SEARCH

SEARCH BY CITATION

Keywords:

  • atonia index;
  • REM sleep atonia;
  • REM sleep behavior disorder;
  • quantification

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD
  5. AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD
  6. OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA
  7. CONCLUSION
  8. ACKNOWLEDGMENT
  9. CONFLICT OF INTEREST
  10. REFERENCES

One of the essential features of rapid eye movement (REM) sleep behavior disorder is REM sleep without atonia seen during nocturnal polysomnographic recordings. In this paper we provide an overview about the varied scoring criteria proposed for visual analysis of loss of atonia during REM sleep. The automatic quantification of loss of atonia overcomes many of the limitations of visual scoring and these new approaches are reviewed. Finally, the contributions of these automatic methods to the understanding of the complex mechanisms underlying muscle atonia and motor suppression during REM sleep are briefly illustrated.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD
  5. AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD
  6. OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA
  7. CONCLUSION
  8. ACKNOWLEDGMENT
  9. CONFLICT OF INTEREST
  10. REFERENCES

Rapid eye movement (REM) sleep without atonia is the polysomnographic hallmark of REM sleep behavior disorder (RBD), a parasomnia where the physiological atonia during REM sleep is absent or greatly diminished and which is characterized by dream-enacting behavior associated with nightmares.1 In its current version the American Academy of Sleep Medicine (AASM) manual distinguishes between sustained tonic activity and excessive phasic muscle activity in REM sleep.2 The AASM features define the presence or absence of loss of atonia but have no quantitative character and are therefore currently not suited to characterize different degrees of loss of atonia, for example, for comparing group of subjects or to quantify the effect of treatments on RBD features. Even more importantly, one criterion of the International Classification of Sleep Disorders, second edition (ICSD-2),3 for the diagnosis of RBD is the “EMG finding of excessive amounts of sustained or intermittent elevation of submental EMG tone”. Without well-established normative values for physiological atonia, it is impossible to indicate cut-off values when sustained electromyography (EMG) tone can be considered to be “excessive”.

In this paper we provide an overview about scoring criteria, visual and automated quantification of loss of atonia in RBD. We will first review the different definitions for loss of atonia applied in studies of RBD. In recent years, some methods have been proposed to automatically quantify the amplitude of the chin EMG during REM sleep4–9 and differences and similarities in these approaches will be detailed next. Finally, we will illustrate applications of the automated quantification of loss of atonia. Although there have been earlier attempts to quantify EMG muscle tone during REM and non-REM (NREM) sleep (e.g.,10–14), we focus here on those criteria and scorings developed specifically for the scoring of chin EMG in RBD and applied in a larger group of subjects.

VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD
  5. AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD
  6. OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA
  7. CONCLUSION
  8. ACKNOWLEDGMENT
  9. CONFLICT OF INTEREST
  10. REFERENCES

There have been a multitude of proposals for the visual scoring and manual quantification of loss of atonia that have focused on tonic activation (Table 1), phasic activity (Table 2), or general or combined measures (Table 3) of chin EMG activation during REM sleep. The first systematic scoring and quantification of increased tonic activation during REM sleep in RBD has been proposed by Lapierre and Montplaisir in 1992.15 They classified all 20-s epochs of REM sleep depending on whether they contained tonic EMG activation for more than 50% of the epoch. This influential approach has been adopted in a large number of studies.16–32 A minor modification over the subsequent years was the adaptation of the original epoch lengths of 20- to 30-s intervals.4,33–41 Because the amplitude for scoring tonic activations was not specified in the original paper, several subsequent studies have explicitly defined amplitude criteria such as an increase of more than 2 µV above the baseline,42 amplitude greater than twice the individual level of atonia and greater than 10 µV,43–46 or greater than twice the EMG amplitude during N3 sleep.47 This last modification is similar to the definition of the AASM,2 which asks for amplitudes greater than the minimum amplitude during NREM sleep. A further, more laborious scoring considered tonic activity with amplitudes greater than four times the lowest amplitude during the present REM episode.48 All except one study49 considered tonic activation of chin EMG to be present when 50% or more of an epoch of 7.5,50 20,15–32,43–46 or 30 s2,4,33–42,47,48,51–53 contained such an activity. As to whole night quantification of tonic activity during REM sleep, there has been broad consensus across studies to quantify loss of atonia as the percentage of epochs containing tonic activity compared with the total number of REM epochs (Table 1).

