Overall, insomnia, circadian rhythm disorder with early awakening, EDS due to frequent napping, nocturnal hallucinations, and confusional nocturnal wandering are frequent. In contrast, SDB and RLS seem to be no more common than in the general population of similar age. RBD is very common and may antedate the onset of dementia by several years. Detection of RBD in a patient with dementia point toward DLB because this parasomnia is rare in other forms of dementia including AD, frontotemporal dementia, and PSP.

In a large multicenter study, Bliwise et al. [2] compared nocturnal sleep disturbance between 339 patients with DLB and 4,192 with AD. Sleep problems were estimated by the informant report through the Neuropsychiatric Inventory Questionnaire (NPI-Q) item “Does the patient awaken you at night, rise too early in the morning, or takes excessive naps during the day?” Nocturnal sleep disturbance was more frequent in DLB (63 %) than in AD (27 %) and was not linked to more advanced disease, depressive symptoms, apathy, hallucinations, delusions, or agitation.

In a retrospective study, Pao et al. [3] reviewed the polysomnographic (PSG) findings of 78 DLB patients (71 male, mean age 71 years) with sleep-related complaints. Seventy-five (96 %) patients had histories of dream-enactment behaviors with 65 (83 %) showing confirmation of RBD during PSG. The remaining 13 subjects did not attain any REM sleep, and hence RBD could not be confirmed by PSG. Mean respiratory disturbance index (RDI, number of apneas and hypopneas per hour of sleep) was 12, and this was greater than 5 in 60 % and greater than 10 in 36 %. The mean periodic leg movements in sleep (PLMS) index (number of PLMS per hour of sleep) associated with arousals was 6. Sleep efficiency was less than 80 % in 72 % of the subjects, and about 75 % of the sample had isolated arousals. Among the six patients who underwent multiple sleep latency test (MSLT) two showed a sleep latency onset of less than 5 min and none showed REM sleep.

Terzaghi et al. [4] evaluated the clinical and video-PSG findings of 29 consecutive DLB patients. Patients were taking levodopa but no dopamine agonists, benzodiazepines, cholinergics, neuroleptics, or antidepressants. Patients were 21 males, their mean age was 75 years and mean disease duration was 3 years. Eleven (38 %) patients reported insomnia, 17 (59 %) EDS, 1 (3 %) had RLS, 3 (10 %) nightmares, 23 (79 %) hallucinations at night, 19 (65 %) episodes suggestive of confusional arousals, and 18 (62 %) episodes suggestive of RBD. Video-PSG showed a mean sleep efficiency of 55 %, mean RDI of 6, and mean PLMS index of 50. REM sleep without atonia was found in 46 %, RDI greater than 5 in 35 % and PLMS index greater than 15 in 61 %. Dissociated or ambiguous sleep was found in six patients who had severe dementia. Disruptive motor behaviors during sleep were found in 70 % and consisted in RBD in 11 subjects, confusional episodes from NREM sleep in 7 cases, and arousal-related episodes from REM or NREM sleep mimicking RBD in 2. Of note, these REM and NREM sleep enactment behaviors have also been described to occur in PD associated with dementia [5].

The most studied sleep disturbance in DLB is RBD. Available data indicate that in subjects with DLB, RBD is common, may be the first symptom of the disease, is associated with less AD pathology in the brain, and can be considered a red flag of the disease. RBD is very rare in AD and other forms of dementia with the exception of PD associated with dementia. Current diagnostic criteria of DLB consider RBD as a suggestive feature of the disease because “it has been demonstrated to be more frequent than in other dementing disorders” [1]. This statement was initially based on a single retrospective study involving 37 consecutive patients with dementia plus RBD [6]. Thirty-four of these patients (92 %) were male. In 35 (96 %) RBD symptoms preceded or occurred simultaneously with the cognitive complaints. Of the 37 patients, 23 fulfilled the 1996-consensus criteria for probable DLB (dementia plus at least two of the following: parkinsonism, visual hallucinations, and fluctuations), and all fulfilled criteria for possible DLB (dementia plus one of the following: parkinsonism, visual hallucinations, and fluctuations) [7]. The diagnosis of DLB was confirmed in the three patients that underwent autopsy and supported the notion that the combination of dementia and RBD most often reflects DLB. This is in agreement with neuropathological studies in patients with antemortem diagnosis of DLB plus RBD showing cell loss and Lewy bodies in the brainstem, limbic system, and neocortex [8, 9]. In a cohort of 234 autopsy-confirmed dementia patients followed longitudinally, a history of definite or probable RBD was present in 76 % of 98 with autopsy confirmed DLB, indicating that RBD is a common feature of DLB. In contrast, only 6 of the 136 patients without autopsy-confirmed DLB exhibited RBD [9]. Thus, inclusion of RBD improves the diagnostic accuracy of DLB [9]. Dugger et al. [10] compared the clinical characteristics of 71 DLB patients with RBD and 19 without RBD. Those with RBD were predominantly male, had shorter duration of dementia, earlier onset of parkinsonism and visual hallucinations, and less AD-related pathology on autopsy. In 54 of the 71 (76 %) RBD patients this parasomnia coincided or developed before dementia onset. This group of patients in whom RBD developed before cognitive impairment were characterized by earlier onset of visual hallucinations and parkinsonism, more severe baseline parkinsonism and shorter duration of dementia.

