2.1 Search strategy

Studies were located by searching PubMED (until 10 June 2022) and PSYCArticles (from 1967 until 10 June 2022) databases using the terms (“dementia with Lewy bodies” OR “dementia with Lewy*” OR “Lewy*” OR “Parkinson's disease dementia” OR “Parkinson's disease with dementia” AND (“sleep” OR “sleep quality” OR “sleep disorder” OR “sleep disturbances” OR “sleep deficit” OR “sleep impairment” OR “RBD” OR “REM behavior disorder” OR “REM behaviour disorder” OR “rapid eye movement behaviour disorder” OR “rapid eye movement behavior disorder” OR “insomnia” OR “parasomnia” OR “restless legs syndrome” OR “Willis-Ekbom disorder” OR “periodic limb movement” OR “periodic limb movement syndrome” OR “periodic limb movement disorder” OR “nocturia” OR “sleep apnoea” OR “sleep disordered breathing” OR “obstructive sleep apnoea” OR “central sleep apnoea syndrome” OR “circadian rhythm” OR “circadian rhythm sleep disorder” OR “excessive daytime sleepiness” OR “hypersomnolence” OR “somnolence” OR “sleepiness”)).

2.2 Eligibility criteria

Studies were eligible if participants had DLB/PDD, diagnosed in line with recognised criteria (or previous versions of this diagnostic criteria).^{2, 17} Studies were included if they: (a) focussed on the prevalence, aetiology and/or treatment of sleep disturbances; (b) compared subjective/objective sleep between DLB and/or PDD, non-LBD neurodegenerative dementias, and/or healthy non-dementia groups; (c) examined associations between sleep and behavioural/neuropsychiatric symptoms; (d) examined if sleep/sleep disorders predicted LBD diagnosis; e) assessed if sleep distinguished LBD from other neurodegenerative dementias or were; (f) were treatment studies with the primary/secondary aim of improving any aspect of sleep or sleepiness (including case study/clinical trial designs). Unpublished studies and pre-print articles were not sought, but were considered if relevant or referenced in eligible studies.

Identified articles were excluded if they were: (a) duplicate; (b) review articles; (c) non-English; (d) concerned with non-LBD dementias, mild cognitive impairment, or individuals without dementia; (e) in abstract format; (f) opinion-based letters; (g) neuropathological studies, unless relevant to sleep; (h) RBD studies assessing non-LBD dementia conversion, or (i) animal studies.

2.3 Data extraction

Eligible articles were exported to EndNote X9.3 (Clarivate, London, UK) and abstracts were screened. The methodological quality of studies, in relation to their main aims, were rated from 0 to 5 (representing the highest methodological quality) using the Mixed Methods Appraisal Tool (MMAT^{18, 19}). The first 10% of identified papers and MMAT evaluations were checked by another member of the study team (PA-M) to verify the search strategy.

2.4 Narrative synthesis

Due to the heterogeneity of studies and methodological approaches (Table 1), these data were not amenable to meta-analysis. Therefore, a narrative description is provided, in line with Synthesis without Meta-analysis (SWiM;²⁰) guidelines.

The studies were grouped as follows: (1) subjective sleep: where the primary aim of the study was to investigate the prevalence or nature of LBD subjective sleep alterations; (2) objective sleep; (3) EDS; (4) RBD, where the aim was to assess the frequency or prevalence in LBD and/or quantify the later conversion to LBD; (5) circadian rhythms, where these were assessed subjectively or objectively (i.e., measuring actigraphy, core body temperature, or other circadian rhythm markers); (6) other sleep symptoms, including obstructive sleep apnoea, sleep-disordered breathing or restless legs syndrome/periodic limb movements (RLS/PLMs); and (7) treatment studies.

3.1 Characteristics of studies

There was a large amount of heterogeneity regarding the main study focus and sleep methodologies employed (Table 1). The included studies (Tables 2 and 3) were conducted between 2000 and 2022. The number of participants ranged from one (single-case studies) to 4531. Fifty-six studies recruited only DLB,^{12, 21-75} 12 studies recruited both people with DLB and PDD,^76-87 and two recruited only PDD patients^{88, 89}; these figures include studies that investigated the conversion from idiopathic RBD (iRBD) to DLB or PDD.

