REM sleep behavior disorder and motor dysfunction in Parkinson ’ s disease e A longitudinal study

Objectives: Longitudinal assessment of a Parkinson ’ s disease (PD) cohort, to investigate the evolution or REM sleep behavior symptoms (RBD) over time and to test the relation between RBD at onset and motor dysfunction progression. Methods: An early stage PD cohort ( n ¼ 61) was assessed at two time points, separated by a two years interval. Diagnostic criteria for RBD were: violent behavior during sleep and body movements or vocalization indicative of dream enacting and at least six af ﬁ rmative answers in the REM sleep behavior disorder screening questionnaire. Motor function assessment was performed with the Uni ﬁ ed Parkinson ’ s Disease Scale part II and III (total and partial scores for tremor, bradykinesia, rigidity, gait/postural instability and dysarthria). Results: 25 Patients had RBD at baseline, vs. 35 at follow-up. Three RBD changed to non-RBD at follow-up, while 10 non-RBD patients developed RBD at follow-up (annual incidence of 12.5%). RBD and non-RBD patients did not differ signi ﬁ cantly at baseline or follow-up. The presence of RBD at baseline was signi ﬁ cantly related to an increase in UPDRS total and bradykinesia scores over time. Discussion: RBD symptoms can vary over time and have a tendency to increase during the early stages of disease. The presence of RBD symptoms could be a risk factor for motor function deterioration and particularly for bradykinesia worsening. (cid:1) 2013 The


REM-sleep behavior disorder (RBD) is a REM sleep parasomnia characterized by violent behavior during sleep, related to loss of
Editor's comment: Rapid eye movement sleep behavior disorder (RBD) is characterized by abnormal behavior during the sleep phase with rapid eye movement (REM), due to loss of muscle atonia (paralysis) during REM sleep.RBD has been linked to various neurodegenerative disorders (such as Parkinson's disease (PD)).Reported studies are usually cross sectional in design with a focus on association with specific clinical factors or disease phenotype (1)(2)(3).In this paper, Bugalho et al. conducted a longitudinal study to evaluate the evolution of RBD in an early PD cohort over a two year period.In their study, 25 PD had RBD at baseline, with 3 of them converting to non-RBD at followup, while 10 non-RBD patients developed RBD subsequently, suggesting that RBD symptoms can vary over time.The presence of RBD at baseline was associated with severity of parkinsonian symptoms, and hence may be a risk factor for motor function deterioration.A longer term followup may be useful to determine the exact relationship between RBD and clinical progression of PD.Eng-King Tan, Associate Editor, Department of Neurology, Singapore General Hospital, National Neuroscience Institute, Outram road, Singapore REM sleep muscular atonia and enactment of vivid dreams [1].RBD can occur in isolation, but is most frequent in patients with neurodegenerative disorders, particularly those associated with Lewy body deposition, like Multiple System Atrophy, Lewy-body dementia and Parkinson's disease (PD) [2].In a previous study, we have presented results from a cross-sectional analysis of a cohort of early stage PD patients, in which worse non-tremor motor symptoms were found to be associated with past or present RBD symptoms [3].Several other cross-sectional studies have suggested that RBD could be associated with specific PD phenotypes, comprising a predominance of postural instability and gait difficulties [4e7].However, there is a need for longitudinal studies, in which the significance of RBD as a risk factor can be determined more accurately.
We conducted a longitudinal assessment of the previously mentioned cohort, in order to assess the evolution of RBD over time and to evaluate the predictive value of RBD for motor dysfunction progression in PD.

Methods
Seventy-five early stage PD patients, diagnosed according to validated criteria [8], were consecutively recruited from Hospital Egas Moniz Neurology Department's outpatient clinic.All patients had clinically satisfactory response to dopaminergic treatment, and none presented with cerebellar dysfunction, signs of autonomic dysfunction, or other signs suggestive of multiple system atrophy or other atypical parkinsonism.Early stage PD was defined, at baseline, as disease duration (time in years from appearance of first motor symptoms to first assessment) up to 5 years and Hoehn and Yahr (HY) [9] stage from 1 to 2.5, included.
Patients were assessed twice, with a two year interval (t0 and t1), with the same instruments and by the same observer.

Motor assessment
Patients were assessed with the Unified Parkinson's Disease Rating Scale (UPDRS) part II and III [10], after receiving their usual medication and while on on state.Separated scores were derived for tremor, rigidity, bradykinesia, speech and gait/postural stability symptoms, from items 20 and 21, 22, 23 to 27, 18, and 29 to 30, respectively.Patients were split into three motor groups, according to the classification system proposed by Jankovic and co-workers [11]: i. tremor predominant; ii.intermediate; iii.postural instability and gait difficulty (PIGD) predominant.Recent studies have shown that motor phenotype variation over time occurs usually in the tremor-PIGD direction and that this progression is predictive of worse prognosis [12].For statistical purposes, motor status progression was defined as a change from the tremor dominant form to the intermediate or PIGD phenotype, versus maintenance of tremor subtype or a change from intermediate of PIGD subtype to tremor (patients who had PIGD subtype at baseline and maintained that status were excluded from this analysis).Dopaminergic treatment was calculated as L-Dopa equivalent doses (DED) [13].

