Reduced Rapid Eye Movement Density in Parkinson Disease: A Polysomnography-Based Case-Control Study

Abstract

Study Objectives:

To explore rapid eye movement density (RD) in patients with idiopathic Parkinson disease (IPD) and to investigate its usefulness as surrogate marker of excessive daytime sleepiness, a frequent complaint in IPD patients.

Methods:

Retrospective polysomnography study on 81 subjects without dementia: 29 patients with early stage IPD (disease duration ≤ 3 y), 21 patients with middle- stage IPD (disease duration > 3 and < 8 y) and 31 healthy controls (HC). Rapid eye movement (REM) sleep was defined as any REM episode with > 3 min of continuous REM sleep. RD was defined as number of ocular movements per minute of REM sleep. Patients with early stage IPD and HC fulfilled the PD-specific sleepiness questionnaires Parkinson's Disease Sleep Scale (PDSS) and the Nonmotor Symptoms Questionnaire for Parkinson's disease (NMSQuest).

Results:

RD was lower in patients with IPD than in HC. The difference was most significant between patients with middle stage IPD and HC (P = 0.001), and most prominent for the third REM episode, again when comparing patients with middle stage IPD and HC (P = 0.03). RD was independent from sex, age, and other sleep parameters. In early stage IPD, RD correlated with the PDSS score (r = −0.63, P = 0.001) and the sleep-related questions of the NMSQuest score (r = 0.48, P = 0.017).

Conclusions:

REM density is reduced in patients with IPD and correlates with subjective scores on sleep impairment. As an indicator of persistent high sleep pressure, reduced RD in IPD is eligible as a biomarker of excessive daytime sleepiness in IPD. It possibly reflects direct involvement of the brainstem REM generation sites by the disease process. RD is a promising new tool for sleep research in IPD.

Citation:

Schroeder LA, Rufra O, Sauvageot N, Fays F, Pieri V, Diederich NJ. Reduced rapid eye movement density in parkinson disease: a polysomnography-based case-control study. SLEEP 2016;39(12):2133–2139.

Significance.

In idiopathic Parkinson disease (IPD) degeneration of sleep regulation centers has been postulated and progressive sleep “de-structuring,” evidenced by polysomnography and rapid eye movement sleep behavior disorder (RBD), identified as a forerunner syndrome. As rapid eye movement density (RD) is known to be highest when sleep pressure is lowest, we explored its usefulness as a marker of sleep pressure in IPD. RD was reduced in patients with IPD compared to controls and correlated with subjective scores on sleep impairment. Application in larger cohorts should replicate these results. Moreover, studies in patients with idiopathic RBD are promising, as in RBD there is increased muscle activity of chin and extremity muscles in rapid eye movement, contrasting with the here reported decreased RD.

INTRODUCTION

In idiopathic Parkinson disease (IPD), sleep complaints are frequent and disease-inherent degeneration of sleep centers has been postulated.^1,2 Polysomnography (PSG) has shown progressive sleep “destructuring” evolving independently from other major disease parameters.³ Rapid eye movement (REM) is a hallmark of REM sleep and is thought to be linked to the generation of pontogeniculo-occiptial waves.⁴ REM density (RD) measures the frequency of eye movements during REM sleep.⁵ In healthy subjects, it increases over the course of the night and is highest when sleep pressure is lowest.^4,6 Briefly, Borbely⁷ and others have proposed a two- process model for sleep regulation composed of the circadian process (process C) and sleep homeostasis (process S). The need for sleep or sleep pressure within the process C depends on prior sleep and wakefulness. Consequently, sleep pressure increases during the day and decreases during sleep. It also increases with sleep deprivation. Finally, the wakefulness signal provided by the process C is opposed to the build-up of sleep pressure during and its decrease during nighttime.⁸ In depressed patients, however, RD is uncoupled from sleep pressure and increased RD is already seen in the first part of the night.⁹

Several REM abnormalities have been reported in IPD. First, REM sleep behavior disorder (RBD) has been identified as a precursor sign of IPD. Second, in advanced IPD stages, REM sleep is substantially reduced.¹⁰ It can be hypothesized that other REM sleep abnormalities remain undetected. Therefore, the current study focuses on RD, a variable not yet explored in IPD. First, we explore the evolution of RD during total sleep time and we compare patients with IPD with age-matched healthy controls. Second, we explore if there is a statistical link between RD and subjective sleep impairment of patients with IPD, and if RD can be proposed as a surrogate marker for sleep dysfunction in IPD.