Table 1. Scoring and quantification of chin electromyography (EMG) tonic activity during rapid eye movement (REM) sleep
No.ReferenceScoringQuantification
Epoch length (s)Duration (% of epoch)Amplitude of EMG activationNameMeasureStudies
  1. AASM, American Academy of Sleep Medicine; NREM, non-REM.

  Tonic activity
1aLapierre and Montplaisir 19921520>50% Tonic EMG activation Atonia %% of 20-s epochs 16–32
 Modification 130>50% Tonic EMG activation  % of 30-s epochs 4,33–41
 Modification 230>50%>2 µV above background   42
 Modification 330≥50%>2 × amplitude during N3 % of 30-s epochs 47
 Modification 420>50%>2 × individual atonia amplitude and >10 µV % of 20-s epochs 43–46
1bSheldon and Jacobsen 19984933–15 sMuscle artifact or ≥50% increase over baseline REM or tibialis activationMovements/REM% of 3-s epochs 
1cSchenck et al. 2003507.5>50%Increased tonic activity% with preserved atonia% of 7.5-s epochs 
1dWinkelman and James 20044830>50%>4 × lowest amplitude during the present REM episode % of 30-s epochs 
1eAASM manual 2007230≥50%>minimum of amplitude during NREM sleep   51–53
Table 2. Scoring and quantification of chin electromyography (EMG) phasic activity during rapid eye movement (REM) sleep
No.ReferenceScoringQuantification
Epoch length (s)Duration (% of epoch)Amplitude of EMG activationNameMeasureStudies
  1. AASM, American Academy of Sleep Medicine; PEM, phasic electromyographic metric.

  Phasic activity
2aLapierre and Montplaisir 19921520.1–5 s>4 × backgroundPhasic EMG %% of 2-s epochs 16–26,29,31,32,45
 Modification 130.1–5 s>4 × background % of 3-s epochs 4,27,33,34,36–38,40–42
 Modification 230.1–5 s>2 × background % of 3-s epochs 35,62,63
 Modification 320.1–5 s>4 × lowest amplitude during present REM episode % of 2-s epochs 48
 Modification 430.1–5 s>4 × background, separated by ≥1 s % of 3-s epochs 47
 Modification 520.3–5 s>4 × background % of 2-s epochs 44
 Modification 620.1–2 s>50 µV % of 2-s epochs 28
2bSheldon and Jacobsen 1998493<3 sMuscle artifact or ≥50% increase over baseline REM or tibialis activationEMG phasic density% of 3-s epochs 
2cEisensehr et al. 20035910>1 s each ≥0.5 s≥50% increase vs preceding atonia amplitude% of long EMG% of 10-s epochs 48
10<0.5 s≥50% increase vs preceding atonia amplitude% short EMG% of 10-s epochs 
>10×
2dBliwise et al. 2006542.5>0.1 sa≥4 × presleep baseline and “superimposed on a discernible background activity of not more than 25% of burst amplitude”PEM% of 2.5-s epochs 55
 Modification >0.1 sa aReturn to baseline had to be clearly present within each 2.5-s epoch
≥4 × background activity   60
2eAASM manual 20072300.1–5 sb≥4 × background   51,61
 Modification300.1–5 sc b≥5|10 3-s epochs
≥4 × background   52
c≥50%
2fMontplaisir et al. 20105620.1–10 s>4 × backgroundPhasic density %% of 2-s epochs 57,58
Table 3. Scoring and quantification of increased chin electromyography (EMG) during rapid eye movement (REM) sleep
No.ReferenceScoringQuantification
Epoch length (s)Duration (% of epoch)Amplitude of EMG activationNameMeasureStudies
  • Refers to definitions listed in Tables 1 and 2. PSG, polysomnography; RBD, REM sleep behavior disorder; REMREEA, REM related EMG activity; RWA, REM sleep without atonia.