RBD assessment
Clinical interview with the patients and bed partners was used to confirm the presence of minimal clinical criteria for RBD, according to the International Classification of Sleep Disorders (ICSD): violent behavior during sleep and body movements or vocalization indicative of dream enacting [14].Assisted by their bed partners, the patients also answered the REM sleep behavior disorder screening questionnaire (RBDSQ), a recently proposed screening tool, which proved to have great sensitivity and reasonable specificity for RBD diagnosis as confirmed by polysomnography [15].RBDSQ contains 13 questions, covering several aspects of RBD symptom spectrum.In the original study a score of at least five positive answers was proposed as the cut-off value for possible RBD.However, we chose to use 6 as the cut-off score, as proposed by a more recent investigation, that validated this scale for use in PD population [16].Patients who fulfilled the ICSD minimal criteria and scored above cut-off on the RBDSQ were classified as possible RBD cases (RBDp).Patients which did not fulfill both criteria were classified has non-RBDp patients.

Statistical analysis
The annual incidence of RBDp was calculated as the number of new RBDp cases divided by the time elapsed (in years), divided by the total number of patients without RBDp at t1.We used the McNemar test to compare the proportion of RBDp and non-RBDp between baseline and follow-up.Two way independent sample t tests or ManneWhitney tests (depending on the distribution of the variables) where used to compare continuous variables at baseline and follow-up, between RBDp and non-RBDp patients.Repeated measures ANOVA was used to test the relation between the presence of RBDp at baseline (between-patients factor) and the variation of motor scores over time (within-patients variables).
Ethics: Patients and controls signed informed consent forms.The ethics committee of the institution approved the investigation protocol.The investigation was performed according to the Declaration of Helsinki.

Results
Of the 75 patients assessed at t0, 61 patients were reassessed at t1.Two patients died, from causes unrelated to PD; 1 refused reassessment; diagnosis changed in 4, who at follow-up were found not to have PD; 7 patients could not be found at t2. Patients who could not be assessed at t1, showed a tendency for older age (75.4 AE 4.55 vs. 71.9AE 7.53, p ¼ 0.099), and older age of onset (72.7 AE 4.75 vs. 69.1 AE 7.76, p ¼ 0.098).There were no significant differences in demographic or motor variables between these patients and those that were available for reassessment.RBDp prevalence was not significantly different: 4 (29%) vs. 28 (46%), p ¼ 0.370.
The number of RBD cases increased from t0 to t1, although the difference did not reach significance (28e35, p ¼ 0.950).Annual incidence of RBD was 12.5%.Variation from non-RBD to RBD was more frequent than in the opposite direction, but most patients maintained the baseline diagnosis (Fig. 1).
Forty-seven patients were on dopaminergic treatment at t0 (mean DED 390 AE 355.8), 18 exclusively on L-Dopa (mean DED 267.2 AE 287.9), 7 exclusively on dopamine agonists (mean DED 170 AE 164.4) and 22 on both (mean DED 674.1 AE 370.7).Four patients were additionally under treatment with MAO inhibitors (2 with selegiline, two with rasagiline), as a co-adjuvant treatment to L-Dopa.At t1, 59 patients were on dopaminergic treatment, 21 exclusively on L-Dopa (mean DED 498.4 AE 213.9), 8 exclusively on dopamine agonists (mean DED 185.0 AE 118.9) and 30 on both (mean DED 751 AE 370.8).To investigate the relation between dopaminergic doses increase (calculated as DED at t1 e DED at t0) and RBD diagnosis change from t0 to t1, we compared total, L-Dopa, and dopamine agonist doses increase between patients that changed from to RBDp to non-RBDp, maintained non-RBDp diagnosis, maintained RBDp diagnosis or changed from non-RBDp to RBDp (ANOVA followed by Bonferroni post-hoc analysis).We found no significant differences.Ten patients were on SSRI at t0 (7 had RBD symptoms) and 1 RBD patient was on clonazepam.There were no patients on rivastigmine.At t1, 15 patients were on SSRI (11 with RBD), 5 were on rivastigmine (3 with RBD) and 3 on clonazepam (all RBD patients).There was no significant relation between the use of these medications and the presence of RBD at t0 or t1.Two patients who were on SSRI at t0 suspended this medication at t1, while three who were not on SSRI at t0, had started this medication at t1. Change in SSRI was not significantly related with change in RBD diagnosis (Chi-square tests).
Fifteen patients presented with PIGD phenotype at t0, and maintained that status at t1, being excluded from motor status progression analysis.Of the other 46 patients, 16 showed motor status progression.Nine of these patients had RBDp at t0, a non significant relation (p ¼ 0.352).
RBDp and non-RBDp groups did not differ significantly regarding motor scores, either at t0 or t1.The presence of RBDp at onset was significantly related to increase in UPDRS total scores and very significantly related to increase of bradykinesia scores (Table 1).Paired samples t test revealed a significant (p ¼ 0.032) increase in bradykinesia in patients who presented RBDp at t0, and a non significant increase in the non-RBD group (p ¼ 0.336).