Twenty studies investigated subjective sleep,^{12, 21-31, 76-82, 88} where 12 used participant, caregiver or informant questionnaires^{12, 21-23, 25-28, 76, 81, 82, 88}; two relied upon clinical evaluations and participant or informant questionnaires,^{79, 80} two relied upon a clinical evaluation^{24, 78} and one study each used a retrospective review of clinical records,²⁹ patient, family or caregiver reports,³⁰ self-report patient sleep duration estimates⁷⁷ or subjective sleep diaries.²²

Seven studies focussed on objective sleep^{32-36, 47, 75, 81}, where five studies used one night of in-laboratory video PSG^{32, 33, 35, 36, 47}; three retrospectively examined PSG records.^{32, 33, 35} One study examined objective sleep in DLB patients who underwent routine clinical visits⁴⁷ and another compared objective sleep between DLB and PD.³⁶ One study used actigraphy, alongside questionnaire and caregiver sleep diaries³⁴ and finally, one study measured objective sleep using a non-wearable commercially available movement sensor, which is placed under a mattress.⁷⁵

Eight studies assessed EDS.^{37-43, 89} One study solely used subjective questionnaire measures.³⁸ Four used subjective measures.^{38-40, 42} Five studies objectively measured cerebrospinal fluid (CSF) levels of the neuropeptide hypocretin-1 (orexin)^{37, 41-43, 89} and two examined the potential underlying neuropathology of EDS, by obtaining neuronal counts within the nucleus basalis of Meynert (nBM), or measuring neocortical hypocretin.^{40, 41} Two used objective measures of sleep and sleepiness (PSG and MSLTs),^{39, 89} where only a subset of patients completed MSLTs.

Of the 14 RBD studies,^{44-46, 48-55, 83-85} eight examined the number of iRBD patients who were subsequently diagnosed with DLB or PDD.^{46, 50-54, 83, 84} Two assessed differences in clinical symptoms, the underlying neuropathology or brain metabolism, in patients with and without RBD,^{45, 48} and one compared sleep-enactment behaviours (including RBD) between LBD and PD.⁸⁵ One retrospective study examined RBD prevalence in DLB patients who attended a sleep laboratory.⁴⁴ Finally, one study examined associations between subjective RBD and CSF hypocretin-1⁴⁹ and a further neuropathological study examined the association between DLB cholinergic pedunculopontine and laterodorsal tegmental nuclei (PPT/LDT) counts, and RBD symptom presence.⁵⁵ All but three RBD-focussed studies used PSG.^{48, 49, 55}

Only one study assessed circadian rhythms,⁵⁶ where DLB core body temperature was measured alongside PSG.⁵⁶ Three studies focussed on PLMs or RLS^57-59 using clinical evaluation and subjective measures,⁵⁷ or objective PSG measures of PLMs.^{58, 59} The latter study also measured aspects of sleep-disordered breathing.⁵⁹

There were 17 specific treatment studies^{60-74, 86, 87} (Table 2); additionally, one open-label study was conducted as an additional component of a DLB observational study.²⁶ Of the 18 treatment studies, 12 used pharmacological interventions,^{26, 61-66, 69-71, 74, 87} including rivastigmine,²⁶ melatonin,⁶¹ galantamine,⁶² gabapentin,⁶³ donepezil,^{64, 69, 71, 74} armodafinil,^{65, 66} memantine,⁸⁷ clonazepam⁶⁹ and levodopa.⁷⁰ Six used predominantly non-pharmacological interventions including bright light therapy,^{60, 72} deep brain stimulation (DBS) of the nBM,⁶⁷ or the herbal remedies yokukansan^{73, 86} and Yokukansankachimpihange.⁶⁸ Most (n = 8) treatment studies were open-label pilot studies^{26, 62, 64, 65, 68, 70, 72, 86}; 5 were single case reports,^{60, 63, 71, 73, 74} two were multiple case reports^{66, 69} and one was a retrospective patient clinical review.⁶¹ There were only two double-blind randomised controlled trials (RCTs), of memantine and DBS^{67, 87}; however, sleep was not the primary outcome measure in either: subjective RBD and sleepiness were secondary outcomes in an existing trial of memantine⁸⁷ and the primary aim of the DBS study was to assess the safety and effects upon episodic memory.⁶⁷ The quality of the treatment studies ranged from very poor to moderate (Table 3).

