Rapid eye movement sleep behaviour disorder: Past, present, and future

Summary This manuscript presents an overview of REM sleep behaviour disorder (RBD) with a special focus on European contributions. After an introduction examining the history of the disorder, we address the pathophysiological and clinical aspects, as well as the diagnostic issues. Further, implications of RBD diagnosis and biomarkers are discussed. Contributions of European researchers to this field are highlighted.


| INTRODUCTION: HISTORY OF RAPID EYE MOVEMENT SLEEP BEHAVIOUR DISORDER (RBD)
The first description of dream-enacting behaviours dates back to the early Roman empire where reports on animals were provided as proof that non-human animals also dream (Pline l'Ancien, 1961). In the 17th century, Miguel de Cervantes described an episode of dream-enacting behaviour (Iranzo et al., 2004), in which Don Quixote shouts and vigorously attacks some wineskins while dreaming that he is fighting a giant. In 1881, Lasègue observed that patients with delirium tremens presented with dream-like behaviours, vocalisations, and rampaging in bed (Lasègue, 1881), features possibly resembling rapid eye movement (REM) sleep behaviour disorder (RBD). A major advance was made in 1965 when Jouvet and Delorme (1965) created an RBD model in cats 20 years before it was identified as a disorder in humans. Sastre and Jouvet sought to determine the origin of REM sleep muscle atonia and injected ibotenic acid in various nuclei to suppress atonia. They observed that lesions of the perilocus coeruleus alpha (an equivalent to the human locus subcoeruleus) resulted in dream-like behaviours ("oneiric behaviours"), including predatory attack, rage, flight, and grooming in cats (Sastre & Jouvet, 1979).
Europeans were also among the first to observe REM sleep without atonia (RWA) in individuals with parkinsonism. In 1969, Traczynska-Kubin et al. (1969) observed patients with Parkinson's disease (PD) having persistent mental electromyography (EMG) activity despite REMs resembling REM sleep. In 1975, Mouret (1975) described a persistence of mental EMG activity during paradoxical sleep in patients with PD compared with healthy individuals. Similar findings were named "stage 1-REM" by Japanese researchers (for a review see Tachibana, 2009) and "stage 7" sleep by Guilleminault et al. (1976) in patients with narcolepsy treated with clomipramine.
The Europeans were also the first to document video-polysomnographic (v-PSG) recording of an RBD episode in a patient with Shy-Drager syndrome (Coccagna et al., 1985). Shortly after, Salva and Guilleminault (1986) described the disappearance of REM sleep atonia and the appearance of complex dream-enacting behaviours in patients with olivopontocerebellar degeneration. Following on from this there were several reports by Japanese researchers of stage 1-REM, possibly resembling RWA in otherwise healthy-seeming elderly individuals experiencing their first manifestations of somnambulism-like behaviours (Tachibana et al., 1991), as well as in patients with brainstem degeneration (Shimizu et al., 1990), and under clomipramine treatment (Niiyama et al., 1993). A more detailed history of the detection of RWA and RBD has been provided in a review (Frauscher & Högl, 2012).
The landmark discovery of RBD as a disorder is attributed to Carlos Schenck and Mark Mahowald who described the first case series of people showing dream-enacting behaviours linked to a polysomnographic confirmation of RWA and gave RBD its name. Schenck and Mahowald identified the sleep-related violence and risk of injury, linked it with Jouvet's cat model, and discovered the benefit of clonazepam (Schenck et al., 1986). In the first brain neuropathological examination of a case with isolated RBD (iRBD), abnormal α-synuclein deposits were found in the locus coeruleus/subcoeruleus (Uchiyama et al., 1995). Schenck and Mahowald followed up their first 30 idiopathic RBD patients finding that one-third of their cohort later converted to parkinsonism (Schenck et al., 1996). This major finding occurred before Braak et al. suggested that abnormal α-synuclein propagates in a bottom-up pattern in prodromal and defined PD, starting in the gut and medulla oblongata (stage 1), then in the locus coeruleus/subcoeruleus (stage 2) and later in the substantia nigra (stages 3 and 4) and cortex (stages 5 and 6) (Del Tredici et al., 2002).
In 2007, Boeve, Silber, et al. (2007b) postulated that dysfunction in the sublaterodorsal nucleus (corresponding to stage 2 in the Braak staging system) could lead to RWA and RBD. This phenoconversion from RBD to parkinsonism and dementia was later confirmed in several countries, and other synucleinopathies were associated with RBD (Fantini et al., 2005;Iranzo et al., 2006).