Discussion
We found a high prevalence of RBDp in our early stage PD cohort.Although most patients retained their baseline RBD diagnosis at follow-up, there was a net increase in RBDp cases over time, with an annual incidence of 12.5%.Lavault and coworkers [17] found a decrease in RBD cases over a two year follow up and a lower annual incidence (9%), although they started with a higher prevalence of RBD at baseline.In Gjerstad's et al. study [18], which had a longer follow-up (8 years, three visits), there was a much larger fluctuation in RBD symptoms, with only 3 out of 89 patients maintaining an RBD diagnosis at all three assessments and total RBD percentages varying from 14.7% at first assessment, to 27.3% on the second visit and 14.6% at the last appointment.Divergences regarding previous studies could be caused by differences in PD populations, as ours was a selected, early stage cohort.A shorter follow-up time and different case ascertainment method (Gjerstad's study was population based) may also have contributed to these differences.Also, each of the three studies based RBD diagnosis in a different questionnaire, which could have resulted in different diagnosis rates.All in all, results from our study and those of Lavault and Gjerstad's, suggest that RBD symptoms suffer fluctuations over time, which are probably less expressive in the early stages of disease, according to our results.This should be taken into account when interpreting RBD prevalence results obtained in questionnaire based, cross-sectional analysis.
Our results suggest that RBD could be associated with worsening of PD motor symptoms, and particularly of bradykinesia.We found only one longitudinal study that has assessed the relation between RBD and progression of motor symptoms [17], which did not find significant associations.This study used a different categorization of motor function: symptoms were aggregated in dopamine and non-dopamine responsive symptoms, and bradykinesia was not studied separately.Several cross-sectional studies have suggested that RBD could be associated with worse motor symptoms.Sixel-Döring and collaborators [6] found that RBD was related to higher HY motor stage and falls.In Lee and collaborators study [4], RBD patients had a lower proportion of their UPDRS scores accounted for by tremor, compared to those without RBD.Kumru [7] also found an association between RBD and non-tremor symptoms, as well as Postuma and collaborators [5].Our own cross-sectional study, showed a higher prevalence of PIGD phenotype in patients with present or past RBD symptoms [3].These findings suggest a relation between RBD and axial, gait and equilibrium related symptoms.This remains a matter of debate, however, as recent studies were unable to find worse gait/ postural dysfunction in PD-RBD patients compared to PD patients without RBD [19].Our data argues against an association between RBD and PIGD phenotype.To our knowledge, our study is the first to find a longitudinal relation between RBD and bradykinesia.This difference is relevant at a pathophysiological level, as bradykinesia shows a strong relation with substantia nigra degeneration and dopaminergic deficits [20].There are studies that point to a relation between the presence of RBD and a decrease in dopaminergic function on the striatum in functional imaging [21] (although other studies have in turn contradicted these results [22]).Our results could suggest that RBD is associated with a faster progression of neuronal loss in substantia nigra.
The present investigation has some limitations.Like in most of the longitudinal studies performed so far, and many of the crosssectional studies, RBD diagnosis was based on clinical criteria and a questionnaire, instead of polysomnography.This method does not take in account asymptomatic cases (i.e.REM sleep without atonia cases).On the other hand, the lower specificity of the RBDSQ could have resulted in an overestimation of RBDp cases.We should also point that the test-retest reliability of RBDSQ was not investigated in the PD validation study.RBD assessment at t0 could have increased bed-partners awareness for mild symptoms, as previously reported by Postuma et al. [23], resulting in an increase of RBD cases at followup.As advantages, we should refer the longitudinal nature of the study, in a field which is lacking this kind of methodology, and the different categorization of motor function, which has allowed us to find a specific relation between RBD and bradykinesia.
In conclusion, our results suggest that RBD symptoms can vary over time, and that the presence of RBD could be a risk factor for motor function deterioration and particularly for bradykinesia worsening.

Table 1
Motor evolution according to RBDp status at onset.Values are mean AE SD.F and p values represent results of repeated measures ANOVA, used to test the relation between RBD status at t0 and score variation from t0 to t1 *p < 0.05; **p < 0.01.HY e Hoehn and Yahr stage; UPDRS e Unified Parkinson Disease Rating Scale total score; DED e L-dopa equivalent doses.