4.1 Nature of sleep disturbances

Overall, these results indicate that sleep disturbances are highly prevalent in LBD,^{24, 79, 80} as despite the heterogeneity in the method of measurement, at least one form of sleep disturbance is present in up to 90% of LBD patients⁷⁹ (Table 4). These results also indicate that subjective sleep disturbances, including poor sleep quality, insomnia, EDS, sleep apnea and RLS, are more frequently observed in LBD than in other dementias, including AD.^{21-23, 25, 28-31, 76, 79-82} LBD sleep disturbances are also more severe, in terms of worse sleep quality and greater daytime sleepiness, relative to other neurodegenerative dementias, including AD.^{22, 23, 26, 76} However, it should be noted that true prevalence studies are needed to confirm these estimates.

It is also highly likely that these sleep disturbances can have a corresponding impact upon the caregivers of people with LBD: in one study, 94% of DLB caregivers reported that patient sleep disturbances were a stressor, compared to 40% of AD caregivers.²⁶ Although only three studies directly compared sleep between DLB and PDD groups,^{76, 77, 80} two indicated that neither subjective sleep quality or EDS differed between DLB and PDD,^{76, 80} and one study demonstrated that the estimated nocturnal subjective sleep duration, estimated in the number of hours per day, was greater in DLB compared to PDD (9.62 h compared to 8.07 h).⁷⁷

Rapid eye movement behaviour disorder, both in terms of probable RBD, which has been measured using subjective methods (e.g. questionnaires), or confirmed RBD, which has been diagnosed using PSG, was found to be especially common in DLB and PDD^{22, 24, 25, 29-31, 44, 47, 79-81, 85}: for instance, one PSG study showed that approximately 90% of DLB patients demonstrated RBD, which was defined in accordance with current diagnostic criteria.⁴⁴ Although one limitation is undoubtedly that PSG studies may be biased towards patients with more severe RBD symptoms, despite this, the identified subjective and cross-sectional studies still clearly demonstrate high rates of RBD are apparent in DLB (approximately 80% of individuals).^{31, 79} Also of note is that an iRBD diagnosis appears to increase the risk of the subsequent development of DLB or PDD.^{46, 50-54, 83, 84} Whilst a weakness of the studies which longitudinally assess iRBD patients is that they typically rely on small sample sizes and variable follow-up duration periods, the study with the largest number of iRBD patients who later developed DLB (n = 32) found a 16% conversion rate from iRBD to LBD at a median 5-year follow-up time point.⁴⁶ The risk of dementia development following an iRBD diagnosis may also increase with time: in one study, 19% of iRBD patients developed DLB or PDD at a mean follow-up of 5.1 years,⁸³ and this rose to 32% by 10.5 years.⁵⁰

Although only five studies have examined objective sleep in DLB using PSG,^{32, 33, 35, 36, 47} these are suggestive of clear impairments to objective sleep. All five studies demonstrated that objective sleep efficiency (SE%), which is a marker of sleep quality, was extremely low in DLB; for example, in one study, patients had a mean SE% value of 56%,³³ which is well below the expected normative values (approx. 80%) for healthy older adults.⁹⁰ High levels of wake after sleep onset (WASO), which indicates the total duration of overnight awakenings, were also apparent from the identified studies.^{33, 36} In one study, relative to normative control data, DLB patients showed altered sleep continuity and architecture including lower SE% values, total sleep and REM sleep duration, alongside increases in the time taken to get to sleep, WASO, and stage 2 sleep duration.³² Compared to PD, three studies found that people with DLB took longer to get to sleep (SL), and had a lower sleep duration, compared to PD.^{32, 33} However, two other studies which have used PSG as a secondary measure, observed no differences in objective sleep between DLB or other groups including AD, PD, iRBD and non-dementia control groups.^{39, 56} However, as impairments to objective sleep are commonly observed in neurodegenerative diseases,⁹¹ more studies are needed to confirm whether or not if there are clear DLB-specific alterations in terms of objective sleep continuity or architecture.

