Meta-analysis on the prevalence of REM sleep behavior disorder symptoms in Parkinson’s disease

Parkinson’s disease (PD) is the second most prevalent neurodegenerative disease with a prevalence of approximately 0.4% in people aged 65 and older, and it increases with age [1]. Apart from motor dysfunction, PD patients also suffer from a variety of non-motor symptoms (NMS), such as hyposmia, rapid eye movement sleep behavior disorder (RBD), constipation and depression [2]. However, certain non-motor symptoms might precede the development of motor symptoms by many years [3]. RBD is characterized by the loss of normal atonia of REM sleep, resulting in an apparent acting out of dream content [4]. It has been reported that RBD as a premotor feature occurred in 38% of patients who developed PD in a mean interval of 3.5 years after a RBD diagnosis [5]. The presence of RBD in prodromal PD is consistent with the Braak hypothesis which implies that it may occur as the result of underlying Lewy pathogenesis at lower brainstem before invading substantial nigra [6]. However, reported prevalence of RBD in PD patients varies greatly from 20 to 72% [7], and the prevalence of probable RBD was estimated to be about 4.9% in the general population [8]. To date, no meta-analysis has examined the magnitude of risk associated with PD and the development of RBD. We therefore conducted a meta-analysis to estimate the prevalence of RBD symptoms among patients with newly diagnosed PD compared with health controls.

The literature search yielded 749 records (include three studies from hand searching of references) after duplicates were removed. Of these, 673 records were excluded on the basis of their title and abstract. Review of the remaining 76 full articles led to 68 exclusions based on criteria described above. Finally, eight studies fulfilled our inclusion criteria and were included in our meta-analysis with a total number of 2462 PD patients and 3818 health controls (Fig. 1).

Of these, seven studies were case-control designs [14‐20], and one was a cross-sectional study [21]. Most of the studies were conducted in Europe and North America [15, 16, 18‐21], with one conducted in Asia [14], and one study conducted in Africa [17]. We divided the eight studies into two subgroups according to different RBD assessment methods described in the articles. If RBD was diagnosed with ICD (International classification of disease), RBD-SQ (RBD- screening questionnaire) or PSG, the symptom was defined as definite RBD [14, 16, 21]. If RBD was diagnosed with NMS-SQ, self-made Questionnaire or MSQ (Mayo Sleep Questionnaire), the symptom was defined as probable RBD [15, 17‐20]. Summary characteristics for all included studies are provided in (Table 1). With NOS quality criteria, all studies scored ≥ 7/9 and six of the included studies scored 8/9.

The overall prevalence of RBD symptoms in PD was 582/2462 (23.6%) compared to 131/3818 (3.4%) in control. And the pooled OR was 5.69 (95% CI 3.60 to 9.01) with a moderate heterogeneity (I² = 70.6%; p = 0.001) (Fig. 2). Subgroup analysis according to different diagnostic methods showed that the pooled OR for “Probable RBD” group was 6.08 (95% CI 3.87 to 11.93) with evidence of moderate heterogeneity (I² = 58.2%; p = 0.048) and 4.14 for “Definite RBD” (95% CI 2.21 to 7.78) with medium heterogeneity (I² = 68.7%; p = 0.041). After excluding the study of low weight [15, 19, 21], the overall polled OR was 3.54 (95% CI 2.77 to 4.52) and the heterogeneity was completely eliminated (I² = 0%; p < 0.00001). Visual inspections of the funnel plot revealed a little asymmetry (Fig. 3) and significant publication bias was detected from results of the statistical test (Begg test: p = 0.019; Egger test: p = 0.011).

The results of our meta-analysis show a definite association between PD and RBD symptoms. Since studies included were of different study design, population bases and methods of exposure ascertainment, the heterogeneity across the studies was high. To identify studies that may have contributed to the pooled heterogeneity, we removed each of the eight studies (one study at a time) and examined the magnitude and direction of the polled relative risk (RR) after the removal of each study. This approach was repeated until the I² statistic was deemed zero. Three studies [15, 19, 21] were identified as the source of heterogeneity in this study. After removing all of those studies, the polled OR was 3.54 (95% CI 2.77 to 4.52; I² statistic = 0%), and the association between PD and the risk of developing RBD symptoms was still statistically significant.

Our systematic review and meta-analysis offers a confirmation of the association between PD and a subsequent diagnosis of RBD. The pooled prevalence of RBD symptoms in PD and control is 23.6 vs. 3.4%, and the patients who had symptoms of PD carried a 3.60 to 9.00-fold increased risk for developing RBD compared to health controls. It is important to quantify the magnitude of the association between PD and subsequent RBD symptoms, as this may underpin efforts to identify higher risk participants for entry to interventional studies with neuroprotective aims.

REM sleep behavior disorder (RBD) is a parasomnia disorder characterized by repeated episodes of dream enactment behavior and REM sleep without atonia (RSWA) during polysomnography (PSG) [22]. The main suspected mechanism of RBD is a lesion in the REM sleep atonia system, which is mainly located in the pontomedullary area [23]. RBD is important as a valuable preclinical marker of PD. Firstly, idiopathic RBD converted to neurodegenerative disease at a high rate ranging from 28 to 82% [24‐27]. Secondly, many potential early marker of PD are significantly abnormal in iRBD, such as decreased striatal dopamine transporter binding, marked EEG slowing, substantial nigra hyperechogenicity and impaired olfaction [28‐30]. Lastly, RBD is also associated with some specific PD phenotypes. It is known that age, gender, disease duration, motor disability, dopaminergic drug, motor phenotype, cognition, and autonomic dysfunction are different in PD patients with RBD compared to PD patients without the disorder [7].

Our study demonstrates that there is potential for publication bias based on the funnel plot and significant publication bias was detected from results of the statistical test (Begg test: p = 0.019; Egger test: p = 0.011). This might result in an overestimation of the association between PD and the risk of RBD symptoms. Considering that we only included eight studies which were assessed via means of the NOS, and all studies included in the main analysis had scores ≥ 7/9, we can conclude that the risk estimate that resulted from this analysis may also be viewed as a fairly strict estimate, and a result of the pooling of data from only the highest quality studies.

Similar to any other meta-analysis of observational studies, our study is subject to several limitations. First, we limited our search to publications in English and therefore might have missed a small number of relevant publications. Second, recall bias may affect the quality of information retrieved from participants in case-control studies and the assessment of RBD symptoms may vary from one study to another. Third, more high quality and large sample sizes observation studies estimating the risk of RBD symptoms in PD should be included in order to avert the publication bias.

In conclusion, despite these limitations, we pooled data from 6280 people across eight studies to provide a consolidated risk that estimated relating PD to the development of RBD symptoms. Most of the evidence was of the highest quality and the conclusion was considered robust. Further prospective research is needed to examine the presence of RBD symptoms and its relationship to PD, in order to better understand the etiology and natural history of PD.