Associations between Sleep Disorders and Impulsive-Compulsive Behaviors in Parkinson’s Disease: A Prospective Cohort Study Open Access

Parkinson’s disease (PD) is the second most common neurodegenerative disorder after Alzheimer’s disease [1, 2], affecting about 1% of the population aged 60 years and over [3]. The pathophysiology of PD involves degeneration of dopaminergic neurons in the substantia nigra pars compacta and depletion of striatal dopamine [4, 5], leading to cardinal motor symptoms (e.g., resting tremor, rigidity, bradykinesia, postural instability, and akinesis) [6, 7]. Besides motor deficits, non-motor symptoms such as impulsive-compulsive behaviors (ICBs), sleep disorders, cognitive impairment, and depression [8], are also increasingly documented in PD, some of which may even appear several years before the onset of motor symptoms.

ICBs are types of psychiatric and neurological symptoms among PD patients, referring to a spectrum of repetitive, excessive, and compulsive behaviors driven by strong desires, and patients often have difficulty in self-control [9]. ICB is characterized by four main impulse control disorders (ICDs) including pathological gambling, hypersexuality, compulsive shopping, and compulsive eating, as well as other compulsive behaviors, such as punding, hobbyism, walkabout, and dopamine dysregulation syndrome [9]. Previous studies reported that up to 20% of PD patients subsequently developed ICD and other compulsive behaviors during the disease course, resulting in serious negative effects on the patient’s quality of life [10, 11].

Sleep disorders are common non-motor symptoms in PD patients [12], characterized by nocturnal and diurnal manifestations, such as excessive daytime sleepiness (EDS), rapid eye movement sleep behavior disorder (RBD), and insomnia [12]. EDS is featured by periods of irrepressible need for sleep or unintended drowsiness or sleep as well as the inability to stay awake and alert during the major wake episodes of the day [13], with a prevalence of 21–76% [14]. RBD, first described in 1986 [15], is characterized by loss of muscle atonia during rapid eye movement sleep and symptoms of dream enactment, typically diagnosed clinically with supportive evidence from a polysomnogram. Probable RBD (pRBD) is commonly observed among PD patients, with a prevalence of 16–47%, and could occur before or after the onset of motor symptoms [16]. Associations between ICB and sleep disorders, such as poor efficiency sleep, EDS, and pRBD, have been implicated by numerous studies of PD [17‒19]. A similar phenomenon was observed in a case-report study, which found that compulsive eating was related to EDS [20]. In addition, impulse-control disturbances may interfere with the transcription of circadian clock genes, altering their activity of circadian rhythm, which in turn contributes to sleep disorders and susceptibility to addictions [21, 22]. Although previous studies have explored the associations between ICB and sleep disorders [23, 24], most of them employed case-control or case-report designs, with scarce longitudinal examinations. This study aimed to explore the longitudinal associations between sleep disorders (EDS and pRBD) and ICB in PD patients using 5-year longitudinal cohort data from the Parkinson’s Progression Markers Initiative (PPMI) program.

The PPMI is an ongoing international, multicenter, observational, and longitudinal cohort study designed to identify biomarkers of PD progression [25, 26]. To this end, patients with early-stage untreated (de novo) PD were recruited at baseline and followed every 3 months in the first year and every 6 months in the following years. For the present analysis, data from baseline and follow-up visits during the first 5 years of follow-up of 807 PD patients were downloaded on June 12, 2023.

For patients with de novo PD, eligibility criteria included a clinical diagnosis of idiopathic PD [27], with resting tremor and/or bradykinesia and rigidity or an asymmetric resting tremor, and Hoehn and Yahr (H&Y) stage ≤2 [28]. The Institutional Review Board approved each participating site, and all study participants have provided written informed consent. More details about the PPMI study are described on ppm-info.org. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline have followed this current study [29].