METHODS

Recruitment Procedures

Throughout the data bank of the Interdisciplinary Sleep Laboratory of the Centre Hospitalier in Luxembourg, PSG data of 81 subjects without dementia (48 men and 33 women, aged 38 to 87 y) were analyzed. All subjects underwent a 1-night polysomnographic study between 2001 and 2012. Based on the clinical examination before PSG they were classified into (1) patients with early stage IPD (ES-IPD), defined by a disease duration ≤ 3 y; and (2) patients with middle stage IPD (MSIPD), defined by a disease duration between 3 and 8 y. Further characterization of ES-IPD patients is presented in Table S1 in the supplemental material. Further characterization of all patients with IPD in terms of Hoehn and Yahr (HY) stage as well as levodopa equivalent daily dosage (LEDD) is presented in Table 1. All except five ES-IPD patients had been prospectively recruited within a project on early nonmotor signs in patients with IPD.¹¹ MS-IPD patients were eligible for PSG when presenting with minor sleep complaints, as described in detail previously.³ After Interdisciplinary Sleep Laboratory data extraction and chart reviewing, the diagnostic classification was revised, based on the latest clinical follow-up examination. Of note, dementia had been excluded by formal testing in 24 ES-IPD patients and 3 MS-IPD patients (Mini-Mental State Examination score equal to or above 24) and by global clinical impression in 5 ES-IPD patients and 18 MS-IPD patients. This procedure led to the exclusion of six patients from the current study as they were reclassified into nonidiopathic parkinsonism (two patients) and vascular parkinsonism (four patients). HCs were recruited as nonconsanguineous family relatives of ES-IPD patients as well as other healthy volunteer controls; on clinical evaluation they had no major sleep complaints and there was no sign indicative of neurodegeneration or depression.¹¹ Finally, four subjects (two HC and two ES-IPD patients) were excluded because there was no detectable REM sleep on the polysomnogram.

Table 1.

Disease characterization of 29 patients with early stage idiopathic Parkinson disease and 21 patients with middle stage idiopathic Parkinson disease.

Polysomnography

Polysomnography was performed with a standardized 21-channel montage (Brainnet-Morpheus, MEDATEC, Brussels): 2 electrooculogram channels, 1 electromyogram channel, 6 electroencephalogram channels, 2 nasal airflow channels, 2 respiratory effort channels, 1 oximetry channel, 2 snore detector channels, 1 electrocardiography channel, 1 pulse transit time channel, 1 body position channel, and 2 limb movement sensors, placed on the right and left leg. Sleep stage classification was performed into REM sleep and nonrapid eye movement (NREM) sleep, further divided into stage 1–2 and stage 3 according to Rechtschaffen and Kales.¹² This scoring system has been preferred to the newer American Academy of Sleep Medicine scoring system in order to guarantee comparability of all datasets, as PSG has been performed several times before the American Academy of Sleep Medicine amendment proposal. The following parameters were analyzed: total sleep time (minutes), sleep latency (minutes), sleep efficiency (percentage), NREM sleep time, and REM sleep time (in percentages of total sleep time and additionally in minutes for REM sleep). Any REM episode with > 3 min of continuous REM sleep was included in the analysis of REM sleep. REM density was defined as the number of ocular movements per REM sleep minute for the total night and, separately, per REM sleep minute for each REM episode 1 to 5. The automated identification of eye movements is generated when the amplitude on the right or left EOG recordings is conjugal or exceeding 15 μV. As eye movements may be mimicked by frontal muscle movements or may have lower amplitude in patients with IPD, manual scoring later validated all obtained values. The criteria for registration of REM density were identical between 2001 and 2012, according to the software provider MEDATEC. Finally, the percentage of REM sleep muscle atonia was assessed according to an established method.¹³ All PSG examinations were performed under the current medication, in order to avoid a medication withdrawal effect.¹⁴ Patients with IPD were taking their usual dopaminergic medication; five ES-IPD patients and one MS-IPD patient were also taking antidepressant drugs. LEDD are outlined in Table 1. ES-IPD patients and healthy controls had given prior written consent and the National Committee for Ethics of Research (CNER) Luxembourg approved the study (CNER N°200401/03). Anonymous interpretation of the data of MS-IPD patients had been similarly approved by the CNER (CNER N°200401/04).