| RBDpathophysiology
Glutamatergic neurons located in a small pontine nucleus localised ventral to the laterodorsal tegmental nucleus (LDT)named sublaterodorsal tegmental nucleus (SLD) in rats and locus subcoeruleus in humansare responsible for inducing muscle atonia during REM sleep. These neurons express the vesicular transporter 2 of glutamate (vGLUT2), the specific marker of glutamatergic neurons, and cFos, a marker of activation in rats displaying a REM sleep hypersomnia (Clement et al., 2011). Unit recordings of these neurons confirmed that they are selectively active during REM sleep (Boucetta et al., 2014). It has also been shown that genetic inactivation of glutamatergic SLD neurons in rats and mice induces RBD and a 30% decrease in REM sleep quantities (Uchida et al., 2021;Valencia Garcia et al., 2017).
It has also been shown that combined microdialysis of bicuculline (a GABA-A antagonist), strychnine (glycine antagonist), and phaclophen (a GABA-B antagonist) in the trigeminal nucleus are necessary to restore jaw muscle tone during REM sleep (Brooks & Peever, 2012). Furthermore, SLD neurons directly excite GABA/ glycinergic neurons located in the ventral medullary reticular nuclei (Boissard et al., 2002;Valencia Garcia et al., 2017;Valencia Garcia et al., 2018). Nearly all c-Fos-labelled neurons localised in these nuclei express GAD67 and glycine transporter 2mRNA after 3 h of paradoxical sleep recovery following 72 h of paradoxical sleep deprivation (Sapin et al., 2009;Valencia Garcia et al., 2018). These neurons directly project to spinal motoneurons (Valencia Garcia et al., 2018).
Inactivation of the GABA and glycinergic neurons of these nuclei in rats and mice induces RBD (Uchida et al., 2021;Valencia Garcia et al., 2018). All these experiments indicate that GABA/glycine neurons located in the ventral medulla project to and hyperpolarise motoneurons during REM sleep leading to muscle atonia.
Interestingly, functional neuroimaging and postmortem brain studies report the presence of Lewy bodies and neuronal loss in SLD and ventral medulla (Arnulf, 2012;Boeve, Dickson, et al., 2007a;Iranzo et al., 2013). It is, therefore, probable that RBD is due to specific neurodegeneration of the glutamate SLD and/or GABA/glycine medullary neurons.
It is well accepted that motoneurons are also phasically excited by glutamate during REM sleep (Burgess et al., 2008). It is, therefore, likely that RBD behaviours are due to phasic excitation of motoneurons by glutamate in the absence of tonic GABA/glycine inhibition.
Glutamate pre-motoneurons are interneurons located in close vicinity of motoneurons and neurons located in ponto-medullary reticular nuclei and the red nucleus (Rekling et al., 2000). These neurons are directly excited by glutamatergic pyramidal neurons of the motor cortex to induce voluntary movements (Rekling et al., 2000). The reticular formation and the red nucleus could play a major role in exciting motoneurons during REM sleep without the need for cortical activation (Blumberg & Plumeau, 2016;Del Rio-Bermudez et al., 2015).
However, patients often show long and complex behaviours, such as singing and giving long speeches, which strongly suggests that the motor cortex drives such behaviours (De Cock et al., 2007). In line with this, activation of the human motor cortex, similar to that observed during a voluntary movement during wakefulness, has been observed during REM sleep (De Carli et al., 2016).
Due to the characteristic emotional component of RBD dreamenacting behaviours and vocalisations, involvement of the limbic system has been postulated. This hypothesis is supported by animal studies on cats and by reports of RBD in humans affected by limbic encephalitis (Iranzo, 2018).
Everything considered, RBD is likely induced by neurodegeneration of the ponto-medullary GABA/glycinergic and/or glutamatergic neuronal system physiologically inducing muscle atonia during REM sleep, although other circuits are also involved in the generation of RWA and dream-enacting behaviours.