Despite the fact that comparatively little work has focussed on objective sleep continuity and architecture, one study showed that sleep architecture alterations were present in 75% of DLB patients, including occipital slowing, delta activity during REM, or the absence of sleep spindles and/or K-complexes.⁴⁷ These abnormalities were not due to technical failures and prevented the accurate classification of overnight brain activity into different stages of sleep.⁴⁷ Two PSG studies also showed greater levels of PLMs in DLB relative to other neurodegenerative conditions, which can in itself be disruptive to objective sleep.^{58, 59} Polysomnography has also been used to demonstrate in the context of a specific test of objective sleepiness (an MSLT), that objectively, DLB patients show greater levels of sleepiness relative to AD,³⁹ but that there is no difference between DLB and PD or PDD.⁸⁹ Taken together, these results very speculatively suggest that objective sleep architecture alterations could be apparent in DLB and are worthy of further investigation.

4.2 Mechanisms of LBD sleep disturbances

The underlying mechanisms of alterations to sleep, and sleep disturbances are unclear from the identified literature, as only five studies included neuropathological evaluations alongside sleep or sleepiness assessments^{40, 41, 45, 55, 78}; therefore, there is a clear need for future work to address this specific knowledge gap. Two EDS neuropathology studies conducted neuronal counts of the nBM and assessed neocortical hypocretin levels, respectively.^{40, 41} The first study showed that nBM neuronal counts were lower in DLB than in controls and that DLB patients who displayed EDS had lower nBM counts than those without EDS.⁴⁰ The second study found reductions in neocortical hypocretin fibres and neurons in DLB relative to AD and control participants.⁴¹ Since the nBM has a key role in sleepiness, and the hypocretin/orexin system is wake-promoting,⁹² speculatively, this could indicate that the presence of EDS in Lewy body dementia is potentially due to LBD-specific alterations to these key wake-promoting neurotransmitter systems.

Despite RBD being an extremely common symptom within DLB, the identified literature also showed that the patho-aetiological investigation of this symptom is currently very limited: in a comparison of DLB patients who exhibited RBD, and those who did not, there were no neuropathological differences between both groups.⁴⁵ One further study found a reduced cell count in the PPT/LDT tegmental nuclei, which are involved in the control of REM sleep, in DLB, relative to controls.⁵⁵ The study concluded that cell losses in this region were not associated with the presence or absence of RBD; however, a major limitation is that very few DLB patients (n = 5) had RBD.⁵⁵ Given that cholinergic neurons in the PPT and LDT have a key role in REM sleep control,⁹² this area is certainly worthy of further investigation. Finally, it should be determined if sleep alterations can themselves lead to neuropathological changes: one study observed that the presence of retrospectively evaluated clinical sleep disturbances were predictive of Lewy body neuropathology at autopsy,⁷⁸ although it was not stated which sleep disturbances were predictive of LBD neuropathology.

Several studies have examined the potential biological or neurobiological correlates of sleep and sleepiness, by measuring relevant biological, circadian, or brain markers.^{37, 42, 43, 48, 49, 89} However, all of the identified studies demonstrate substantial heterogeneity in terms of the experimental methods used and rely on small sample sizes (Table 2). These are discussed in the following section.

Five studies have examined the neurobiological correlates of EDS.^{37, 42, 43, 49, 89} These studies examined CSF hypocretin-1 levels, and found no differences in LBD relative to non-dementia controls or other neurodegenerative disease groups.^{37, 42, 43, 49, 89} Despite the key role of hypocretin-1 in excessive sleepiness, this would suggest that specific neurobiological alterations may not be the main driver of this common symptom. Intriguingly, one pilot study of circadian measures suggests that alterations to circadian rhythms, which refers to the approximately 24-h oscillatory rhythms exhibited by the body's various physiological and behavioural processes and which exert a significant influence over the timing and quality of sleep,⁹³ might be a feature of DLB.⁵⁶ In this highly-controlled sleep laboratory study in a small group of patients, reductions were observed in core body temperature amplitude (i.e. where the lowest point of the body temperature has been subtracted from the temperature activity), which was measured from the evening until the following morning, in DLB relative to controls, PD, and RBD groups.⁵⁶