  Overall/combined measures
3aGilman et al. 20033430 Tonic (def. 1a) or phasic activity (def. 2a)RBD PSG measure% of 30-s epochs 32
3bSchenck et al. 2003507.5 Increased EMG tone (def. 1c) and/or limb twitch burstsTonic/Phasic% of 7.5-s epochs 
3cArnulf et al. 2005691Yes/no≥amplitude during quiet wakefulness%RWA% of 1-s epochs 70–75
3dConsens et al. 20054130/3 Tonic (def. 1a, mod. 1) and phasic activity (def. 2a, mod. 1)PSG scoreAverage of % tonic and phasic activity 4
3eNightingale et al. 200578  Any degree of EMG activity regardless of specific amplitude in submandibular muscleIncrease in chin EMGSeconds 
3fStiasny-Kolster et al. 200568 >1 s or (>10× <0.5 s)b≥50% increase over preceding baseline atoniaEMG activity (% REM)% of REM time 
bIn 10 s
3gTuin et al. 20086430≥50%Tonic activity (def. 1a) or phasic events ≥2 s%RWA% of 30-s epochs 
3hZhang et al. 20083730/3 Tonic activity (def. 1a, mod. 1) or phasic activity (mod. def. 2a, mod. 1)REMREEA% of 30-s epochs with tonic activity + % of 3-s epochs with phasic activity 65
3iGagnon et al. 200676  ≥2 × background or >10 µV% muscle activity% of REM time 77
3jBonakis et al. 20096730≥50%Tonic activity (def. 1e) in submental muscles or phasic activity (mod. def. 2e) in submental, tibialis anterior musclesREM loss of atonia %% of 30-s epochs 
3kLin et al. 20096130 Tonic (def. 1e) or phasic activity (def. 2e)%RWA% of 30-s epochs 
3lSasai et al. 20116630 Tonic submentalis (def. 1a) or phasic submentalis and anterior tibialis muscles (def. 2a, mod. 1)%RWA% of 30-s epochs 

Compared with tonic activation, approaches to the scoring of phasic activity during REM sleep have been distinctly more diverse (Table 2). Again, the first definition was proposed by Lapierre and Montplaisir in 1992.15 They defined phasic activity as EMG activations between 0.1 and 5 s with more than four times the amplitude of the background activity. Other criteria included durations between 0.1 and 2 s,28 2.5 s,54,55 or 10 s,56–58 between 0.3 and 5 s,44 or all activity shorter than 3 s.49 In addition, Eisensehr and coworkers classified 10-s epochs as containing long EMG activity when the sum of the individual activations at least 0.5 s long was longer than 1 s, or with short activity when the same interval contained more than 10 activations shorter than 0.5 s.59 The multitude of duration criteria is matched by the diversity of amplitude definitions, which have been given as greater than four times the background activity,2,4,16–27,29,31–34,36–38,40–42,44,45,47,51,52,56–58,60,61 two times the background,35,62,63 or greater than 50 µV.28 Other approaches proposed amplitudes greater than four times the lowest amplitude during the current REM episode,48 1.5 times the preceding atonia amplitude,48,49,59 or four times the presleep baseline.54 The combination of these diverse amplitude criteria with the various duration criteria resulted in a considerable number of different proposals for the scoring of phasic activity during REM sleep. The quantification of phasic activity, however, generally followed the same approach of quantifying the percentage of 2-, 2.5-, 3-, or 10-s epochs containing such an activity (Table 2).