Questionnaires

All participants fulfilled the Epworth Sleepiness Scale (ESS) on the eve of the PSG. ESS scores equal to or higher than 10 were considered to be pathologic.^15,16 All ES-IPD patients and all healthy controls fulfilled the Parkinson's Disease Sleep Scale (PDSS) and the Nonmotor Symptoms Questionnaire for Parkinson's disease (NMSQuest).^17,18 Only questions 22 through 25 of the NMSQuest, directly or indirectly dealing with sleep quality and thus reflecting also sleep propensity, were used for further statistical analysis.¹⁹

Statistical Analyses

The statistical analyses were performed with the SPSS Statistics program (Version 23.0. Chicago). RD was designated as the primary outcome variable and all other outcome variables qualified as secondary outcomes. For descriptive analysis, conventional statistical parameters including means and standard deviations were used to describe continuous outcomes, whereas percentages and frequencies were used for categorical outcomes. Normality was checked using the Shapiro-Wilk test. Associations between outcome variables and IPD (categorical variable with three values: MS-IPD, ES-IPD, and HC) were assessed with the appropriate statistical tests. Concerning continuous outcomes, the Kruskal-Wallis test was used for non-normal outcomes, whereas analysis of variance was used for normal outcomes. For significant associations, post hoc tests were performed (Mann-Whitney for nonnormal and Student t-test for normal outcomes). Regarding categorical outcomes, the chi-square test was used. Concerning correlations between RD and subjective sleep complaints, Pearson (respectively Spearman) correlations were applied for normal (respectively not normal) distributions. All P values were considered signifi-cant at the level of 5%.

RESULTS

Demographics

Demographic data, presented in Table 2, did not show any significant difference between the two IPD groups and the control group. HY stage, presented in Table 1, shows no significant difference between ES-IPD patients and MS-IPD patients (P = 0.124). LEDD showed a significant difference between ES-IPD patients and MS-IPD patients (P = 0.003).

Table 2.

Demographic data of 29 patients with early stage idiopathic Parkinson disease, 21 patients with middle stage idiopathic Parkinson disease, and 31 age-matched healthy controls.

Polysomnography Results

The PSG data are shown in Table 3. RD was lower in ES-IPD and MS-IPD patients than in HC. However, the difference was only significant between MS-IPD patients and HC (P = 0.001) (see Figure 1). The difference in RD is most prominent for the third REM episode: in this episode, RD of MS-IPD patients compared to HC was again significantly lower (P = 0.03) (see Figure 2). RD was independent from sex, age, sleep efficiency, sleep latency, duration of total sleep, and duration of the different sleep stages (data not shown). The total number of REM episodes was not different between the groups (see Table 3). Concering other sleep parameters such as total sleep time, sleep efficiency, or percenteges of different NREM sleep stages, there were no statistical differences (see Table 3). HC always showed better scores than patients with IPD (see Table 3), but the differences were not significant. RD correlates negatively with LEDD in ES-IPD patients (r = −0.49, P = 0.009, 95% confidence interval [CI] = −0.73 to −0.14), but not in MSIPD patients (see Table 1). Age had a significant effect on several sleep parameters in patients with IPD, but not in controls. Therefore, reduced sleep efficiency was only linked with age in patients with IPD (r = −0.32, P = 0.022, 95% CI = −0.55 to −0.05). In HC, no correlation was seen between age and sleep efficiency (r = 0.05, P = 0.78). In ES-IPD patients, age was also negatively correlated with overall REM sleep time (r = −0.35, P = 0.006, 95% CI = −0.64 to 0.002). In HC, age showed no correlation with overall REM sleep time (r = −0.03, P = 0.87) (data not shown). In patients with IPD and healthy controls, a linear regression analysis to investigate the influence of age, disease duration, and HY stage on RD was performed. Age (β = −0.022, P = 0.34) and HY stage (β = −0.09, P = 0.79) was not significantly correlated with RD, whereas disease duration (β = −0.21, P = 0.15) was negatively correlated with RD.

Table 3.

Polysomnographic data on 29 patients with early stage idiopathic Parkinson disease, 21 patients with middle stage idiopathic Parkinson disease, and 31 healthy controls.

Figure 1.

