Video polysomnographic findings in non‐rapid eye movement parasomnia

Abstract Although video polysomnography (vPSG) is not routinely recommended for the evaluation of typical cases of non‐rapid eye movement (NREM) parasomnias, it can aid diagnosis of unusual cases, other sleep disorders and complicated cases with REM behaviour disorder (RBD), and in differentiating parasomnias from epilepsy. In this study, we aimed to assess vPSG findings in consecutive patients with a clinical diagnosis of NREM‐parasomnia covering the whole phenotypic spectrum. Five hundred and twelve patients with a final diagnosis of NREM parasomnia who had undergone vPSG were retrospectively identified. vPSGs were analysed for features of NREM parasomnia and for the presence of other sleep disorders. Two hundred and six (40.0%) patients were clinically diagnosed with sleepwalking, 72 (14.1%) with sleep terrors, 39 (7.6%) with confusional arousals, 15 (2.9%) with sexsomnia, seven (1.4%) with sleep‐related eating disorder, 122 (23.8%) with mixed phenotype, and 51 (10.0%) with parasomnia overlap disorder (POD). The vPSG supported the diagnosis of NREM parasomnia in 64.4% of the patients and of POD in 98%. In 28.9% of the patients, obstructive sleep apnea (OSA) or/and periodic limb movements during sleep (PLMS) were identified, most commonly in older, male, sleepy and obese patients. vPSG has a high diagnostic yield in patients with NREM parasomnia and should be routinely performed when there is diagnostic doubt, or in patients where there is a suspicion of OSA and PLMS.

The current American Academy of Sleep Medicine (AASM) guidelines on the use of vPSG in parasomnia suggest that it be reserved for patients with violent or potentially injurious behaviours, cases where there is a failure to respond to therapy, or cases with an atypical or forensic presentation (Kapur et al., 2017). These guidelines do not comment on the identification of concomitant sleep pathology in patients with parasomnia, such as obstructive sleep apnea (OSA) and periodic limb movements during sleep (PLMS). Identifying these may be meaningful as they may serve as precipitating factors for NREM parasomnia behaviours, facilitating different approaches to the treatment of parasomnia patients (Fois et al., 2015;Guilleminault et al., 2005).
The aim of this study was to assess the findings of vPSG performed in a large cohort of patients with a diagnosis of NREM parasomnia with or without coexisting RBD. We aimed to evaluate the diagnostic utility of vPSG both in NREM parasomnia and POD, particularly in the identification of other sleep disorders that may serve as precipitants of parasomnia behaviours.

| Patient selection
Using an internal sleep laboratory database, we retrospectively iden-   (Berry et al., 2015). Time-synchronized video recordings were also performed, while a multidirectional microphone placed above the bed captured audio recordings. In patients requiring multiple sleep latency testing for diagnostic purposes of concomitant sleep disorders, standard guidelines were applied (Carskadon et al., 1986). PLMS, respectively (Berry et al., 2015). Additional analyses were performed using the more clinically relevant AHI cut-offs of moderatesevere OSA (i.e. AHI ≥15 events/hr) and OSA syndrome (OSAS; i.e. AHI ≥5 events/hr with daytime sleepiness). The presence of spontaneous arousals from NREM3 with typical NREM parasomnia behaviours (NREM3-A), spontaneous arousals accompanied by more subtle behaviours such as raising the head, sympathetic activation, such as tachycardia, or rhythmic delta activity on EEG (NREM-a), were recorded.
All cases of concomitant diagnosis of hypersomnia of central origin were retrospectively reviewed and diagnosis was made de novo according to ICSD-3 criteria (AASM 2013).

| Priming and precipitating factors
Medical records were searched for priming factors associated with NREM parasomnia, including sleep deprivation, stress, insomnia, hyperthyroidism, migraines, history of head injury, encephalitis and stroke, and drugs including lithium, phenothiazines, anticholinergic agents, hypnotic drugs and sodium oxybate. Sleep deprivation was captured through sleep dairies or actigraphy, and stress was recorded based on subjects' reports. The diagnosis of insomnia was made based on the existing ICSD criteria at the time of the diagnosis.
Potential precipitating factors, including external stimuli such as noise, light and bed partner's movement, and internal stimuli such as OSA and PLMS, were extracted from patients' notes and vPSG.