In addition to the scoring of tonic and phasic activity during REM sleep, several authors have proposed combined or overall measures of EMG activity (Table 3). One group of studies quantified the percentage of epochs containing either tonic or phasic activity as defined above.32,34,61,64 Other approaches averaged4,41 or summed37,65 the percentages of tonic and phasic activity. Still, other combined the amount of tonic activity of the chin EMG with limb twitch bursts,50 or with phasic activity of chin EMG and anterior tibialis muscles.66,67 Stiasny-Kolster et al. followed a proposal of the task force for motor activity during sleep of the German Sleep Research Society and quantified all activity 1.5 times higher than the preceding baseline atonia and either longer than 1 s or shorter than 0.5 s but occurring more than 10 times within a 10-s window.68 Other overall measures included the scoring of all activity with amplitudes greater than the amplitude during quiet wakefulness,69–75 greater than two times the background or 10 µV,76,77 or any EMG activity regardless of its amplitude.78

It is not known how these different scorings and quantifications compare to each other. Given the diversity of definitions, substantial differences in the amount of loss of atonia scored may be assumed. In addition, visual scoring and manual quantification are labor-intensive and susceptible to inter- and intra-rater differences. So far, only two studies have assessed inter-rater reliability for the scoring of phasic activity during REM sleep. Frauscher et al. compared phasic activity scorings (def. 2a, mod. 2 in Table 2) of 100 3-s mini-epochs between raters from different sleep laboratories and reported mean kappa values of 0.872 (range, 0.72–1.0).63 Bliwise and coworkers compared two different scorers who quantified the phasic electromyographic metric (def. 2d) in 27 recordings of patients with Parkinson's disease.54 Phasic electromyographic metric was scored for chin, arm, and leg muscle activations and during both REM and NREM sleep and the median inter-rater reliability across these different scorings was 0.77. It should also be considered that the visual methods of quantification of the EMG amplitude during REM sleep have been developed mostly during the paper polysomnographic recording era and that the same type of methods published later did not take advantage of the opportunities offered by digital recordings, continuing on the old path. This, together with the time-consuming features of this type of approach and with the relatively low inter-rater reliability found, constitutes a strong point against the extensive use of these methods in the clinical and research settings and against their potential use in a new, more quantitative definition of RBD.

AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD
  5. AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD
  6. OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA
  7. CONCLUSION
  8. ACKNOWLEDGMENT
  9. CONFLICT OF INTEREST
  10. REFERENCES

Automated quantification of loss of atonia promises to overcomes the limitations of the visual approach and several measures have been proposed in recent years. Burns and coworkers quantified the variance of the chin EMG for each 3-s mini-epoch during NREM and REM sleep.4 Mini-epochs during REM sleep with increased muscle activity were identified as those with variance values above the 5th percentile of variance values during NREM sleep. The percentage of 3-s mini-epochs during REM sleep with increased muscle activity was quantified as suprathreshold REM EMG activity metric (STREAM). Automatically computed STREAM values were compared with visually scored percentage of tonic and percentage of phasic activity according to Lapierre and Montplaisir:15 the correlation between STREAM scores and the average of percentages of epochs with tonic and phasic activity was 0.87.