On the other hand, patients initially diagnosed with idiopathic RBD frequently are diagnosed with DLB and other synucleinopathies (mainly PD and less frequently MSA) with time. We reported that in a cohort of 44 IRBD subjects, 36 (82 %) were eventually diagnosed with a defined neurodegenerative syndrome: 14 with DLB, 16 with PD, 1 with MSA and 5 with mild cognitive impairment [11]. All 14 subjects diagnosed with DLB were men. In these subjects with DLB, recurrent visual hallucinations occurred in 13 (93 %), parkinsonism in 11 (79 %), and fluctuating cognition in 9 (65 %). Dementia was preceded by a recognized period of mild cognitive impairment characterized by executive, visuospatial and memory dysfunction. The median interval between the diagnosis of mild cognitive impairment and the diagnosis of DLB was 2 years. Median age at DLB diagnosis was 76 years, median RBD duration at the time of the diagnosis of DLB was 12 years, and the median interval between diagnosis of RBD with PSG and clinical diagnosis of DLB was 7 years.

Sleep onset insomnia and interrupted sleep are common complains. PSG studies have consistently found low sleep efficiency between 40 and 60 %, increased sleep latency and excessive sleep fragmentation, with relatively long periods of wakefulness throughout the night, sometimes without a clear cause [13–22]. However, there is a small subgroup of patients that report no sleep problems despite PSG shows dramatic fragmented sleep, severe PLMS and intense RBD. In MSA, many causes may contribute to sleep fragmentation and these include urinary incontinence, anxiety, depression, PLMS, inability to change body position in the bed because of parkinsonism, and the use of several medications. Abnormal nighttime sleep increases with progression of the disease. It is unclear, though, if interrupted sleep contributes to EDS in MSA.

EDS is frequent but in most of the cases is not a major complaint. Many variables may induce EDS in MSA but studies have not found consistent clues of which are the most relevant. Moreno-López et al. [21] evaluated EDS in 86 European MSA subjects (73 with the parkinsonian and 13 with the cerebellar subtypes) with the Epworth sleepiness scale (ESS). Mean EES score was higher (more indicative of hypersomnia) in MSA than in healthy subjects (7.72 versus 4.5). EDS (defined as an ESS greater than 10) was present in 28 % of the patients and in 2 % of the healthy subjects. EDS was associated with SDB and low sleep efficiency, and not with disease duration and dopaminergic therapy. Shimohata et al. [22] evaluated EDS in 25 Japanese patients (21 with the cerebellar and 4 with the parkinsonian subtypes) and found that the mean ESS was 6.2 and that the score was greater than 10 in 24 %. PSG detected SDB in 96 % and PLMS in 44 % but these variables were not linked to ESS score. Guo et al. [19] found a mean EES score of 8.2 in a sample of 37 subjects from China. In one small study [23] in which MSLT was performed in five MSA patients, the sleep latency was normal in four, despite poor nocturnal sleep quality. Hypocretin neurons have been found to be moderately decreased in the brain of patients with MSA [24] but hypocretin-1 levels in the cerebrospinal fluid were normal [23]. Sleep attacks induced by the introduction of levodopa have been reported in a few MSA patients with the parkinsonian subtype.

There are only a few studies addressing whether RLS is common in the backdrop of MSA and they have shown different results. An optimal strategy would be to study untreated MSA patients and to exclude other conditions with symptoms that may resemble RLS. None of the published studies, however, have excluded other forms of sensory and motor problems that are common in MSA and that may mimic the symptoms of RLS (e.g., rigidity, stiffness, central pain, dystonia, etc.). Most of these studies evaluated the presence of RLS in subjects already treated for parkinsonism with dopaminergic agents, a therapy that may mask the symptoms of true RLS. Thus, the prevalence of RLS given in these studies may have been either underestimated or overestimated. Moreno et al. [21] found RLS in 24 (28 %) of 86 subjects, 23 with the parkinsonian subtype and 1 with the cerebellar subtype, and RLS was unrelated to the amount of dopaminergic therapy. Other studies reported RLS in 3 % [20] and 12 % [16, 22] of the patients.

It is not clear whether PLMS are more frequent in MSA than in the general population of similar age. One small study comparing PLMS in untreated ten MSA patients, ten PD patients, and ten matched controls found an increase in PLMS in PD only [25]. Compared to controls, MSA patients had more PLMS (mean PLMS index of 34 versus 14), but the difference did not reach the level of significance. Overall, PSG studies commonly disclose PLMS in untreated and dopaminergic-treated patients with MSA, both with the predominantly parkinsonian and cerebellar presentations [15, 19, 20, 22]. Most patients with MSA who experience PLMS are unaware of these leg movements probably because they are generally not associated with arousals. Therefore, PLMS in MSA do not appear to be a main contributing factor for developing sleep fragmentation and EDS. In MSA, it is common to find aperiodic leg and upper limb movements in non-REM sleep which in some cases resemble visually those typical jerky and brisk movements seen in RBD during REM sleep.

A majority of patients with MSA have RBD with a prevalence of 70–100 % [13, 15, 20, 26–28]. The finding that in MSA brainstem cell loss is consistently widespread and severe may explain the high prevalence of RBD in this disease. In one study, 21 consecutive MSA patients without sleep behavioral complaints underwent video-PSG that demonstrated RBD in 19 (90.5 %) [27]. In another study, video-PSG showed RBD in 35 out of 37 (95 %) consecutive patients [13]. In our experience, all 78 MSA patients who were referred to our sleep center from April 1997 to October 2013 for different reasons (suspected RBD, stridor, or sleep fragmentation) had RBD on video-PSG. Taken together, we think that in a patient with suspected MSA, the absence of RBD (particularly if it is formally excluded by video-PSG) should seriously question the diagnosis of this disease. RBD is currently considered a red flag for the diagnosis of MSA [12].