| Statistical analysis
Data are reported as mean ± standard deviation if not otherwise indicated. In analyses of demographic, clinical and sleep parameters between parasomnia subtypes, comparisons between groups were performed using the Kruskal-Wallis test with Dunn's multiple comparison test when needed for continuous variables, whereas chi-squared with Cramer's V product was used for comparisons between nominal variables. To estimate the likelihood of concomitant sleep-disrupting pathology (i.e. OSA or PLMS), a logistic regression model was built incorporating parasomnia phenotypes, age, gender, obesity and subjective daytime sleepiness, with the presence or absence of sleep-disrupting pathology as the dependent variable. A value of p < 0.05 was considered to be statistically significant. IBM SPSS Statistics V24.0 (SPSS, Chicago, IL, USA) was used for all statistical analysis.

| Diagnosis and priming factors
Based on clinical history, 627 patients with a possible diagnosis of NREM parasomnia or POD were identified. Of those, 512 had undergone vPSG with an ultimate diagnosis of NREM parasomnia, which could overlap with RBD ( Figure 1). The majority (n = 98; 88.3%) of the patients excluded from this report did not have any investigations, primarily (95%) because of patient preference or existing investigations from elsewhere. Sleepwalking was the most commonly identified phenotype, and the majority of patients with a mixed phenotype (23.8% of the cohort) had sleepwalking and sleep terrors. The most common NREM parasomnias within the POD group (10.0% of the cohort) were sleepwalking and confusional arousals (Table 1) Patients diagnosed with POD were the oldest among the groups.
There was a male predominance in the sexsomnia group, whereas

| Polysomnography results
No significant differences between phenotypes were observed in the vPSG sleep macrostructure (Table 3). At least one precipitating factor was identified on vPSG in 28.9% of patients (  (Fois et al., 2015), and a study of 50 sleepwalkers undergoing polysomnography found an AHI >5 events/hr in 24% (Guilleminault et al., 2005). At a population level, a diagnosis of OSA appears to confirm a nearly four-fold increase in the likelihood of frequent sleepwalking (Ohayon et al., 2012 considered, particularly in older, male, obese or sleepy patients presenting with an apparent NREM parasomnia. This is particularly the case in patients with sexsomnia, confusional arousals and POD, amongst whom there was a prevalence of physical pathology of almost 40% or above (Fois et al., 2015), and in older more obese patients with subjective daytime sleepiness. However, a direct association between concomitant sleep pathologies and NREM parasomnia-related arousals was not confirmed in our study, with parasomnia behaviours being directly preceded by sleep-disrupting pathology in only a minority of cases.
Up to 28% of patients with sleepwalking or sleep terrors have been reported to present with POD features (Schenck & Howell, 2013;Schenck et al., 1997). An updated classification included RBD-SRED and RBD-sexsomnia (AASM; Schenck & Howell, 2013). Subclinical RBD (RWA) can also coexist with NREM parasomnias (Schenck & Howell, 2013). RBD is considered a precursor of neurodegenerative diseases; RWA may also be relevant in this regard although more research is required (Boeve, 2010;Boeve et al., 2013;McCarter, St Louis, & Boeve, 2012).  (Aldrich & Jahnke, 1991;Fois et al., 2015). In the previously mentioned study by Fois et al. (2015), vPSG provided decisive diagnostic information in 33.9% of patients and was supportive of the diagnosis in a further 26.6%. Nonetheless, the evidence base in this area otherwise remains relatively immature, largely consisting of small case series or case reports (Blatt et al., 1991;Guilleminault et al., 2005;Kavey et al., 1990).
Although the exact mechanisms behind NREM parasomnias are unknown, it is thought that priming factors that increase the amount of NREM3 sleep, and precipitating factors that increase the number of arousals from deep sleep are required to coexist in a genetically predisposed patient for a parasomnia event to occur (Schenck & Howell, 2013 that the real diagnostic yield of performing vPSG in parasomnia patients may be significantly lower than our estimate. Although there are some data to support the notion that treating comorbid sleep pathology in parasomnia patients can improve control of parasomnia events (Drakatos et al., 2018;Guilleminault et al., 2005), it may be the case that the OSA and PLMS identified in this cohort are coincident phenomena rather than precipitating factors. Only a small minority of parasomnia events within our cohort were immediately preceded by either sleep-disordered breathing or PLMS; a prospective study with more rigorous attention given to relationships between sleep-disrupting pathology and parasomnia events could yield a stronger relationship. Finally, although we attempted to adjust our statistical analyses appropriately, we cannot discount the possibility that any apparently significant relationships we identified were type 1 errors contributed to by multiple comparisons.

| CONCLUSION
vPSG has a high diagnostic yield in patients with NREM parasomnia and can aid diagnosis in unusual cases and cases complicated with RBD, and in differentiation from epilepsy. Clinicians should have a high clinical suspicion for physical sleep-disrupting pathology in cases of NREM parasomnia, particularly older, male, obese and sleepy patients.

CONF LICT OF I NTEREST
None of the authors have any relevant conflicts of interest to declare.