Only two imaging studies have been conducted, where both used 18F-fluorodeoxyglucose position emission tomography (FDG-PET). In a study of DLB patients with probable RBD, these patients displayed greater levels of reduced metabolism in a number of regions (e.g. dorsolateral medial frontal, left precuneus, bilateral superior parietal lobule, rolandic operculum and amygdala regions),⁴⁸ which the authors speculated reflects a more severe degeneration of the cholinergic system relative to DLB patients without RBD. In the other imaging study, glucose uptake of the left pulvinar of the thalamus in DLB was positively associated with actigraphically measured total sleep time, and negatively correlated with activity during sleep.³⁴ The thalamus has a key role in the control of sleep-wake cycles,⁹⁴ and as thalamic dysfunction is a feature of DLB, and may have a causative role in the core symptom of cognitive fluctuations,^95-97 the role of the thalamus in DLB-specific sleep alterations should be explored further.

Finally, several of the identified studies would indicate that sleep, sleepiness, and other LBD related symptoms are associated.^{12, 49, 76, 88} For instance, two studies found that greater subjective excessive daytime sleepiness was associated with worse motor symptoms in PDD⁷⁶ and worse subjective depression in DLB.¹² Other studies have found associations between sleep problems (assessed using the neuropsychiatric Inventory) and cognition and motor symptoms,⁸⁸ and between subjective sleep quality and subjective depression.¹² Finally, one study showed that hypocretin-1 levels were positively associated with subjective RBD, and night-time sleep disturbance severity, in DLB.⁴⁹ Taken together, this would suggest that there is a complex interplay between sleep, sleepiness and the neuropsychiatric and motor symptom profile of LBD. However, due to the wide heterogeneity in the identified studies, the exact nature of the association between sleep, sleepiness and LBD symptoms are as yet unclear, and this should be investigated further. Well-designed studies are needed to explore the underlying mechanisms behind these associations, and clarify if there is a direct casual mechanistic link between specific sleep alterations, or sleep disturbances, and particular aspects of the LBD symptom profile, since this may lead to new therapeutic opportunities.

4.3 Treatment of LBD sleep disturbances

Despite the clinical relevance and ubiquity of sleep disturbances in LBD, it is perhaps very surprising that few treatment studies have been conducted. In particular, the identified literature demonstrated that there is a dearth of high-quality RCTs in LBD. Of the 18 treatment studies identified,^{26, 60-74, 86, 87} only two were double-blind RCTs^{67, 87}; neither of these RCTs had the primary aim of improving sleep or sleep-related symptoms. Despite the low quality of the identified studies, and the fact that the majority were open-label pilot studies with small sample sizes,^{26, 62, 64, 65, 68, 70, 72, 86} or case reports,^{60, 63, 66, 69, 71, 73, 74} several interventions are of interest and may warrant further investigation, ideally in the context of a sufficiently powered, high-quality, RCT.

Of the open-label studies, in terms of DLB pharmacological interventions, the identified results indicated that rivastigmine and galantamine may benefit subjective sleep quality,^{26, 62} and that donepezil may improve nocturnal sleep disturbances and fragmented objective sleep.⁶⁴ Rivastigmine and armodafinil were shown to reduce subjective EDS and armodafinil also objectively improved the ability to stay awake.^{26, 65} In the only double-blind RCT of a pharmacological intervention, memantine improved subjective RBD in DLB and PDD, albeit as a secondary outcome measure.⁸⁷ Mechnicanistically, armodfinil is wake-promoting⁶⁵ and whilst acetylcholine has a role in the complex process of sleep and wake regulation,⁹⁸ future work should investigate whether cholinesterase inhibitors directly benefit sleep, and the mechanistic reasons for doing so, or if the benefit observed to sleep is secondary due to improvements in dementia symptom severity.^{26, 62, 64}

Of the non-pharmacological interventions, the Japanese herbal intervention yokukansan improved LBD insomnia symptoms, although this was measured using an AD insomnia subscale⁸⁶; similarly, the related yokukansankachimpihange herbal intervention reduced subjective nocturnal disturbance behaviour frequency, although these were open-label studies.⁶⁸ An open-label study showed that bright light therapy did not improve DLB sleep disturbances, but did so in AD⁷²; however, a single case study in DLB suggested that light therapy improved subjective daytime sleepiness and reduced subjective sleep disturbances.⁶⁰ Finally, deep brain stimulation of the nBM did improve sleep and daytime sleepiness in DLB; however, the primary aim of this study was to examine the safety and efficacy of this procedure upon cognitive outcomes, and there are outstanding concerns regarding the adverse effects of this procedure.⁶⁷ The underlying mechanisms of these interventions are less clear relative to pharmacological interventions, although light is a very important cue for sleep timing and quality,⁹⁹ and nBM stimulation may be beneficial due to this region's key role in regulating sleepiness.⁹²

