Abstract
To investigate the development of REM sleep behavior disorder (RBD) and REM sleep behavioral events (RBE) not yet fulfilling diagnostic criteria for RBD as markers for neurodegeneration in a cohort of Parkinson disease (PD) patients between their de novo baseline assessment and two-year follow-up in comparison to healthy controls (HC).
Clinically confirmed PD patients and HC with video-supported polysomnography (vPSG) data at baseline were re-investigated after two years. Diagnostic scoring for RBE and RBD was performed in both groups and related to baseline findings.
One hundred thirteen PD patients and 102 healthy controls (HC) were included in the study. Within two years, the overall occurrence of behaviors during REM sleep in PD patients increased from 50% to 63% (P = 0.02). RBD increased from 25% to 43% (P < 0.001). Eleven of 29 (38%) RBE positive PD patients and 10/56 (18%) patients with normal REM sleep at baseline converted to RBD. In HC, the occurrence of any REM behavior increased from 17% to 20% (n.s.). RBD increased from 2% to 4% (n.s.). One of 15 (7%) RBE positive HC and 1/85 (1%) HC with normal REM at baseline converted to RBD.
RBD increased significantly in PD patients from the de novo state to two-year follow-up. We propose RBE being named “prodromal RBD” as it may follow a continuous evolution in PD possibly similar to the spreading of Lewy bodies in PD patients. RBD itself was shown as a robust and stable marker of early PD.
The polysomnographic data that we obtained from this long-term cohort of de novo PD patients shows an increase in the frequency of RBD in early PD over time. Our data demonstrate that RBD does not resolve with dopaminergic treatment in early PD. We found that REM sleep-associated motor behaviors not fulfilling diagnostic criteria of RBD—“REM behavioral events (RBE)”—are a precursor of RBD in 38% of cases, and as such could be referred to as “prodromal RBD.”
INTRODUCTION
Over the last two decades, rapid eye movement (REM) sleep behavior disorder (RBD) has become a prominent premotor marker for Parkinson disease (PD) and is now recognized as a prodromal state of neurodegenerative diseases associated with misprocessing of intracellular α-synuclein,1–3 thus facilitating the identification of an at-risk population. In clinically manifest PD, several studies have shown an association between RBD and a more severe course of the disease.4–6 Our own data from a large cohort of PD patients in various stages of the disease supports the concept that RBD signals a more advanced stage of neurodegeneration,7 suggesting RBD is also a possible clinical progression marker. However, little is known about the evolution of RBD in the individual patient.
In the context of a prospective observational longitudinal cohort study of initially de novo PD patients and healthy controls (“DeNoPa” cohort), we previously identified motor behaviors and/or vocalizations in REM sleep with a purposeful component other than comfort moves in 51% of these de novo PD patients investigated with video-supported polysomnography (vPSG).8 Only half of these, i.e., 25% of the total cohort, fulfilled current diagnostic criteria for RBD. We presented preliminary evidence that these “REM behavioral events” (RBE) not fulfilling RBD defining criteria corresponded to dreaming suggesting that RBE may be a precursor to RBD.9 We now present the two-year follow-up data to further clarify the role of RBE as a possible very early marker for neuro-degeneration and investigate the progression of REM sleep associated behaviors.
METHODS
The presented data are part of the 2-year follow-up evaluation of a prospective longitudinal observational cohort study of, at baseline, de novo PD patients and healthy controls (HC) for assessment of biomarkers in PD (for further details see Mollenhauer10). PD patients were clinically reassessed at 2-year follow-up, and only those with confirmation of the diagnosis of PD were included in the study. PD patients and HC required complete vPSG data at baseline and at 2-year follow-up.
Standard Protocol Approvals, Registrations, and Participants' Consent
All participants gave consent for their data to be scientifically evaluated, and signed additional consent to use their nighttime videos for scientific evaluation. (Ärztekammer Hessen, Approval No. FF89/2008).