Ferri et al. proposed the REM sleep atonia index that automatically creates a measure of tonic activation during REM sleep from the rectified, band-pass (1–100 Hz) and notch (50 Hz) filtered submentalis muscle EMG signal.9 In short, the average amplitude of the filtered and rectified EMG signal is computed for each 1-s mini-epoch. The average amplitude is then classified into 20 distinct categories (amp ≤ 1 µV, 1 µV < amp ≤ 2 µV, . . . , 18 µV < amp ≤ 19 µV, amp > 19 µV). The REM sleep atonia index is computed as the percentage of average amplitudes ≤ 1 µV divided by the sum of percentages of average amplitudes > 2 µV (i.e., amp ≤ 1 µV/[100 − 1 µV < amp ≤ 2 µV]). This index expresses mathematically the preponderance of atonic mini-epochs (amp ≤ 1 µV) with respect to all other mini-epochs with the exception of amplitudes > 1 µV and ≤2 µV, which are thought to reflect both atonia and EMG activation. The REM sleep atonia index can vary from 0 (complete loss of atonia, i.e., absence of mini-epochs with average amplitudes ≤ 1 µV) to 1 (all mini-epochs with amplitudes ≤ 1 µV). Other measures derived with this approach include the number of movements during REM, defined as the number of consecutive 1-s mini-epoch with average amplitudes > 2 µV, and a classification of the duration of these movements into 20 distinct categories (see Table 4). Results of the automatic analysis, when compared to visual scorings of loss of atonia and phasic density, as proposed by Lapierre and Montplaisir,15 showed adequate agreement: the automatically computed REM atonia index was compared with visually scored percentage of atonia, and the number of automatically detected EMG activations (≤5 s) to visually scored 2-s mini-epoch containing phasic EMG events. Average correlations in four large groups of subjects (young controls, old controls, idiopathic RBD [iRBD], multiple system atrophy [MSA]) varied between 0.745 and 0.963 for the REM sleep atonia index, and between 0.628 and 0.915 for the number of EMG activations. Subsequent to the original publication, an additional noise reduction method was introduced5 where the 1-s average amplitude was corrected for the local level of noise by subtracting from it the minimum amplitude of the EMG signal in a moving window of 60 s surrounding the 1-s mini-epoch. This was based on the assumption that this minimum value was a good estimate of the local level of noise that affects all values within such a 60-s period. The noise reduction correction made the REM sleep atonia index more easily applicable and suited for the analysis of larger number of recordings. Analysis of the corrected index in a large group of 89 subjects (young controls, aged controls, untreated and treated iRBD, MSA, obstructive sleep apnea syndrome (OSAS)) showed that the corrected index was numerically higher compared with the original index but distinguished reliably between patients and controls, affording the establishment of clinical cut-off values.

Table 4. Algorithms for the automated scoring and quantification of chin electromyography (EMG) during rapid eye movement (REM) sleep
No.ReferenceNameAlgorithm/definitionStudies
  1. NREM, non-REM.

1Burns et al. 20074SignalBand-pass (10–70 Hz), notch (60 Hz) filtered chin EMG signal, sampling rate 200 Hz 8
Algorithm(1) Variance of signal during all 3-s epochs (REM and NREM)
(2) Cut-off 5th percentile of variances during all NREM epochs
STREAMPercentage of 3-s REM epochs with average variance above the 5th percentile of variances during NREM
2Ferri et al. 20089SignalRectified, band-pass (10–100 Hz), notch (50 Hz) filtered signal, sampling rate 200/256 Hz 79
Algorithm(1) Average amplitude computed for each 1-s mini-epoch
(2) Individual stage-specific percentages of average 1-s amplitudes (amp) in 20 categories (amp ≤ 1 µV, 1 µV < amp ≤ 2 µV, . . . , 18 µV < amp ≤ 19 µV, amp > 19 µV)
REM sleep atonia index% amp ≤ 1/(100 – %1 < amp ≤ 2)
Movements/h REMNumber of consecutive 1-s mini-epochs with amp > 2 µV/h REM
REM movement durationIndividual percentages of duration (dur) of movements (consecutive 1-s mini-epochs with amp > 2 µV) in 20 categories (dur = 1 s, dur = 2 s, . . . , dur = 19 s, dur > 19 s)
Noise-reduction modification5 Modification of step 2: 80
1-s mini-epoch amplitude = average amplitude – minimum amplitude of surrounding moving window ± 30 s
3Mayer et al. 20086SignalHigh- (5 Hz), low-pass (120 Hz) filtered signal, 50 Hz digital filter, sampling rate 200 Hz 
Algorithm(1) Amplitude as difference between smoothed (0.025 s/5 samples) upper and lower envelope 
(2) Threshold curve as 2 × smoothed (200 s) amplitude curve
Muscle activityAll sampling points with amplitude (1) above threshold curve (2); clusters with <1-s distance were defined as one event 
Short activityMuscle activity < 0.5 s 
Long activityMuscle activity ≥0.5 s 
Mean muscle toneAverage of amplitude curve for 1-s mini-epochs 
4Shokrollahi et al. 20097SignalBand-pass (10–100 Hz) filtered signal, sampling rate 256 Hz 
Algorithm(1) Fixed segmentation of REM EMG signal from all subjects (four with RBD, four controls) 
(2) One level discrete wavelet transform (Daubechies, dB4) of signal of each segment, extraction of approximation coefficients (0–128 Hz)
(3) Continuous wavelet transform (Daubechies) of approximation coefficients, extraction of scale coefficients (15–100)
(4) Computation of mean scale values per segment, selection of scale with largest coefficient per segment; selection of most frequently chosen scale across segments as classifier
(5) Classification of continuous wavelet transform for chosen scale of all segments with linear discriminant analysis
5Kempfner et al. 20108SignalBand-pass (20–70 Hz), notch (48–52 Hz) filtered signal, sampling rate 200 Hz 
Algorithm(1) REM signal divided into 3-s mini-epochs and four features extracted for each segment: 
• Kurtosis
• Percentile ratio (75th percentile of absolute amplitude values in REM segment/75th percentile of absolute amplitude values of all NREM segments)
• Log-power (logarithm of the sum of squared amplitudes in segment divided by the number of sampling points [600])
• Median frequency (frequency at median of power spectrum density of segment)
(2) Principal component analysis of the four feature vector; extraction of the first two components
(3) Quadratic discriminant analysis of the two components with leave-one-out technique to determine optimal decision boundaries