4.4 Future research directions

Overall, as these results indicate that there is a very high prevalence of sleep disturbances and EDS in Lewy body dementia, and that they are complex in nature. Importantly, our results indicate that the majority of the studies have focussed on DLB.^{12, 21-75} Whilst we have identified clear knowledge gaps in a range of areas (e.g. whilst there is a high prevalence of subjective and objective sleep disturbances in LBD, their underlying mechanisms are unclear), currently, very little is known if there are specific differences between DLB and PDD in terms of sleep and more comparative studies should investigate this; it is possible that there may be differing symptom profiles and treatment responses with regards to sleep between both groups.

There are a number of other specific future directions. Firstly, high-quality LBD-specific treatment studies are urgently needed for these symptoms. From the identified literature, the quality of treatment studies appears to be weak and is primarily limited to poor-quality open-label designs with insufficient statistical power. Current clinical management guidelines for LBD sleep disturbances mainly include the use of sleep hygiene, or pharmacological agents such as clonazepam, melatonin and memantine for RBD¹⁰⁰; however, some agents are associated with negative side effects, which can impact on cognition and cause further sleepiness.⁷ Despite their clinical use, high-quality RCTs of these agents are yet to be undertaken. This is important because the current evidence base which is used to inform the treatment of sleep disturbances in LBD is very weak, and based on the findings from PD and iRBD, rather than LBD populations.⁷ It is particularly surprising that there were no trials which assessed melatonin in LBD. Melatonin is considered to be a first-line treatment for symptomatic RBD, and whilst the exact mechanism of action is not well-understood, it does have a favourable safety profile.¹⁰¹ Given the high preponderance of RBD in Lewy body dementia patient groups, this agent is particularly worthy of further investigation. Additionally, given that neuroleptic sensitivity and polypharmacy is an issue in LBD^{102, 103} non-pharmacological behavioural methods including sleep hygiene, exercise or increased ambient light exposure should be trialled; these techniques have certainly shown promise in other neurodegenerative conditions.¹⁰⁴

Secondly, the exact nature of the potential link between sleep, sleepiness, and LBD behavioural and neuropsychiatric symptoms^{12, 34, 88} remains to be elucidated. For instance, two studies observed a link between subjective sleep, EDS, and subjective depression, and between objective total sleep time, nocturnal activity, and hallucinations, respectively.^{12, 34} It is not clear if LBD directly affects sleep, or EDS, which may in turn, exacerbate particular symptoms; if there is a bi-directional sleep and symptom link, or a common underlying pathway. For instance, in the case of depressive symptoms and daytime sleepiness, both might be driven by LBD-specific noradrenergic deficits.^{105, 106} This should be examined further as if there is a bi-directional link between sleep, EDS and behavioural and neuropsychiatric symptoms, then this may provide a therapeutic opportunity: it is possible that by improving sleep quality or timing, this may also improve behavioural and neuropsychiatric symptom severity. Depression is likely to be one such target, as depression is particularly common in DLB, with a prevalence of approximately 60%,¹⁰⁷ and there is a very strong bi-directional relationship between sleep, the clinical sleep problem of insomnia disorder, and depression.¹⁰⁸ As an indication of the potential therapeutic applications in LBD, a recent clinical trial showed that in older adults without dementia, who had comorbid insomnia disorder and depression, the non-pharmacological treatment of cognitive behavioural therapy for insomnia improved both insomnia and depression severity.¹⁰⁹