| Clinical features of isolatedRBD(iRBD)
The prevalence of iRBD in the general population over 60 years old is 0.5-1% (Haba-Rubio et al., 2018;Pujol et al., 2017), and the mean age at clinical consultation is usually in the seventh decade. iRBD is rare in those under 50 years (Fernandez-Arcos et al., 2016). For unknown reasons, people seeking medical consultation for iRBD at sleep centres are more frequently men than women, although this finding may be due to a selection bias as RBD in men is more violent than in women. Ageing, head injury, farming, pesticide exposure, antidepressant therapy, and GBA mutations are all risk factors for developing iRBD (Postuma et al., 2012). Hyposmia, depression, and constipation are more common in iRBD than in controls (Aguirre-Mardones After 15 years of follow-up, about 95% of iRBD patients will be clinically diagnosed with the synucleinopathies dementia with Lewy bodies (DLB, 45%), PD (45%), or multiple system atrophy (MSA, 5%) . Indeed, the presence of abnormal synuclein in the cerebral spinal fluid (CSF), olfactory mucosa, and peripheral organs (colon, skin, and salivary glands) is detected in most iRBD patients.
Markers of short-term conversion to a clinically overt synucleinopathy are hyposmia, abnormal DAT-SPECT, mild cognitive impairment, and minor parkinsonian signs. Patients with long-standing iRBD of more than 15 years may show synuclein in the CSF and organs, hyposmia and DAT deficit, indicating a slow but ongoing neurogenerative process (Hogl et al., 2018;Iranzo et al., 2017;Teigen et al., 2021).

| VIDEO-POLYSOMNOGRAPHY (V-PSG) AS A DIAGNOSTIC REQUIREMENT AND PROGRESSION MARKER INRBD
The current criteria for the diagnosis of RBD (American Academy of Sleep Medicine, 2014) require the demonstration of RWA on v-PSG.
Several manual/visual methods have been proposed to quantify RWA, but the most validated method that is also recommended by recent international guidelines  is the one proposed by the Sleep Innsbruck Barcelona (SINBAR) group (Frauscher et al., 2012). This method measures "any" (i.e. phasic or tonic) muscular activity in the chin and phasic muscular activity in both flexor digitorum superficialis muscles in the upper limbs, either in 30 s epochs or in 3 s epochs, and the respective cut-offs of 27% and 32% have shown to be sensitive and specific to distinguish RBD from controls (Frauscher et al., 2012). Quantification of RWA in the lower extremities has lower specificity than in the upper extremities . Because manual/visual RWA quantification is extremely laborious, several (semi-)automatic methods have been proposed. Of these, the REM atonia index (Ferri et al., 2010) is the most validated one, but it quantifies RWA only in the chin. Only one semi-automatic method measures muscular activity both in the chin and the upper extremities (Frauscher et al., 2014).  (Frauscher et al., 2007). Research to develop automatic tools that can identify such movements and, therefore, help to diagnose RBD (Waser et al., 2020), is essential.