We assessed EDS and pRBD as two separate time-varying exposure variables. EDS was defined from the self-reported Epworth Sleepiness Scale (ESS), which is an eight-item, self-administered tool that is widely used to evaluate the likelihood of dozing off or falling asleep in different scenarios [30, 31]. A participant with self-reported ESS ≥10 out of a total score of 24 points was classified as having EDS, and such EDS definition has been validated [30]. pRBD was defined from the RBD Screening Questionnaire (RBDSQ), a 10-item self-rating questionnaire about the clinical features of pRBD [32]. A participant with a self-reported RBDSQ ≥5 out of a total score of 13 points was considered to have pRBD. The sensitivity and specificity of such pRBD definitions have been shown to be as high as 0.84 and 0.96, respectively [32, 33].

The short version of the Questionnaire for Impulsive-Compulsive Disorders rating scale (QUIP-S) is a self-completed screening instrument rather than a diagnostic tool, specifically developed and validated to estimate the existence of current ICBs in individuals with PD over a 5-year follow-up period. Following a previous study, the presence of ICB was defined by a total score of >0 [34].

Demographic and clinical data were collected for all participants. Covariates were adjusted in either a time-varying or time-invariant fashion. For the time-invariant covariates, gender, family history (no or yes), and disease duration (months) were assessed at baseline. For time-varying covariates, they were defined using data from baseline and all 5-year follow-up visits. Demographic variables included age at visit (years), clinical characteristics included Movement Disorders Society-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) part III Total Score, levodopa equivalent daily dose (LEDD), Geriatric Depression Scale (GDS-15), and State-Trait Anxiety Inventory-Form Y (STAI-Y).

The severity of PD was assessed by the MDS-UPDRS part III total score [35]. Total LEDD was calculated for every participant with PD [36]. The 15-item Geriatric Depression Scale (GDS-15) was used as a screening tool for depression [37] and has been commonly studied and validated in young and old PD patients [38]. The STAI-Y was considered to be a valid instrument for assessing anxiety symptoms in individuals with early-stage PD [39].

Data were described as median (interquartile range, IQR) for continuous variables and as frequency or percentage for categorical variables. Comparison between two groups was performed via the Kolmogorov-Smirnov test or Mann-Whitney U test for continuous variables and by χ² test for categorical variables. To evaluate the relationship between repeatedly measured EDS and pRBD, an unadjusted Spearman’s rank correlation coefficient was calculated.

Generalized linear mixed-effects models were used to assess longitudinal associations between these traits over the 5-year follow-up period, with the intercept and slope of age at visit fitted as random effects at the participant level. For longitudinal analyses, three separate models were fitted for the entire cohort and for PD patients: (1) adjusted for age at visit and gender, (2) further adjusted for family history, GDS score, and STAI-Y score, (3) additionally adjusted for MDS-UPDRS part III score, LEDD, and disease duration. These covariates were known to be associated with both sleep disorders and ICB. The results were reported as odds ratio (OR) with a corresponding 95% confidence interval (CI).

To test the robustness and potential variations in different subgroups, we conducted effect modification analysis by population characteristics. Subgroup analyses were stratified by age at baseline (<65 or ≥65 years), gender (men or women), family history (no or years). Potential modifying effects were examined by testing the corresponding multiplicative interaction terms.

Two sensitivity analyses were performed to test the robustness of the findings. First, to avoided potential effect other subgroups on outcome, we only chose subgroup with sporadic PD participants, while excluding hyposmia, leucine-rich repeat kinase 2 (LRRK), glucocerebrosidase, alpha-synuclein, or other rare mutation. Second, we avoided the possibility of reverse causation by further excluded PD patients more than 2 years disease duration at baseline. The statistical analyses were performed on R software (version 4.1.1) with the lme4 package, and a two-tailed p < 0.05 was considered statistically significant.