Sleep Data
At 2-year follow-up, all participants were studied for one night in the sleep laboratory with vPSG according to American Association of Sleep Medicine (AASM) criteria, using the same techniques for recording and evaluation as for the baseline assessment,8 with the exception that at follow-up both flexor digitorum superficialis (FDS) muscles were also measured in addition to the activity of the mentalis muscle. Real-time video review of all REM sleep phases was performed. Comfort moves, neck myoclonus, respiratory noises, and events related to arousals were discounted. All motor behaviors and/or vocalizations with a purposeful component, seemingly expressive of a subject's mentation, were noted. Each behavioral event was scored by 2 of the 3 experienced raters (AW, CT, FSD). In case of a divergent scoring, the event was reconsidered by all 3 reviewers. Only events with a unanimous decision were counted. At least 2 separate behavioral events and/or vocalizations during REM sleep had to be present. REM without atonia (RWA) was measured as any surface EMG activity of the mentalis and both flexor digitorum superficialis (FDS) muscles in 3-s mini-epochs during REM sleep using the SINBAR method and cutoff values specific for RBD set at EMG activity rates of 18.2% for mentalis and 16.8% for FDS muscles.11 Additionally, mentalis muscle activity changes over the 2-year time span were analyzed. RBD was defined according to the second edition of the International Classification of Sleep Disorders (ICSD-2)12 with RWA above these cutoffs. All other subjects with at least 2 episodes of motor behaviors and/or vocalizations during REM sleep but RWA below the specific cutoffs were classified as RBE positive. History of possible RBD was assessed with the RBD screening questionnaire (RBD-SQ).13
Assessment of Clinical Data
To define possible phenotypes in the PD cohort we correlated sleep parameters with the following clinical variables: demographic data (including gender and age at the time of vPSG), motor impairment measured by the Unified Parkinson's Disease Rating Scale motor score (UPDRS III) and classification as tremor versus postural instability/gait disorder (PIGD) phenotype according to Jankovic.14 We used the motor assessments at baseline (without any dopaminergic medication) to determine the motor subtype. Cognitive performance was assessed with the Montreal Cognitive Assessment Test (MoCa) at follow-up. Levodopa equivalent daily dosage (LEDD) at the time of the 2-year follow-up vPSG was calculated according to Tomlinson.15 We also documented the use of co-medications potentially influencing RWA and RBD such as selective serotonin/noradrenalin re-uptake inhibitors (SSRI/SNRI), tri-cyclic antidepressants (AD), and benzodiazepines (melatonin was not used) and compared within-subject changes for these substances from baseline to follow-up.
Statistical Analysis
We used McNemar's χ2 tests to explore whether the proportions of PD patients and HC with RBE and RBD changed from baseline to follow-up. Additionally, mentalis muscle activity changes and changes in awakenings over the 2-year time span were analyzed using paired sample t-tests. In the PD cohort, effects of selected predictor variables on the occurrence of RBE and RBD at follow-up were tested by means of multi-nominal regression analysis. Group differences in continuous variables between PD patients with normal REM, PD patients with RBE and PD patients with RBD at follow-up were tested using analysis of variance (ANOVA) based on the heteroscedasticity-corrected covariance matrix.16 Significant omnibus effects were explored using the Tukey Honest Significant Difference method, controlling for the inflation of Type 1 error within the ANOVAs. Group differences in categorical variables (e.g., gender) were tested using χ2 tests. All analyses were carried out with the statistical platform R, using the packages “nnet,” “car,” and “multcomp.”
RESULTS
Altogether, 113 clinically confirmed PD patients and 102 HC were available for analysis with complete vPSG data from both baseline and 2-year follow-up. At follow-up, REM sleep-associated motor behaviors and/or vocalizations were detected in 71 (63%) of PD patients and 20 (20%) of HC (P < 0.001). In the PD cohort the occurrence of REM-sleep associated motor behaviors and/or vocalizations increased significantly from 57 (50%) at baseline to 71 (63%) at follow-up (P = 0.02). This was due to an increase in the number of PD patients fulfilling diagnostic criteria of RBD from 28 (25%) to 49 (43%) (P < 0.001). The number of patients with RBE did not change significantly (P = 0.44) with 29 patients at baseline (26%) and 22 patients at follow-up (19%). In the HC cohort, the increase of overall REM sleep-associated motor behaviors and/or vocalizations from 17 (17%) to 20 (20%), as well as the increase of RBD from 2 (2%) to 4 (4%) failed to reach significance. All individuals with RBD at baseline (PD: n = 28; HC: n = 2) were again identified with RBD after 2 years. In addition, 11/29 patients with RBE at baseline and 10/56 patients with initially normal REM were now classified as RBD at follow-up (Figure 1). In the HC cohort, 1/15 RBE positive individuals at baseline and 1/85 subjects with initially normal REM sleep converted to RBD (data not shown).
Study tree of 113 clinically confirmed Parkinson patients of the DeNoPa cohort at baseline and 2-year follow-up. *Cutoff for REM without atonia as measured on mentalis and flexor digitorum superficialis (FDS) EMG according to Frauscher11 with 100% specificity for RBD set at 18.2% and 16.8%, respectively. PD Parkinson disease; RBD, REM sleep behavior disorder; RBE, REM sleep behavioral events; FU, follow-up.