One particular knowledge gap, which is worthy of further investigation, is regarding the potential association between EDS and cognitive fluctuations. Speculatively, this may be due to LBD-specific alterations to circadian rhythms, as cognitive performance across multiple domains and the underlying neural activity shows a marked circadian modulation.^110-113 Typically, the circadian drive for sleep is highest in the early morning hours¹¹⁴ and one pilot study has shown that alterations to circadian rhythms may be a feature of DLB.⁵⁶ Clinically, the frequency of cognitive fluctuations varies considerably: cognitive fluctuations have been reported to occur over short periods (e.g. minutes or hours) and longer periods (e.g. daily).^{115, 116} Although it is well-established that the timing and duration of sleep, and sleep deprivation, has a direct impact upon cognitive performance,¹¹⁷ the contribution of sleep and sleep disruption to cognitive fluctuations is currently less clear, despite multiple brain regions which are relevant to sleep (e.g. the thalamus) being involved in this symptom.^{118, 119} As immediate fluctuations may reflect LBD-specific impairments to information processing, or transient cortical or subcortical synaptic disturbances,¹²⁰ future work should attempt to clarify the precise impact of sleep upon this poorly understood symptom. Intriguingly, one pilot study of circadian measures suggests that alterations to circadian rhythms might be a feature of DLB: one highly controlled sleep laboratory study in a small group of patients found reductions in core body temperature amplitude (i.e. where the lowest point of the body temperature has been subtracted from the temperature activity), which was measured from the evening until the following morning, in DLB relative to controls, PD, and RBD groups.⁵⁶ Given the rhythmicity of cognitive fluctuations,^{115, 116} the potential mechanistic role of circadian rhythm dysregulation, and the circadian interaction with sleep, should be examined further in relation to this core symptom.

Similarly, the natural history of sleep disturbances in LBD is yet to be determined: post mortem neuropathological investigations, alongside a detailed longitudinal assessment of subjective and objective sleep, may help delineate the specific pathoaetiological contributions. One particularly important area for future research is to examine the link between subjective sleep quality and continuity, or sleep disturbances, and cognition. The impact of poor sleep upon cognition is well established: multiple domains of cognition, including sustained attention, working memory, and decision-making, are negatively impacted by insufficient sleep duration and quality.^121-123 Additionally, evidence now increasingly shows that sleep disturbances are mechanistically involved in neurodegeneration.^{11, 124}

Thirdly, relatively few studies have objectively measured sleep using PSG. Although those studies which have been conducted suggest LBD-specific disruptions to sleep continuity,^{32, 33, 35, 36, 47} there are potential methodological confounders, since a particular issue is that in all studies, sleep was only measured on one night. This is problematic as alterations to objective sleep continuity and architecture are consistently observed during the first night of a PSG study, potentially due to the unfamiliarity of the sleep laboratory environment, and subsequent nights are therefore more representative of normal sleep.^{125, 126}

Polysomnography is also relevant as objective sleep continuity and architecture is of increasing importance: studies in healthy individuals and AD suggest that disruption to specific stages of sleep influences neurodegeneration.^{127, 128} Notably, the link between sleep architecture and cognition has not been examined in LBD; two relevant and specific non-rapid eye movement sleep (NREM) features are sleep spindles and slow-wave sleep; both of which reflect thalamocortical network integrity.^{129, 130} Sleep spindles are particularly relevant as they originate in the thalamus, and reduced sleep spindles in PD predict subsequent PDD.^{129, 131} These NREM measures might influence or reflect cognitive symptom severity: sleep spindles are strongly associated with plasticity, learning and memory.^132-134 One DLB case study found that donepezil increased sleep spindle activity⁷¹; if NREM measures are associated with cognition, sleep spindle manipulation may also benefit LBD cognition. Additionally, sleep spindles have a protective role against sleep disruption¹³⁵ and an increase in spindles may also improve sleep quality.

Future studies should consider home sleep measurement, which enables sleep information to be collected over longer time periods, as this might provide a greater insight into the impact of sleep upon symptoms or disease progression. For instance, home-based PSG recording, which has been used to examine sleep architecture in dementia caregivers¹³⁶ may be cheaper and more tolerable for patients than attending a hospital or research sleep laboratory.¹³⁷ Alternative sleep measurement methods might include behind-the-ear EEG electrodes, which are easy to apply, less intrusive than PSG, and show a good level of agreement with PSG.¹³⁸ Finally, actigraphy, as a non-invasive method of measurement, could be used to assess natural sleep/wake patterns over an extended period of time. LBD symptoms may have a circadian element, or be driven by sleep-related brain regions: cognitive fluctuations are periodic in nature and may have a thalamocortical basis^{116, 139}; actigraphy may provide an insight into this symptom.