First, because excessive daytime sleepiness greatly affects quality of life in PD patients [50], and because this is a primary symptom of SDB [51], one might expect that some sleepiness in this patient group is due to SDB [37]. Some findings support this association. For example, Shpirer et al. [39] found that PD patients who had an Epworth Sleepiness Scale (ESS) score greater than 10 had a higher AHI than those with scores below 10 (see also [42, 52]). However, when measured as continuous variables, AHI and ESS did not correlate significantly [39]. Other studies have reported no significant correlation between AHI and ESS [43, 44], and we did not observe a correlation in our 48-h protocol [29].

When sleepiness has been measured objectively using the Mean Sleep Latency Test, with few exceptions [53], there is typically no correlation between AHI and mean sleep latency [30, 40]. In the 48-h protocol, we incorporated up to eight Maintenance of Wakefulness Tests (MWTs) in which PD patients were instructed to stay awake while lying in bed for 40 min [29]. In this study, we did not observe a relationship between AHI and MWT scores. Therefore, the relationship between SDB and quality of life in PD [32] does not appear to be mediated by excessive daytime sleepiness. However, the causes of sleepiness are multifactorial (e.g., the presence of dopaminergic medications), which may explain why SDB is a weaker predictor of sleepiness in PD patients.

As previously described, common correlates of SDB in healthy controls include hypertension, cardiovascular events, and higher body mass index [13, 14]. Though studies of SDB in PD are not as well powered as those in the general community, it is still surprising that the typical (aforementioned) correlates of SDB have not been observed in studies of PD patients [42, 43, 46]. Valko et al. [54] evaluated the association between heart rate variability and presence or absence of obstructive sleep apnea syndrome in 62 PD patients and 62 age-matched controls. Their control group demonstrated several significant associations between obstructive sleep apnea syndrome and heart rate variability, particularly for the low-frequency power band and the low-frequency/high-frequency power band ratio. By contrast, heart rate variability did not correlate with SDB in the PD group. The authors suggested that there is a blunted sympathetic response to sleep breathing events in PD.

SDB has also often been connected to the development of dementia, mild cognitive impairment, or subtle cognitive effects in the general population [12]. It should be noted, however, that this literature has produced variable results, and in the large-scale, multicenter Apnea Positive Pressure Long-Term Efficacy Study (APPLES), few associations were observed [11]. Few studies have examined SDB in relation to cognition in idiopathic PD patients (though much work has been done for REM sleep behavior disorder [55]). Cochen De Cock et al. [43] found no correlation between SDB and Mini-Mental State Examination scores in PD patients. However, more recently, Stavitsky et al. [56] observed a positive correlation between actigraphy-defined sleep efficiency and an executive function composite. Though not necessarily implicating SDB, it is possible that some of the actigraphy-defined poor sleep efficiency might be due to breathing events due to breathing events or periodic limb movements [57].

There is an important distinction to be made between impaired cognition as assessed by neuropsychological testing (as a stable trait feature of PD), versus an emerging “memory consolidation” literature that connects cognitive test performance improvements to the sleep obtained between repeated cognitive tests (for review, see [58]). In the 48-h protocol, we gave PD patients eight digit span backward (executive function) tests across 2 days [59]. As illustrated in Fig. 7.1, PD patients taking dopaminergic medications demonstrated significant digit span backward improvements across the 48-h study. The improvement was not simply due to practice effects because no performance improvements were observed across the repeated daytime tests. Instead, performance improvements were localized to the nocturnal sleep interval (Fig. 7.1). The degree of digit span backward improvement was significantly associated with higher levels of slow-wave sleep and higher oxygen saturation during the night that constituted the training interval, and not during the pre-experimental night (laboratory adaptation night). PD patients who demonstrated 5 or more minutes of oxygen saturation below 90 % during the training interval night did not show significant digit span improvements. Though this (sleep) effect appears robust for repeated executive function testing, it has not been observed for motor memory testing in PD patients [60, 61]. We suggest that correcting SDB might improve the ability to acquire some new skills (as suggested by digit span backward training), which could have a positive impact on PD patients’ quality of life on a day-to-day basis (cf. [32]).

SDB might also be associated with poorer quality of life because nocturnal hypoxia could potentially exacerbate certain aspects of PD neuropathology. Basic science studies have shown experimentally that sleep apnea could cause the loss of catecholamine neurons [15] or reduce extracellular dopamine [16]. Furthermore, oxidative stress has been implicated in dopamine cell degeneration, mitochondrial dysfunction, excitotoxicity, and inflammation in PD [62, 63]. Therefore, one might expect that higher AHI or indices of nocturnal hypoxia would correlate with measures of disease severity in PD. Consistent with this idea, Efthimiou et al. [47] observed more apneas in PD patients with more severe disease (Hoehn and Yahr scale). Moreover, Maria et al. [36] found significant correlations between log-transformed UPDRS motor scores and log-transformed AHI as well as median oxygen saturation levels, even after controlling for age. Cochen De Cock et al. [43] also observed a significant correlation between AHI and UPDRS motor scores, with SDB being more frequent and severe in the most disabled PD patients. One additional study [32] observed a nonsignificant trend for higher SDB risk (Berlin Questionnaire [4]) in PD patients at more severe disease stages, and another study [42] observed significant correlations for greater sleep fragmentation and more severe disease stages, but three studies have reported no significant correlations between SDB variables and either UPDRS or Hoehn Yahr scores [41, 42, 64].

We contend that a limitation in prior studies of the clinical correlates of SDB in PD patients is that most studies have examined polysomnography in relation to disease severity in PD patients at a single time point in the course of disease. A current direction of our research is examining the association between polysomnographic variables and change in motor disease severity over the course of disease. We recently evaluated whether change in motor disease severity (UPDRS motor subscale) in 29 idiopathic PD patients was associated with SDB variables [65]. The two time points (Time 1 [T1] and Time 2 [T2]) were separated by an average interval of 265 days, and at T2 patients underwent 2 nights of polysomnographic recording. Variables derived from overnight polysomnography were not associated cross-sectionally with UPDRS motor scores at T1 or T2 in our data. Similar absence of significant findings with cross-sectional analyses has been observed previously [41, 64]. However, in our data poorer sleep was strongly associated with declining function in UPDRS motor scores (i.e., from T1 to T2). As illustrated in Fig. 7.2, there were strong correlations with mean oxygen saturation levels (but not AHI), particularly during REM sleep. The UPDRS-change correlations with mean oxygen saturation during REM sleep was not negated when controlling for time between UPDRS assessments, overt dream enactment, low REM sleep amounts, dopamine dosage, age, gender, education, years since diagnosis, cognitive status, or mean oxygen saturation while awake. Thus, nocturnal oxygen saturation levels may represent an important biomarker of change in disease severity.