In the PD cohort, multinominal regression analysis controlling for age, gender, LEDD, cognitive performance, motor impairment measured by UPDRS III and motor phenotype classification at baseline, showed that only RBE was associated with an increased risk for RBD at follow-up (P = 0.002). RBDSQ scores were higher in the PD+RBD group compared to PD patients with normal REM (P < 0.001) but did not differentiate to the PD+RBE group. Co-medication potentially influencing RWA and RBD such as SSRI/SNRI, AD and benzodiazepines did not differ between the subgroups (Table 1). Within-subject changes revealed that 2/5 PD+RBD patients with SSRI/SNRI at follow-up had RBD at baseline without SSRI/SNRI, 1/5 of this subgroup who had converted from RBE to RBD at follow-up, had been newly started on SSRI/SNRI. The PD+RBE patient with SSRI/SNRI came from the PD with normal REM sleep group at baseline and had also been newly started on SSRI/SNRI. One patient with newly diagnosed RBD took AD at baseline and had discontinued this medication at follow-up. All PD+RBD patients with benzodiazepines at follow-up were pre-diagnosed with RBD at baseline and had newly been started on these substances (data not shown). Due to small numbers no statistical analysis was performed. Awakenings as a measure for sleep fragmentation did not differ between the subgroups and did not change from baseline to follow-up (Table 2 and Table S1 in the supplemental material). The frequency of sleep disordered breathing also showed no difference between subgroups and no difference over time.
Demographic and clinical data of Parkinson patients with and without REM sleep associated motor behaviors.
Demographic and clinical data of Parkinson patients with and without REM sleep associated motor behaviors.
Sleep parameters of Parkinson patients with and without REM sleep associated motor behaviors.
Sleep parameters of Parkinson patients with and without REM sleep associated motor behaviors.
RWA Measurements
RWA measurements revealed a higher percentage of muscle activity on mentalis and FDS EMG during REM sleep in the PD+RBD group compared to PD+RBE and PD with normal REM sleep (P < 0.001). Whereas mentalis EMG did not differentiate between PD+RBE and PD with normal REM sleep, patients with RBE showed more muscle activity on FDS compared to patients with normal REM sleep (P = 0.02). All other sleep parameters did not show any significant difference (Table 2).
Over the two years from baseline to follow-up we detected a mean 6% increase in mentalis muscle activity (P < 0.001) in PD patients. This was mainly due to the PD patients converting from normal REM to RBD (P = 0.006), as well as those converting from RBE to RBD (P = 0.008). However, we could also demonstrate an increase in mentalis muscle activity within the group of PD patients with normal REM sleep at both assessments (P = 0.02) (For details see Table 3).
REM without atonia (RWA) change from baseline to 2-year follow-up.
REM without atonia (RWA) change from baseline to 2-year follow-up.
DISCUSSION
Compared to HC polysomnographic data of this long-term cohort of initially de novo PD patients show a significant increase in the frequency of RBD in early PD over a time span of two years. The proportion of PD patients with RBD according to ICSD-2 criteria12 increased from 25% at baseline to 43% after two years, with confirmation of RBD in all patients previously diagnosed with RBD at baseline when untreated. Of note, this could also be shown for HC with RBD at baseline. However, as an observational period of two years is deemed too short to yield sufficient information on the development of sleep in an otherwise healthy ageing population, the further results in HC are not used for comparison in this study. Instead our analyses focus on the development of REM sleep motor behaviors in early PD patients. Our data prove that RBD does not resolve with dopaminergic treatment in early PD. When including RBE and RBD, the proportion of PD patients with any REM sleep-associated motor behaviors increased from 50% at baseline to 63%. Furthermore, we provide evidence that REM sleep associated motor behaviors not fulfilling RBD diagnostic criteria and therefore named “REM behavioral events (RBE)” may be a precursor of RBD in PD in 38% of cases. We propose naming these events “prodromal RBD.” RBE at baseline was incrementally associated with RBD at follow-up even after controlling for age, gender, LEDD, cognitive performance, and motor impairment as measured with UPDRS III and motor phenotype classification. Analysis of potentially REM sleep influencing co-medication revealed only very few or single cases in whom a change in the intake of SSRI/SNRI, tricyclic antidepressants, or benzodiazepines may have prompted manifestation of REM associated motor behaviors.
Although RBD-SQ results statistically discriminated RBD patients from those with normal REM and RBE, the wide range suggests limited sensitivity and specificity. These findings are in line with a previous study on PD patients with RBD in more advanced stages of the disease.17 Reduced self-awareness for abnormal nocturnal behaviors, the absence of a bed partner as well as the presence of other motor phenomena during sleep in PD may contribute to this limitation. These findings underline that a screening questionnaire such as the RBD-SQ will not facilitate identification of RBE and cannot replace vPSG in the diagnosis of RBD.