A further automatic method was proposed by Mayer et al.6 They computed a smoothed upper and lower amplitude envelope of the chin EMG over NREM and REM epochs and defined amplitude as the difference between these two amplitudes. A threshold curve was constructed as twice the amplitude curve smoothed over 200 s and muscle activity during REM was defined as all activity with amplitudes above the threshold curve. From these computations mean amplitudes per 1-s mini-epochs and per sleep stage were computed for descriptive purposes. In addition, short (<0.5 s) and long (≥0.5 s) muscle activity was distinguished. The results of the automated quantification were not compared with visual scoring but the number of short and long muscle activities was higher in patients with RBD than controls.

A very different approach to the analysis of chin EMG during REM was published by Shokrollahi et al.7 In contrast to the other algorithms, they did not aim to quantify muscle activity during REM sleep directly but were interested in separating fixed-length segments of chin EMG during REM sleep of four RBD patients from those of four subjects without RBD. To that end they used linear discriminant analysis of the coefficients of a continuous wavelet transform applied to the approximation coefficients of a one-level discrete wavelet transform (for details see Table 4). They report that 97.5% of the segments from control subjects were classified as “normal” whereas 91.3% of segments of RBD patients were classified as “abnormal” with an overall accuracy of 94.3%. Besides the lack of direct quantification of loss of atonia, also the assumption that all segments of REM sleep in RBD will contain muscle activity makes this approach difficult to apply in clinical practice and research.

A further approach aiming at classifying rather than quantifying muscle activity during REM sleep was proposed by Kempfner et al.8 They used quadratic discriminant analysis of the first two components extracted from principal component analysis of four features (kurtosis, percentile ratio, low power, median frequency) extracted for all 3-s REM mini-epochs of six patients with Parkinson's disease and RBD and six healthy controls (for details see Table 4). Nevertheless, for each of the 12 subjects they computed the percentage of 3-s epochs with muscle activity during REM as classified by the discriminant function and compared it with STREAM scores computed according to Burns et al.4 Receiver operating characteristic (ROC) analysis for the classification of RBD subjects versus controls showed a perfect classification for this algorithm (100% sensitivity and specificity) while STREAM yielded 100% sensitivity only with 33% specificity. Because of the data-based optimizing approach an independent cross-validation of this approach will be needed before considering it for clinical practice.

OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD
  5. AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD
  6. OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA
  7. CONCLUSION
  8. ACKNOWLEDGMENT
  9. CONFLICT OF INTEREST
  10. REFERENCES

There is a need to establish normative values for the quantification of loss of atonia in subjects suspected to have RBD. The analysis of chin EMG in large patient groups needs objective, reliable, and time-efficient approaches and the automated quantification of REM sleep muscle activity satisfies these criteria. As detailed above, several automated approaches have been proposed and we close this review with a short overview of the results obtained so far with these algorithms.