Among the 938 participants enrolled in the PPMI program, 807 participants were included in the present analysis (Fig. 1). The baseline characteristics of participants according to ICB status are summarized in Table 1, comprising 496 (61.5% men) patients with PD, with a median (IQR) age of 63.1 (55.6–69.3) years. Compared to non-ICB participants, ICB participants were likely to be younger, had longer disease duration, had lower MDS-UPDRS part III score, and tended to have higher LEDD, GDS score, ESS score, and RBDSQ score (all p < 0.05), whilst no substantial differences were observed in gender and family history (p > 0.05).

Additionally, online supplementary Table S1 (for all online suppl. material, see https://doi.org/10.1159/000536555) depicted the baseline characteristics of participants stratifying by gender. Men exhibited lower disease duration, LEDD, GDS score, and STAI-Y score and tended to have higher MDS-UPDRS part III score and RBDSQ score (all p < 0.05), whereas the other differences were not statistically significant (p > 0.05).

LEDD for PD patients with or without ICB over 5-year follow-up was further shown in online supplementary Table S4. The mean (SD: standard deviation) baseline LEDD was of 118.96 ± 273.86 in the non-ICB group, with a median (IQR) of 0.00 (0.00–0.00). Furthermore, LEDD was administered at each time point in the non-ICB group. The proportion of those who were administered levodopa at baseline was 149 (24.6%) in the non-ICB group and 87 (43.3%) in the ICB group.

Longitudinal changes in ESS and RBDSQ Score with age stratified by QUIP status for each participant during 5-year follow-up were depicted in online supplementary Figure S1. The generalized linear mixed-effects models demonstrated that all models observed significant associations in Table 2 (p < 0.05). EDS (OR = 1.09, 95% CI: 1.05, 1.12) and pRBD (OR = 1.07, 95% CI: 1.03, 1.12) among PD patients were essentially associated with higher odds of developing ICB over time, after adjusting for age, gender, family history, GDS score, STAI-Y score, MDS-UPDRS part III score, LEDD, and disease duration.

Consistent results were observed in association of EDS and pRBD with ICB when analyses were stratified by age at baseline, gender, and PD family history (Table 3; online suppl. Fig. S2). We did not find statistically significant interactions after accounting for multiple testing (p < 0.05).

The results were robust in sensitivity analysis. The positive association between sleep disorders and ICB was not materially changed when participants without sporadic were excluded (online suppl. Table S2). After excluding PD patients with more than 2 years of disease duration at baseline, the results remained substantially unchanged (online suppl. Table S3).

Based on 5-year follow-up data from the PPMI, this longitudinal study demonstrated that EDS and pRBD were associated with a higher risk of ICB in early PD. To our knowledge, this is the first longitudinal study to specifically investigate the association between EDS and ICB in PD patients over time.

A previous study indicated an increased frequency of EDS in untreated PD patients over a 5-year follow-up, suggesting that the occurrence of EDS might be mainly explained by disease progression, which did, however, not analyze the association of EDS with ICB [40]. In a previous study based on PPMI, subjective impulsivity appeared to be higher among individuals with EDS at 1-year and 2-year follow-up visits, compared to baseline, although no difference was observed at year 3 [41]. This study did, however, not focus on longitudinal associations [41]. Further, escalated risk has been reported in the association between daytime sleepiness and impulsivity in PD [18]. It is unclear, however, whether sleep disturbances enhance the vulnerability of impulsivity or are caused by addictive behaviors [18].

Several studies have examined the associations between probable RBD and ICB in PD patients. While some reported that probable RBD was associated with an increased risk of ICB [24, 42‒44], others did not [45‒49]. Based on a 5-year longitudinal study among 423 PD patients, baseline RBD was suggested to be an independent predictor of the development of ICDs over time [42]. Of note, in previous cross-sectional studies, there was a positive association between RBD and the risk of ICB among patients with PD [24, 43, 44]. However, in a previous longitudinal study that recruited 921 patients with PD, RBD did not impose an elevated risk of ICDs [45]. Similar associations were reported in another longitudinal study including 401 newly diagnosed PD patients, showing that probable RBD was not associated with increased ICB development in PD [46]. A nonsignificant association of RBD with increased ICD risk was reported in other cross-sectional studies [47, 48] and a case-control study [49]. A meta-analysis of 10 observational studies, including 2,781 PD patients, suggested that pRBD was a risk factor for ICB in PD, with a more than two-fold (OR = 2.12, p < 0.01) increase in the pooled estimate for ICB [50]. However, in our study, the association was smaller than that reported in the meta-analysis. This difference may stem from the fact that most studies in the meta-analysis utilized a cross-sectional design, which could overestimate the associations. In contrast, our study was the first to employ a longitudinal design with repeated measurements over 5 years in a relatively large sample, providing new evidence for the associations between pRBD and ICB.

