Abstract
In previous studies, idiopathic RBD was associated with a different risk factor profile from PD and dementia with Lewy bodies, suggesting that the PD-RBD subtype may also have a different risk factor profile.
INTRODUCTION
Rapid eye movement sleep behavior disorder (RBD) is a parasomnia characterized by loss of REM atonia and a corresponding acting out of dream material [1]. Individuals suffering from RBD tend to be male with symptom onset in later middle age. Idiopathic RBD is a very strong prodromal marker of synuclein–mediated neurodegenerative disease, including Parkinson’s disease (PD), dementia with Lewy Bodies (DLB), and multiple system atrophy. Approximately 30–50% of PD patients have associated RBD [2]. Many recent studies have demonstrated that RBD marks a subtype of disease, characterized clinically by increased cognitive impairment / dementia, more autonomic dysfunction, and a tendency towards more akinetic-rigid features [3–7]. Pathologic studies have suggested that patients with RBD have a more diffuse and severe deposition of synuclein [8].
Recently, a comprehensive study of epidemiologic risk factors for idiopathic RBD versus controls was conducted [9–12]. This study found that patients with idiopathic RBD had a risk factor profile that differed from PD. Although like PD, pesticide exposure and head injury was more common, well-established risk factors thought to be protective for PD such as caffeine use and smoking, and putative risk factors such as calcium channel blocker use, education level, and cardiovascular disease differed substantially from what is known about PD. Moreover, prospective follow-up of the cohort demonstrated that caffeine and smoking were not risk factors for conversion to PD/DLB, but family history of dementia, motor symptoms, and autonomic symptoms predicted outcome [13]. One potential explanation for these findings is that RBD-PD is a unique subgroup of PD, with its own risk factor profile. If so, adjusting for subgroups of PD may be essential in properly analyzing studies of environmental risk factors in PD. However, there has been no study of epidemiologic risk factors in PD stratified to the presence of RBD. Therefore, we conducted a multicenter assessment of environmental risk factors in PD patients who were referred for clinical assessments, stratified to with and without polysomnography-proven RBD.
METHODS
Patients
Patients were selected (convenience sampling) from three academic centers [4]. The ethics boards of all centers approved the study and informed consent was obtained from each patient. All eligible patients had idiopathic PD. All participants completed a polysomnogram (one night) to confirm presence or absence of associated RBD. Diagnosis of RBD was made according to the International Classification of Sleep Disorders as a loss of REM atonia concomitant with either a history of dream enactment, or dream enactment observed during REM sleep on polysomnography. Each laboratory used their own existing clinical protocols for definition of REM atonia; the Mayo Clinic site used clinical visual analysis without specific quantification (which would correspond to REM atonia loss of >20%), the Montreal site used the definitions in the 2010 Montplaisir study (i.e. 15% tonic and/or 30% phasic) [14] and the Montpellier site used a cutoff of ‘any’ tone of 20% . History of dream enactment was elicited upon clinical interview at the time of polysomnography; however, because the risk factor/symptom questionnaires were administered 1–5 years after the polysomnography, an additional screen for RBD symptoms was added to the questionnaire. Patients initially classified as RBD negative were excluded from the study if they reported development of dream enactment behavior since the time of last polysomnography.
Questionnaire administration
A structured questionnaire was mailed to 425 participants after initial telephone confirmation of willingness to participate. Participants were encouraged to have family members, friends or caregivers assist them in completion of the questionnaire if details were difficult to remember. In addition, contact information for the researchers was provided so participants could ask for any necessary clarification by telephone. If questionnaires were not returned, a follow-up phone call with resend was conducted.
The questionnaire was similar to our recently published risk factor questionnaire for idiopathic RBD [9–12, 15]. Areas of interest included: Demographic Variables: Age, Sex, and Level of Education. Medical Variables Other illnesses and comorbidities- from a list [11] supplemented by open-ended self-report. Atherosclerotic risk factors. Head trauma, including any head trauma with loss of consciousness. Medication usage - all past medications that were taken regularly (i.e. more than 3 years), as well as current medications [11]. Overall self-perceived health status, assessed on a 5-point Likert scale. Family History of RBD, PD, Alzheimer disease, other dementias, and atherosclerotic disease. Environmental and Lifestyle Variables. Pesticide exposure. Rural living and well water usage. Occupation-assessed by open-ended occupation list as well as by direct query of potential PD-linked occupations. Smoking-including amount, duration and type of smoking (cigarettes, cigars, etc.). Caffeine intake including type of caffeine intake (coffee, tea, colas), amount of use, duration of use. Alcohol intake–including intake frequency, amount, and frequency of binge drinking.