Distribution of rapid eye movement (REM) densities in healthy controls (HC), in patients with early stage IPD (ES-IPD) and in patients with middle stage IPD (MS-IPD).

Figure 2.

Distribution of rapid eye movement (REM) densities in third episode of REM in healthy controls (HC), in patients with early stage IPD (ES-IPD), and in patients with middle stage IPD (MS-IPD).

Figure 2 shows the evolution of RD during total sleep time; all subjects started with a low RD value in the first REM sleep episode, then RD increased. Of note, in the fourth REM sleep episode ES-IPD patients showed a larger RD increase than healthy controls, but these differences were not significant.

Questionnaire-Based Results

IPD patients reported higher ESS scores than HC (Table 4). However, all group scores were below the cutoff of 10 and the differences were not significant. In contrast, on the disease-specific PDSS questionnaire, ES-IPD patients performed less well than HC, with a trend toward a significant difference (P = 0.05). No significant difference between ES-IPD patients and HC was observed for the four selected questions of the NMSQuest (P = 0.159). In ES-IPD patients and HC, RD correlated negatively with the PDSS scores (r = −0.63, P < 0.001, 95% CI = −0.81 to −0.34). Only in ESIPD patients, RD positively correlated with the four selected questions of the NMSQuest scores (r = 0.48, P = 0.017, 95% CI = 0.10 to 0.74).

Table 4.

Results of questionnaires on 29 patients with early stage idiopathic Parkinson disease, 21 patients with middle stage idiopathic Parkinson disease, and 31 healthy controls.

DISCUSSION

As first shown by the sleep research pioneer Aserinsky⁴, RD in healthy subjects is an index of sleep satiety. RD increases with sleep duration and eventually saturates at a plateau.^4,6,20 The current study is the first to report a pathological reduction of RD in patients with middle stage IPD in comparison to healthy, age-matched controls. The reduction is most important for the third REM episode, again when comparing patients with middle stage IPD to controls. Furthermore, MSIPD patients do not show the physiological increase of RD at the end of the night, whereas this physiological increase is still preserved in the early stage of the disease, suggesting that at an early IPD stage increased sleep pressure can still be partially reduced, similar to the still mostly efficient motor compensation at this stage.³ RD is independent of sex, age, and various sleep parameters, including sleep efficiency and duration of the different sleep stages. REM sleep cycling is not reduced in patients with IPD who have a total number of REM sleep episodes similar to that of HC. In ES-IPD patients, RD robustly correlates with the scores obtained on two questionnaires dealing with subjectively perceived sleep impairment. As lower values of the PDSS scores and higher scores on selected questions of NMSQuest indicate higher, subjectively perceived sleep impairment, lower RD correctly indicates nonrestorative sleep. However, these subjective scores on sleep impairment had not yet been available in MS-IPD patients.

Our findings suggest that RD is an independent marker of sleep quality in ES-IPD patients, possibly directly reflecting disease-inherent involvement of REM generation sites.⁴ In particular, a causal direct relationship with disease involvement of the pontomesencephalic tegmentum has to be discussed. Studies^21,22 have pointed out that in IPD stage two, Lewy bodies are found in the medullary nuclei of the level setting system and that in stage three, the tegmental pedunculopontine nucleus is also affected. In addition to reduced cerebrospinal fluid hypocretin levels in IPD, reduced RD is another surrogate marker for sleep dysfunction in IPD, in particular for excessive daytime sleepiness (EDS).²³ In this context, the correlation of RD with the sleep-related questions of NMSQuest in IPD is remarkable as a direct link of EDS in IPD with other PSG parameters has not been found previously.^10,11

RD as phasic REM activity is linked to direct activation of the motor cortex during REM sleep.²⁴ In patients with IPD, reduced RD may induce less activation of the motor cortex in REM sleep, a striking parallelism with bradykinesia, which also has less motor cortex activation in wakefulness. However, it is important to highlight that RD evolves differently from other phasic REM sleep muscle activities. Increased, and not decreased, muscle activity of the chin and the extremities is seen in idiopathic RBD and in IPD with RBD. Independent-generation pathways of these different expressions of phasic REM activity and, consequently, independent disease involvement has to be discussed.^25,26