| Sex issues inRBD
Since the seminal first description of five patients with RBD, a male predominance has been reported (Schenck et al., 1986). This was supported by subsequent studies that reported up to 80% of RBD patients being male (Schenck et al., 2019).
A more recent epidemiological study, however, suggested that the prevalence of RWA does not differ between the sexes (Haba-Rubio et al., 2018). As women with RBD often present less violent dreams and behaviours, it has been suggested that RBD might be underestimated in women due to milder symptoms or lack of perception by bed partners (Fernandez-Arcos et al., 2016).
Differences in dream content between men and women, investigated in people with PD-RBD, showed that women dream more about activities of daily living, family and friends, while men have more performance-related dreams, including about sports, employment and can also be more aggressive (Borek et al., 2007), although data on this in iRBD are lacking. Nonetheless, differences in dream content could explain why men more often show harmful behaviours during RBD episodes (Mahale et al., 2016) and have a higher risk of injuries compared with women (Comella et al., 1998).
In line with this, a polysomnographic study investigating motor features of sleep behaviours reported that men have higher EMG phasic activity, more myoclonic movements, and more movements involving the trunk, whereas segmental movements were more frequent in women (Bugalho & Salavisa, 2019). Another polysomnographic study investigating RWA in both legs and arms, reported a higher RWA index in the legs in men compared with women, and a higher RWA index in the arms in women. The authors postulated that this finding might reflect a different character of RWA among the sexes (Borek et al., 2007;Tatman & Sind, 1996).
Despite frequent reports of sex differences in RBD phenotypes, this aspect has yet to be studied extensively. Future studies are needed to better elucidate the reasons for sex differences in the prevalence of RBD, as well as to characterise potential differences in muscle activity and motor events during REM sleep between women and men (Bodkin & Schenck, 2009;Borek et al., 2007;Bugalho & Salavisa, 2019;Comella et al., 1998;Fernandez-Arcos et al., 2016;Haba-Rubio et al., 2018;Mahale et al., 2016;Schenck et al., 1986;Tatman & Sind, 1996;Zhou et al., 2015).

| The genetics ofRBD
Great strides forward have been made in iRBD genetic research over the past several years. iRBD likely has a distinct genetic risk profile compared with overt α-synucleinopathies, with evidence of both overlapping and contrasting risk loci across the conditions. RBD heritability, explained by common variants, is estimated at 12.3%, which is similar to the current estimation for DLB, and five RBD genetic loci have been identified by a genome-wide association study (GWAS) (Krohn et al., 2021). These genes are concentrated in the autophagylysosomal pathway (ALP) and are more specific to this pathway than any similarly powered GWAS of PD or DLB. Each of these RBD genes, SNCA, GBA, TMEM175, INPP5F, and SCARB2, are nominated PD risk loci (Nalls et al., 2019). However, the driving risk variants in SNCA and SCARB2 are independent, meaning different genetic mechanisms are driving the risk for RBD and PD at these loci. The top SNCA variant associated with an increased risk for PD may have an opposite effect in RBD, decreasing the risk (Krohn et al., 2020). There is evidence that these RBD and PD variants may be affecting gene expression differently, and in different brain regions, with RBD variants localised in cortical regions (Krohn et al., 2021). Additionally, prominent PD genes such as LRRK2 (Fernandez-Santiago et al., 2016), MAPT, and autosomal recessive genes (Mufti, Rudakou, et al., 2021a) are not associated with isolated RBD. A similar pattern is found between RBD and DLB, where ALP genes SNCA, GBA, and TMEM175 are shared risk factors in both conditions, however, DLB genes APOE and BIN1  are not associated with RBD (Gan-Or et al., 2017;Krohn et al., 2021). Mutations in PSAP, encoding for saposin C, a lysosomal activator of GBA, have also been implicated in iRBD (Sosero et al., 2022), as well as rare variants in LAMP3 (encoding the lysosomal associated membrane protein 3) and other genes (Mufti, Yu, et al., 2021b). Overall, the shared loci across RBD and overt α-synucleinopathies are localised in the ALP, and genes associated with other neurodegenerative mechanisms (e.g., tau aggregation, mitochondrial dysfunction) do not appear to play a major role in RBD susceptibility. Genetics may also contribute to the phenoconversion rate from RBD to overt neurodegeneration; studies in SNCA (Krohn et al., 2020) and GBA (Honeycutt et al., 2019) show evidence that risk variants are associated with rapid phenoconversion, however, with limited confidence at these sample sizes.