Statistical analysis
The primary statistical analysis was a comparison via logistic regression between participants with vs. without RBD, adjusted for age, sex, center, and disease duration. Given that this is an exploratory study, no adjustment for multiple comparisons was used [16].
RESULTS
Demographics, lifestyle and environmental factors
243/425 (57%) of the questionnaires mailed were returned; 7 were not complete enough to include,leaving 236 participants. 126 were from Rochester, 23 from Montreal, and 87 from Montpellier. 98 (41.4%) had RBD based on polysomnography and 138 did not. Among the 138 participants without RBD, 41 endorsed possible recent-onset dream enactment behavior on the questionnaire; these were removed from the analysis. Therefore, we included 189 patients, 98 with RBD, and 91 without.
RBD was somewhat less common among women (OR = 0.53 (0.28–1.01)), although this difference was not significant. Age was similar, but PD disease duration was longer in PD-RBD patients (9.8 + /–6.8 years vs. 7.4 + /–5.7 years, OR = 1.07 (1.02–1.13)) (Table 1).
Among lifestyle factors, PD-RBD patients were more likely to have been regular cigarette smokers (47.4% vs. 28.6% , OR = 1.96 (1.04–3.68)). However, alcohol and caffeine intake were not different between groups. In terms of occupation, PD-RBD patients were more likely to report history of a welding occupation (16.3% vs. 5.5% , OR = 3.11 (1.05–9.23)). This effect was attenuated after adjusting for years of education (OR = 2.79 (0.92–8.4)). No other occupations were significantly linked to RBD (Table 2); inparticular, farming and occupational pesticide use did not differ between groups. Neither rural living nor well water use was associated with RBD. Current or past body mass index were not associated with RBD.
Medical and family history
There was no association found with a diverse list of medical comorbidities, including cardiac and atherosclerotic risk factors, stroke, and head trauma. Family history of neurological symptoms, PD, Alzheimer’s disease, and RBD did not differ between the groups (Table 3). However, when analysis was assessed according to a combined history of PD and dementia (i.e. suggestive of DLB/PD dementia), patients with RBD were more likely to report a positive family history (13.3% vs. 5.5% , OR =3.28 (1.07–10.0)).
RBD before and after PD
In a secondary exploratory analysis, we stratified PD-RBD patients to those who had onset of RBD symptoms before (i.e. those who passed through an idiopathic RBD stage) vs. only after PD. 89/98 patients with RBD provided onset of dream enactment behavior (Supplementary Table 1). As would be expected, RBD duration was longer in those with RBD before PD and PD duration was longer in those with RBD after PD. Although power was limited, patients with RBD onset before PD had non-significant borderline increases in family history of dementia (34.6% vs. 20.8), PD (21.1% vs. 13.5%) and family history of both dementia and PD (20.0% vs. 10.2%).
DISCUSSION
This study has delineated a number of risk factors that differ between PD patients with and without RBD. Among environmental risk factors, we found that PD-RBD patients are more likely to be male, have a combined family history of PD and dementia, have been in a profession involving welding, and have been a smoker.
If PD is a heterogeneous condition in terms of clinical features and neuropathology, it may also be heterogeneous in terms of risk factors. In previous studies, RBD has served as a powerful identifier of a specific subgroup in PD. Now, we provide some evidence that it has some differences in risk factor profile.
One of the most consistent findings in PD epidemiology is that smokers have a lower risk of PD [17]. The mechanism for this effect is controversial. Among the possibilities are a true neuroprotective effect against synucleinopathy, a specific neuroprotective effect on the substantia nigra only, a symptomatic effect (nicotine improves motor PD), reverse causality (prodromal PD reduces smoking, perhaps via changes in reward circuitry), or confounding by an unmeasured variable (e.g. the Parkinson personality). In our recentmulticenter study of RBD, we found that patients with idiopathic RBD (almost all of whom are in prodromal stages of neurodegenerative synucleinopathy) actually had an increased rate of smoking, and that smoking did not protect against development of PD on prospective follow-up [9, 15]. At the time, we suggested that perhaps smoking had differential associations with PD according to disease subtype; we now provide evidence that this is the case. Moreover, although differences in geographical patterns mean that direct comparison cannot be reliably made (i.e. smoking rates are generally lower in America, which is more represented in this study), it is notable that our PD patients without RBD indeed had lower smoking rates than the controls from our idiopathic RBD study [9] (28.6% vs. 52.8% in idiopathic RBD), an effect consistent with other epidemiological studies [17].