To the best of our knowledge, only one study has so far directly addressed RD in neurodegeneration. Patients with narcolepsy were compared to patients with RBD. RD was higher in patients with narcolepsy than in patients with RBD, but lower in patients with RBD than in controls. Because RBD is a synucleinopathy, just as is IPD, these findings are in line with the findings of the current study. A direct comparison of RD between patients with IPD and RBD is thus promising as RBD is an established precursor syndrome of IPD: parkinsonian syndromes, mostly IPD, will develop in 80% of patients with RBD within two decades.²⁷

Contrasting RD findings have been reported for depression. In major depression and in affiliated depressive syndromes RD has been explored most extensively. In general, an increase in RD has been seen in depressed patients, with the most important increase occurring during the first part of the night.²⁸ Increased RD correlates with the severity of the depression and its clinical outcome.²⁹ Even unaffected family members of depressed patients may show increased RD.^30–33 REM sleep disturbances in depression persist during lifetime and RD has been proposed as biomarker of depression.^34,35 In alcoholics with secondary depression, increased RD may predict relapses by 3 mo.³⁶ In borderline personality disorder higher RD in the first REM sleep episode has been hypothesized to be “a marker of liability to mood disorders”.³⁷ These observations suggest a link between possibly inborn abnormalities of REM sleep regulation and depression. The clinical dataset of our patients did not contain validated data on depressive symptoms of the IPD patients in order to evaluate if depressed patients with IPD preserve higher RD scores.

As current medication was maintained during PSG, in order to avoid a medication withdrawal effect, direct medication effects on REM sleep are possible. With this maintenance strategy we also could not evaluate the general effect of dopaminergic treatment on sleep architecture.³⁸ Antidepressants used by several patients with IPD could partially compensate sleep fragmentation. This potential interference could have been avoided by the recruitment of antidepressant-naive patients. However, for ethical reasons, such a patient selection was not applicable.

Our study has several limitations. The retrospective recruitment of MS-IPD patients may have introduced a recruitment bias. We have circumvented this problem for the ES-IPD patients and HC by the prospective recruitment of consecutive patients and HC.³⁹ However, subjective evaluation of sleep impairment has not been performed in the MS-IPD patients and the numbers of the subjects in the subgroups are small. The number of the subjects is limited and larger study groups are warranted to confirm the results. Furthermore, exact numbers for sample sizes are not known, because “the effect size between various sleep disorders and IPD has not been established”.³⁹ It is possible that in a 1-night PSG, the results are influenced by a first-night effect.^40,41 However, such an effect should similarly affect the sleep of HC. The scoring of PSG in patients with IPD is a demanding task, as disease-inherent sleep fragmentation, abrupt sleep-wakefulness transitions, and further electroencephalographic abnormalities may occur.^42–46 However, in the current study automatic pattern recognition had always to be confirmed by manual scoring, based on the long experience of this sleep laboratory in patients with IPD.⁴⁷ In addition to reduced REM density, there may also be reduced mean amplitude of the eye movements in patients with IPD. Although the manual rescoring probably has circumvented the danger of non- recognition of such eye movements in patients with IPD, we have not added the mean amplitude of REM as another parameter susceptible to distinguish patients with IPD from healthy controls. As night-to-night variability of quantitative REM sleep parameters is low,⁴¹ a single-night PSG has provided valid REM sleep data in other studies on IPD patients.^48,49 The strengths of the study include the differentiation of early and middle stage PD and the verification of the initial diagnosis by the follow-up examinations: several patients with vascular and nonidiopathic parkinsonism, in particular with progressive supranuclear palsy, had been excluded from the study, as this latter group of patients may already have reduced eye movements in wakefulness.^50,51

We conclude that with longer disease duration, patients with IPD become “lazy REMers” and that RD further decreases. RD may be a valid surrogate marker for the sleepiness complaints of patients with IPD. Adding RD to the diagnostic tool panel for investigating sleep dysfunction in IPD is worthwhile, as it not only gives a further clue for EDS in these patients, but also sheds light on REM sleep generation sites directly involved by the disease process.

DISCLOSURE STATEMENT

This was not an industry supported study. The study has been supported by an unrestricted grant from the Foundation Think, Luxembourg. Recruitment of the patients has been partially performed in a study supported by “Fonds National de Recherche,” Luxembourg (FNR) (FRN/06/04/05). The authors have indicated no financial conflicts of interest. The work was performed in the Department of Neurosciences, Centre Hospitalier de Luxembourg, Luxembourg-City, Luxembourg.