| Biomarkersclassical and recent
Clinical trials with putative neuropreventive strategies in PD have been largely unsuccessful in the past, in part because the target populations were early/de novo PD patients with characteristic motor symptoms already present. At this stage, the degeneration of dopaminergic neurons is advanced with over 50% of nigrostriatal dopaminergic neurons already being affected by neurodegeneration. Therefore, strategies to prevent future disease conversion need to focus on risk cohorts. To achieve this important step, studies should be conducted in iRBD patients to assess progression biomarkers in the early disease stage in parallel to larger cohorts of at-risk individuals. With an average of only 6.3% of iRBD patients converting to disease per year (Postuma et al., 2019), conversion alone cannot serve as an outcome measure for future clinical trials and further objective trait markers are needed. Also, individuals with iRBD convert into different α-synuclein aggregation diseases (i.e., PD, MSA, DLB). It may, therefore, be important to stratify these groups in the prodromal state. For prodromal individuals and population-based screens, which are becoming more popular for identifying individuals at risk, we need biomarkers that are less invasive than CSF, as well as widely applicable screening instruments. Several strategies are currently being investigated, including imaging studies (e.g., dopamine transporter imaging, 18 F FDG-PET), objective quantitative assessment of emerging mild motor disease, cognitive testing and tissue analyses with aggregation chemistry for α-synuclein (Miglis et al., 2021).
The total α-synuclein in CSF is 15% lower in PD and MSA, while in iRBD levels are slightly higher for, as yet, unknown reasons . More promising is the detection of α-synuclein aggregation by seeding aggregation assays (SAA; RT-QuiC or PMCA) with sensitivities and specificities above 90-95% for PD in CSF (Shahnawaz et al., 2017). One study in iRBD showed a sensitivity of 90.4% and a specificity of 90.0%. In one individual, positivity of SAA was detected 10 years before the conversion to disease  Major hurdles faced by SAA are the lack of quantification and the fact that the best performance is shown in CSF compared with colon, salivary glands, olfactory mucosa, and skin Doppler et al., 2017;Fernandez-Arcos et al., 2016;Iranzo et al., 2018;Sprenger et al., 2015;Stefani & Hogl, 2021;Vilas et al., 2016). The current mechanism of the molecular processes leading to α-synuclein aggregation and disease are still unknown and cannot yet be explained in relation to the assay parameters. While for CSF the results are promising, the first studies also report SAA positivity in skin samples in PD as well as in olfactory mucosa (De Luca et al., 2019;Wang et al., 2020). Some groups also show different seeding dynamics in patients with MSA and PD, which could indicate different α-synuclein strains (Shahnawaz et al., 2020).
A more peripherally acting biomarker is the neurofilament light chain (NfL), which can now reliably be quantified in peripheral blood.
Although nonspecific and elevated in several other neurological diseases, slightly higher values of plasma NfL are shown for PD and RBD and markedly higher for MSA patients (Mollenhauer et al., 2020).
In the future, we will know if other biomarkers such as inflammatory panels, faecal microbiome, and miRNA hold their promise of being adequate screening tools for those at risk of converting to disease. Unfortunately, none of the previous biomarkers, including SAA, can be currently used for individual quantification to differentiate between individuals at risk for conversion and those remaining free of a neurodegenerative disease. Larger cohorts are needed.