We also found that PD-RBD patients were three times as likely to have conducted welding as part of their occupation. This was partially (but not completely) confounded by education. Of interest, idiopathic RBD patients in our previous study also had increased welding likelihood (OR = 1.53 (0.98–2.40)) compared to healthy controls [9]. It is hard to explain this effect. Manganese, a known neurotoxic substance, is a common component in welding fumes. Manganese has been documented to cause damage to the basal ganglia in manganese miners [18], and is controversially linked to parkinsonism in welders in some studies (although the majority find no increased risk) [19–21]. It is unclear why welding should be particularly associated with the RBD subtype. Given that adjustment for education partially attenuated the effect, it is possible that another unrecognized confound (e.g. socioeconomic status independent of education) could explain the effect.
Although there were slight trends towards increased family history of dementia and PD as individual conditions, it was when we specifically analyzed the combination of dementia and PD that differences emerged. The combination of dementia and parkinsonism is the hallmark of both dementia with Lewy bodies (in which RBD is highly prevalent) and the PD subtype that has associated RBD [22–25]. This may suggest a genetic factor that is specific to the RBD subtype. One strong candidate would be glucocerebrosidase A, which has been linked to dementia in PD [26], and also recently to the presence of RBD in PD [27].
Our study has some limitations. First, to ensure diagnostic accuracy, 1/3 of non-RBD patients had to be excluded because of possible dream enactment on questionnaire. Ideally, each patient would have had the questionnaire administered at the time of the polysomnogram, to prevent this exclusion. Despite the relatively large number of patients, power to detect risk factors with low prevalence was still limited. Many risk factors may be subject to recall inaccuracy, particularly family history, which was proxy-reported. As this was the first study to assess risk factors for RBD in PD, we did not adjust for multiple comparisons; therefore, some findings may be simply due to chance, and this study should be considered exploratory in nature [16]. Of note, this study assessed patients for RBD at a single time point. Of course, some non-RBD patients will eventually develop RBD. In general, this form of ‘non-differential misclassification’ would bias towards the null (i.e. tend to wash out differences between groups). The true ‘lifetime’ proportion of RBD in PD is unknown, so it is impossible to determine to what degree this would bias results. There was an interval between PSG and questionnaire in many patients. It is possible that this could confound some of the assessment of current exposures (e.g. current smoking) or recent-onset medical events. Of note, none of these variables was significantly associated with RBD. Note that although PSG protocols were broadly similar, there was no pre-specified REM atonia cutoff required for inclusion; it is possible that some patients classified as RBD in one lab may not have been classified as such in another. Again, this might lead to non-differential misclassification, washing out differences between groups. On the other hand, this study has strengths. Most notably, we had a large sample size of polysomnography-confirmed RBD, ensuring accurate assessment of RBD status. Sample size ensured sufficient power to assess moderate-frequency variables. The study assessed a comprehensive range of variables, including many that have never been assessed in PD-RBD.
In conclusion, PD patients with comorbid RBD differ from those without, with regards to several environmental variables. Recognition that different subtypes of disease may have different underlying risk factor profiles allows more targeted testing of epidemiologic hypotheses. Recognition of distinct PD subtypes can help to develop more specifically tailored personalized treatment strategies.
STUDY FUNDING
This study was supported by the Fonds de Recherche du Québec –Santé and the Canadian Institute of Health Research.
DISCLOSURES
RB Postuma received grants from the Fonds de la Recherche en Sante Quebec, the Canadian Institute of Health Research, the Parkinson Society, the Weston-Garfield Foundation, and the Webster Foundation, as well as funding for consultancy from Biotie, Roche and speaker fees from Novartis Canada and Teva Neurosciences. JF Gagnon is funded by grants from the Fonds de Recherche du Québec –Santé, the W. Garfield Weston Foundation and by the Canadian Institutes of Health Research. He holds a Canada Research Chair on Cognitive Decline in Pathological Aging. Y Dauvilliers has received funds for speaking and board engagements with UCB Pharma, Jazz and Bioprojet. S. Rios Romenets receives grant and contract support from the NIA, Banner Alzheimer’s Foundation, Genentech, Colciensius, and an anonymous foundation to develop the API ADAD Registry and help conduct the API ADAD Trial in Colombia. EK St. Louis is funded by the Mayo Clinic Center for Clinical and Translational Science and the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant Number 1 UL1 RR024150-01. Marie Jacobs, Stuart McCarter, Amélie Pelletier and Mahmoud Cherif have nothing to disclose.