In the original publication of the STREAM algorithm,4 17 patients with neurodegenerative disorders and six control subjects were classified as no, probable, or possible RBD according to ICSD-based clinical impression. Both STREAM and the visual scoring of muscle activity separated subjects with and without probable or possible RBD with comparable sensitivity and specificity. In addition, an overall RBD symptom score was derived from a questionnaire filled in by the bed partner and both STREAM and the visual scoring showed a modest correlation (rho = 0.42) with the symptom score.

The REM sleep atonia index, in its original9 and corrected forms,5 has been successfully applied in several studies so far.5,9,79,80 Among the first results from these studies was the observation that REM sleep atonia is the lowest in patients with MSA, even lower than in patients with iRBD, who nevertheless also exhibit values distinctively lower than young or old control subjects.9 In a subsequent study,79 application of the atonia index in 34 patients with narcolepsy, half of whom had RBD, and 35 controls demonstrated that atonia was lower in narcolepsy patients irrespective of the diagnosis of RBD. Further application in large samples of patients and controls has led to the establishment of normative values for muscle activity during REM sleep.5 Based on the joint analysis in 25 young controls, 10 aged controls, 31 untreated and 8 treated patients with iRBD, 10 patients with MSA, and 5 patients with obstructive sleep apnea, discriminative thresholds of the atonia index (AI) could be established: AI below 0.8 and AI between 0.8 and 0.9. All young control subjects and 7 of 10 old controls had AI values above 0.9, the remaining 3 old control subjects had AI values between 0.8 and 0.9. In contrast to this, all MSA patients has values below 0.8 and 74% of RBD patients had values below 0.9. Importantly, the combination of REM sleep atonia index, and number and duration of movements has shown that different clinical disorders such as RBD, MSA, and narcolepsy are characterized by different patterns of motor behavior during REM sleep.9,79,80 Another important result of the application of this approach is the repeated observation of a mono-modal distribution of the duration of chin EMG activations, that is, a progressive decrease of frequencies from the shortest to longer distributions.5,80 This indicates that the classical distinction between phasic and tonic activations may need to be reconsidered as this form of distribution argues for a single underlying process.

Finally, in the most recent study80 the development of REM sleep muscle atonia over the life span was studied in a large group of subjects showing that increasing levels of atonia develop during a long period from childhood to young adulthood, with a possible partial reversal of this trend during advanced age. Figure 1 summarizes some of the study results obtained by the application of the REM sleep atonia index so far and demonstrates its successful application in clinical practice and large samples.

image

Figure 1. Rapid eye movement (REM) sleep atonia index5 in different groups of subjects. RBD, REM sleep behavior disorder; iRBD, idiopathic RBD; MSA, multiple system atrophy; PD, Parkinson's disease; NC, narcolepsy with cataplexy; OSAS, obstructive sleep apnea syndrome.

Download figure to PowerPoint

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD
  5. AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD
  6. OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA
  7. CONCLUSION
  8. ACKNOWLEDGMENT
  9. CONFLICT OF INTEREST
  10. REFERENCES

In summary, over the last 20 years, a multitude of approaches to the scoring and quantification of loss of atonia during REM sleep have been proposed. The availability of objective, reliable, and time-efficient automated quantification algorithms such as the REM sleep atonia index has the potential to replace visual scoring with the added advantage of being easily and internationally applicable in large samples. This, in turn, is expected to yield new information potentially able to expand our understanding of the complex mechanisms underlying muscle atonia and motor suppression during REM sleep.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. VISUAL SCORING AND MANUAL QUANTIFICATION OF LOSS OF ATONIA IN RBD
  5. AUTOMATED QUANTIFICATION OF LOSS OF ATONIA IN RBD
  6. OUTLOOK – APPLICATION OF THE AUTOMATED QUANTIFICATION OF LOSS OF ATONIA
  7. CONCLUSION
  8. ACKNOWLEDGMENT
  9. CONFLICT OF INTEREST
  10. REFERENCES