention 2 studies that indicate that dopamine agonist treatment of RLS/PLMS leads to a coincident significant decrease in sleep disordered breathing.^56,57 Whether these effects are clinically significant is unknown because subjects with RLS and significant sleep disordered breathing (eg, respiratory disturbance index > 10/h) were excluded from both of these studies. This clinical experience, albeit limited, suggests that brain mechanisms modifying expression of PLMS interact at some level with those responsible for respiration. As 1 example of how changes in the periphery might lead to cardiovascular and cerebrovascular disease by way of RLS, we mention the study of Haxiu et al.,⁵⁸ who demonstrated that sciatic nerve stimulation can modify upper airway tone. This represents a plausible functional anatomy whereby the abnormal leg sensations of RLS might contribute to sleep disordered breathing and its downstream consequences. These putative associations are of particular relevance to RLS, since radiculopathy and peripheral neuropathy are oft-cited causes of ostensibly “secondary” RLS.⁵⁹ RLS could also produce cardiovascular disease, hypertension, and stroke via decrements in sleep time, which show a modest but significant association with these outcomes in patients without RLS.^60–63

Interactions among comorbidities associated with RLS (insomnia, obstructive sleep apnea, anemia, renal failure) and sympathetic overactivation and arterial disease^16,59 represent additional avenues through which RLS might promulgate cardiovascular disease and stroke (see Figure 2).

Heart Disease, Stroke, or Peripheral Vascular Disease Leading to RLS/PLMS

Cardiovascular disease could alternatively lead to RLS and PLMS via vascular pathologies in the central nervous system or the periphery.²² As we have discussed, RLS, PLMS, or both can appear immediately after the onset of acute clinical stroke,^41–43 and PLMS can occur coincidentally with sleep disordered breathing encountered in the setting of CHF.^25,27,29,30 Hypoxia associated with apneic events might also lead to impaired synaptic release of spinal dopamine, as has been shown for nigrostriatal dopaminergic terminals after exposure of newborn rats to intermittent hypoxia.⁶⁴ Additional insights into pathophysiologic networks linking RLS/PLMs to cardiovascular consequences may derive from knowledge of the genes conferring susceptibility to RLS/PLMS.^8,9 Among its many functions as a homeobox gene, for example, Meis1 is critical to vascular patterning/endothelial cell development in both mice and zebra fish.^65,66 Mice that are molecularly engineered to lack Meis 1 exhibit vascular dysplasia that leads to lethal hemorrhagic events in the embryo.⁶⁵ Venous, as opposed to arterial, pathology may also play a role in the pathogenesis of RLS. Thirty six percent of 174 consecutive patients presenting to a phlebology practice had RLS, as opposed to only 19% of 174 matched control subjects.⁶⁷ In 2 separate studies, endovenous laser ablation or sclerotherapy improved RLS symptoms in a large percentage of patients with duplex-proven superficial venous insufficiency.^68,69 These findings will have to be validated further in blinded studies. LaBan et al.⁷⁰ offered a theoretical explanation for the association between venous pathology and RLS. The authors suggested that an increase in paraspinal venous volumes within the reduced confines of a stenotic lumbar spine could contribute to leg pain and paresthesias arousing patients from a sound sleep (viz., Vesper's curse). Symptomatic relief observed with assumption of an erect or semireclining posture in the 6 subjects examined is highly suggestive, but not diagnostic, of RLS.

Patients with RLS, PLMS, or both, also have cold feet, suggesting sympathetically mediated vasoconstriction (see previous discussion). Not only is sympathetic hyperactivity associated with RLS/PLMS, but that same sympathetic hyperactivity might in turn worsen the symptoms of RLS/PLMS. Work by Ware et al.¹³ and Ancoli-Israel et al.¹⁴ indicates that a reduction in sympathetic tone can lead to a reduction in PLMS. Rajaram et al. also explored the possibility that vasoconstriction produced by sympathetic hyperactivity was causal to RLS symptoms by employing enhanced external counterpulsation to increase long-term vascular flow to the limbs in patients with RLS. The results were initially encouraging, but the double-blind study did not show any improvement in RLS symptoms.^71,72 Medications used to treat cardiovascular disease might also play a role in the provoking RLS/PLMS although there is at present, no evidence to support this hypothesis.