It is an ongoing discussion how many nights are needed to definitely classify RBD in a patient.18,19 It has been demonstrated that there is no difference between EMG activity scores on two consecutive PSG nights, and that one night may suffice for diagnosis of RBD if EMG criteria were combined with video analysis, although video analysis had poorer night-to night reliability.20 However, as patients with RBE did not show pathological RWA, EMG measurements alone will not allow for the differentiation of RBE positive subjects and those with completely normal REM sleep. Video analysis is mandatory for the identification of RBE and therefore the restriction to only one night of sleep study for the follow-up investigation, compared to two nights for the baseline assessment, may be responsible for the failure to detect RBE after two years in nine of those 29 patients initially classified with RBE. A considerable night-to-night variability in the severity of RBD manifestation has been described,21 and this may also hold true for RBE. As further limitations of this study we must point out that internationally accepted diagnostic criteria for RBE as well as inter-rater reliability have not yet been established. Although video analysis was rigorous excluding comfort moves, neck myoclonus, respiratory noises, and events related to arousals, as well as requiring a minimum of 2 events and a unanimous decision among the raters, RBE may still be a heterogeneous category with a wide variety of motor manifestations. Another possibility is that RBE may only appear as an intermittent phenomenon in some patients and may still resolve or show larger fluctuations. Whether or not the significant increase in RWA, even in those PD patients with normal REM sleep at baseline as well as after 2 years, indicates a progressive disintegration of muscle atonia regulation during REM sleep and thus predicts future REM motor behavior manifestation, cannot be answered at this time-point in our longitudinal study. The fact that PD+RBE patients showed more activity on FDS EMG than PD patients with normal REM sleep stresses the utility of this muscle for RWA analysis and may eventually become a diagnostic criterion for RBE. In our experience, RBE often manifests in upper limb movements, thus predestining FDS measurements to pick up this activity. Further follow-ups are needed to prove these hypotheses.
Previous studies provide evidence that PD associated with RBD predicts a more severe course of the disease with earlier cognitive decline, preferably affecting the PIGD phenotype.4–6 In this early stage of the disease we failed to demonstrate a difference in motor impairment, cognitive function, age, and gender between PD patients with RBE, with RBD, and those with normal REM sleep. Dopaminergic medication may not influence dream-enacting behaviors as there was no difference in LEDD in patients with or without RBD or prodromal RBD. At baseline, cognitive screening as well as detailed neuropsychological testing failed to detect differences in PD patients with RBD or RBE in comparison to those with normal REM.8 These findings were reproduced at follow-up (detailed tests not shown), summarized here by nonsignificant MoCa results. However, when looking for possible progression markers, the presence of RBD at baseline predicted a change in MoCa results over the two years,22 possibly as an early indication of emerging cognitive decline.
The hypothesis of RBD indicates a continuous increase from mild to more severe motor events during REM sleep. Data from animal studies reflecting the pathophysiology of REM sleep control23,24 are in accordance with the concept of PD with a slow intrusion of α-synuclein aggregation pathology within the REM sleep circuits modifying motor activity. Autopsy findings of α-synuclein aggregation pathology in RBD patients with or without neurodegenerative disease supports these concepts.3,25 Recent data from F-fluorodeoxyglucose positron emission tomography visualizing metabolic network activity identified similar metabolic patterns in patients with idiopathic RBD and those with early PD, showing a decrease in the RBD-specific metabolic network expression as the disease progressed.26 Whether or not these findings are mirrored by a decrease in the clinical manifestation of RBD/RBE is currently not known. Further longitudinal polysomnographic investigations are necessary to observe the natural development of REM sleep abnormalities in PD patients.
We were able to demonstrate in this cohort that motor behaviors and/or vocalizations during REM sleep, regardless of whether they fulfill diagnostic criteria for RBD serve as early markers for PD. At this early stage the presence of RBD or RBE is not yet definitely linked to the emergence of a specific phenotype of PD, but RBD at baseline may predict eventual earlier cognitive decline. In accordance with the terminology of prodromal PD,27 we propose the term “prodromal RBD” for RBE.
ACKNOWLEDGMENTS
Author contributions: Dr. Sixel-Döring: Research project (conception, organization, execution), statistical analysis (review and critique), manuscript (writing of the first draft). Dr. Zimmermann: Research project (execution), statistical analysis (design and execution), manuscript (review and critique). Dr. Wegener: Research project (organization, execution), statistical analysis (review and critique), manuscript (review and critique). Dr. Mollenhauer: Research project (conception), statistical analysis (review and critique), manuscript (review and critique). Dr. Trenkwalder: Research project (conception and organization), statistical analysis (review and critique), manuscript (review and critique).
Comments