By Donald L. Bliwise, PhD, FAASM

Synucleinopathies are a family of neurodegenerative conditions that have as their hallmark pathologic feature the deposition of a misfolded protein, alpha synuclein, throughout the central and autonomic nervous systems. Parkinson’s disease (PD) is the most well-known of these, but they also include other conditions that may share some motor and autonomic features of PD, including dementia with Lewy bodies and multi-system atrophy (MSA). Older literature often included reference to MSA subtypes using earlier terminology (e.g., Shy-Drager syndrome, Olivopontocerebellar atrophy).As developments in diagnosis and, eventually, disease-modifying strategies continue to evolve for synucleinopathic disease, it is becoming increasingly evident that sleep medicine specialists may be asked increasingly to “weigh in” on one particular potential biomarker for these conditions: REM sleep without atonia (RSWA).

RSWA, of course, is most often associated with REM sleep behavior disorder (RBD), and the AASM International Classification of Sleep Disorders – Third Edition, Text Revision (ICSD-3-TR) defines RBD, in part, by the presence of polysomnogram- (PSG) defined RSWA. RSWA constitutes only one of many other potentially diagnostic biomarkers, which include: orthostasis, hyposmia, urinary dysfunction, color blindness, subtle motor abnormalities on neurologic examination, specific neuroimaging-based findings (e.g., downregulation of dopamine transporter in the striatum), visuospatial impairment on neuropsychological testing and, most recently, a positive alpha synuclein seeding assay obtained from cerebrospinal fluid or phosphorylated alpha-synuclein noted by skin biopsy.

The reader is directed elsewhere for more complete descriptions of these.1,2 The importance of the congruence of such biomarkers for clinical diagnosis and decision-making cannot be overstated, as disease classification across the synucleinopathic spectrum has been proposed recently to move from classic neurologic diagnoses to categorization based on the presence of such signs.3, 4 Although formal neurological consultation is probably indicated for adjudication across many of these domains, specialists in sleep medicine should be able to offer an informed opinion on the presence or absence of RSWA on a PSG. To rephrase that question more succinctly: How much muscle activity is too much muscle activity? A simple question indeed, for sure, but not one easily answered.

The AASM Manual for the Scoring of Sleep and Associated Events Version 3 stipulates that, in a given 30-second REM epoch, at least five 3-second intervals must show phasic activity above baseline to qualify as RSWA. Although seemingly specific, these criteria do not delineate how many epochs of REM must meet these criteria. A single 30-second epoch? Half of the REM epochs? All REM epochs?  Without such specifics, the current AASM criteria cannot be applied uniformly, which creates complexity for clinicians reviewing PSGs to provide meaningful interpretations. The ICSD-3-TR indicates that a 50% criterion is likely too high and suggests using additional limb muscle recordings to lower the threshold, but it notes that REM quantity remains an open issue.

In view of these considerations, it may be worthwhile to consider how sleep researchers over more than 30 years have dealt with this issue and then offer some simplified suggestions for how clinicians in sleep medicine can use that accumulated fund of data to more judiciously make diagnostic judgments regarding levels of “excessive” EMG activity. Admittedly, the specifics regarding the derivation of the numeric data on REM EMG measures can be overwhelming, and the reader interested in the excruciating details regarding differences in electrode placement site, specific amplitude and duration definitions of phasic and/or tonic activity, and hardware/software used is directed to those studies.5,6,7,8,9,10 These can also be reviewed in the excellent summary by an international group of RBD researchers.11 In lieu of this, I would offer the following suggestion.

Based on detailed quantification of all REM sleep epochs on the PSG for a given night, the upper limit (95th%/97th% [= 2 SD]) of the density of mini-epochs containing submentalis phasic EMG activity in normal subjects ranges from 8.6% to 11.9% to 14.3% to 14.8% to 20.8% to 24.0%.12,13,24,15,16,17 These data originate from a richly diverse compilation of well over 300 adults without known neurologic disease of varying age and sex and were recorded using different hardware/software and visually scored with admittedly different definitional approaches, any or all of which likely impacted the findings. Still, these measurements can be useful.  The upper bound of such activity estimated from all recorded REM on a given night yields a very practical “rule of thumb” schematic (expressed as a density) for classifying submentalis EMG activity during REM, which is:

  • > 25% activity (high activity, likely outside of normal range)
  • 10-25% activity (equivocal/borderline abnormal activity)
  • < 10% activity (within normal limits)

We have found that such pragmatic estimates have resonance when training our sleep fellows, since they are easy to understand and apply. Additionally, for any given PSG, density measurements can still be derived though detailed manual quantification, should they be required (e.g., research, medical/legal purposes), although the considerable time and expense to provide such measurements may not be necessary in most routine cases. These suggested estimated thresholds can be implemented into textual PSG interpretation to facilitate communication across centers. Ultimately, it is highly probable that characterization of the density of EMG activity in REM will be amenable to digitally based machine learning approaches. Although several machine learning approaches have attempted to summarize EMG activity, these systems typically employ proprietary software and do not yet have what might be termed “turnkey” applicability at this time.18,19,20

(As a parenthetical note, the minimal duration of REM sleep required to make such a determination of RSWA remains an open question. In the NAPS Consortium, a lower bound limit of a cumulative duration of at least five minutes of REM sleep across the night has been employed and has proved a workable solution. Of nearly 400 PSGs performed on RBD patients, only about 1% have failed to meet that criterion on a single diagnostic night of in-lab PSG.)

Finally, one must keep in mind that adoption of the aforementioned approach to determine whether REM submentalis EMG activity exceeds normal limits should not in itself signify a biomarker of incipient synucleinopathic disease in the absence of dream enactment behavior.

RSWA may indeed represent an early disease biomarker, but its utility must occur in the context of other corroborating biomarkers. The NAPS Consortium has shown considerable overlap between RSWA and other biomarkers.21 The positive predictive value (PPV) and true limits of sensitivity/specificity of RSWA as a sole biomarker for synucleinopathy remains an open question. Numerous worldwide cross-sectional studies have examined the potential discriminative value of RSWA, but the selected ratio of patients (including idiopathic/isolated RBD or even frank PD) to controls often approached 1:1 or an even greater imbalance in favor of patients versus controls (e.g., 3:1).12,13,14,15,16,17,22,23,24 By contrast, classic epidemiology teaches that the determination of PPV of any test to detect disease in a population requires that the sample under study mirrors the incidence of the disease in that population.25,26 In the case of the synucleinopathies, this would translate into disease/non-disease participant ratios approximating 2:100.27

The field has already learned that certain medications, sleep apnea, and posttraumatic stress disorder can be associated with RSWA.28.29,30 Until population-based, longitudinal studies clarify these issues with greater resolution, the sleep medicine clinician is advised to interpret judiciously excess EMG activity during REM sleep, to consider this finding a fundamentally descriptive observation, and to always seek further clinical correlation for such results.

Donald L. Bliwise, PhD, is a professor of neurology at Emory University School of Medicine in Atlanta, Georgia.

This article appeared in volume 11, issue 1 of Montage magazine.