| Implications of theRBDdiagnosis
The implications of establishing the diagnosis of RBD are multifaceted. These encompass (i) clinical implications centred on the management strategies and risk mitigation of potential injurious RBD behaviours, (ii) implications related to very strong associations between RBD and evolving synuclein-specific neurodegeneration, (iii) the need to increase awareness about this disorder among physicians, allied health professionals, and the general public, and (iv) the opportunity to engage numerous stakeholders around RBD as a unique disorder that is positioned on the intersection of neurology, sleep medicine, and neuroscience.
The clinical implications of RBD as a parasomnia are centred on symptomatic management and safety precautions in sleeping environments aimed at preventing injuries from RBD-related motor behaviours (St Louis & Boeve, 2017). The relative paucity of well-designed randomised symptomatic clinical trials should be a call for action for a more robust clinical trial RBD pipeline (Gilat et al., 2020;Shin et al., 2019). The success of such trials will largely depend on the adequate selection of outcome measures and trial duration, which will allow for proper ascertainment of the intervention's effectiveness.
Implications of RBD as a prodromal stage of an evolving α-synucleinopathy are complex. RBD patients are increasingly becoming aware of their risk of developing one of these disorders in the future. This necessitates proper counselling on the risk of phenoconversion so that appropriate planning for the future can be considered. Currently, there is a substantial gap in prognostic counselling offered to the patient as reports indicate that only 50% of patients with RBD receive counselling (Feinstein et al., 2019). There is, therefore, a need to develop best practices for prognostic counselling in iRBD.
One of the most important implications of RBD as a prodromal synucleinopathy is an opportunity to position the iRBD population as an ideal study cohort for testing disease-modifying treatments aimed at delaying or preventing phenoconversion to a synucleinopathy (Hogl et al., 2018). Selecting the ideal patient candidates for disease-modifying trials, having effective and reliable screening and recruitment methods, employing robust clinical trial designs, and choosing appropriate outcome measures capable of demonstrating disease modification within the reasonable trial duration are the critical aspects of trial planning for iRBD. Some of these elements are more straightforward than others. The International RBD study group published two consensus statements on clinical trials in the RBD population that detail these aspects relevant to organisations and implementation of RBD clinical trials (Schenck et al., 2013;Videnovic et al., 2020).
Finally, raising awareness and promoting education about RBD is a very important implication of RBD, which is overall underdiagnosed, or the diagnosis is quite delayed. Bringing together various stakeholders such as clinicians, scientists, patients, disease-specific foundations, government agencies, industry, and the general public will enable us to advance various aspects of RBD clinic care, research, and therapeutic development.

| FUTURE DIRECTIONS
Since the first description of RBD, the disease has been better characterised through pathophysiologic, clinical and v-PSG studies, as well as through extensive research on biomarkers and genetics.
Despite these advances, the phase of phenoconversion from iRBD to overt α-synucleinopathy still needs to be further investigated.
A combination of biomarkers will likely allow better identification of those with iRBD at high risk of short-term phenoconversion, whereas it is still unclear how iRBD individuals that go on to develop PD will be differentiated from those phenoconverting into DLB or MSA.
Although a combination of biomarkers might be useful, it is also likely that a better phenotypic characterisation will be achieved through the identification of biomarkers or α-synuclein aggregates in biofluids or tissues, which are specific for one α-synucleinopathy (i.e., DLB, PD or MSA).
Another aspect that will become more relevant in the future and deserves further investigation is prodromal RBD. It has been defined as a stage in which symptoms and signs of evolving RBD are present, but do not yet meet established diagnostic criteria Hogl et al., 2018). International guidelines for the identification of prodromal RBD have been published recently by the International RBD study group, providing a framework that will ensure harmonised studies and, therefore, a better understanding of the clinical relevance of this condition and its evolution into iRBD.
All the previously mentioned aspects, as well as the identification of RBD in the general population, will likely take advantage of the implementation of artificial intelligence. Machine-learning approaches may provide new ways of identifying RBD patients (e.g., considering both muscle activity and movements), as well as new ways of phenotypic characterisation (e.g., based on a combination of biomarkers), dramatically improving detection and classification of RBD. Some open issues in the RBD research field concern treatment. New insights into pathophysiology deriving from basic science may lead to the development of novel treatments acting on specific pathways involved in the generation of RBD. Doubleblind, randomised, controlled trials are still needed to assess the efficacy of symptomatic RBD treatment. Moreover, clinical trials testing neuroprotective drugs may represent a turning point in the development of disease-modulating treatments for alphasynucleinopathies.

AUTHORS' CONTRIBUTIONS
BH and AS planned, coordinated, and edited the manuscript. All authors drafted a section of the work and revised the entire work critically for important intellectual content and gave final approval of the version to be published.

CONFLICT OF INTEREST
Author conflicts of interest in relation to the content of this manuscript.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.