The mechanisms underlying the relationship between sleep disorders and ICB in PD patients are, however, not well understood. The mesocorticolimbic system appears to play a critical role in the reward system [51]. It is likely that sleep deprivation gives rise to a dysfunction of the prefrontal cortex and its connection to the limbic system, leading subsequently to a reduction in inhibitory control [52]. Sleep disorders may disrupt the dopaminergic system and yield an imbalance in dopamine levels, or may affect frontal lobe function, further contributing to increased risk of ICB [53, 54]. Increased hypersexuality in idiopathic RBD was found to be correlated with altered functional connectivity between the limbic striatum and posterior cortical regions [55]. A probable interpretation is that the change might finally predispose individuals to the occurrence of ICDs when receiving dopaminergic medications for patients with PD [55].

Our study has several strengths and limitations. First, over a 5-year follow-up period, we were able to estimate long-term dynamic changes in ICB based on multiple repeated measurements. Second, our results are relatively robust, with similar findings observed in subgroup and sensitivity analyses. Third, to our knowledge, this is the largest longitudinal cohort of patients with PD to examine the association between EDS and ICB so far. However, several limitations should also be acknowledged. First, the assessment of ICB and sleep disorders was based on self-reported screening instruments such as QUIP-S, ESS, and RBDSQ, rather than diagnostic tools, which may be susceptible to information bias. Second, some participants were lost to follow-up, which may affect the reliability of results to a certain extent. Third, given the nature of observational studies, we could not eliminate the possibility of residual confounding and establish a firm causal association. Accordingly, to enhance the validity of the findings, future research is required by expanding the sample size and the follow-up.

In summary, our findings demonstrated that EDS and pRBD were longitudinally associated with an increased risk of ICB among patients with PD. The findings emphasize the significance of evaluating and addressing sleep disorders in PD patients as a potential approach to managing ICB.

PPMI was funded by the Michael J. Fox Foundation for Parkinson’s Research and multiple funding partners, whose names can be found at www.ppmi-info.org/fundingpartners.

PPMI was conducted in accordance with the Declaration of Helsinki for all the participating sites. Each participating PPMI site received local ethical approval from an ethical committee before study initiation. Because data from PPMI was made publicly available, our present study only involved data analysis rather than data collection and has therefore been granted an exemption from ethics approval according to the Ethical Review Board of School of Public Health (Shenzhen) at Sun Yat-Sen University. Consent to participate statement: Written informed consent for research was obtained from all participants.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The study was supported by Sun Yat-Sen University. Yiqiang Zhan was supported by the Pearl River Scholar Program of Guangdong Province (Health Science Section, No: 0920220206). Karin Wirdefeldt was supported by Region Stockholm (clinical research appointment).

Y.Z., X.K., I.K.K., and X.X. designed and conceptualized the study; X.K., XX, S.H., F.F., and Y.Z. conducted the study and critiqued the manuscript; I.K., YW, Q.J., Q.L., and F.F. extracted the data and performed statistical analysis; MY, X.K., F.F., K.W., S.S., and I.K.K. wrote the first draft of the manuscript. All authors have reviewed and approved the final version of the manuscript.

Data supporting the findings of this article were obtained from the Parkinson’s Progression Markers Initiative (PPMI) database (www.ppmiinfo.org/data). For up-to-date information on the study, visit www.ppmiinfo.org.