Cognitive Performances and Mild Cognitive Impairment in Idiopathic Rapid Eye Movement Sleep Behavior Disorder: Results of a Longitudinal Follow-Up Study

Abstract

Study Objectives:

To investigate the capacity of neuropsychological deficits in idiopathic rapid eye movement sleep behavior disorder (iRBD) to predict the development of dementia and/or parkinsonism.

Design:

Prospective longitudinal follow-up study.

Setting:

Tertiary sleep center.

Patients:

Twenty patients with initial iRBD (19 males, mean age 66.1 ± 7.1) underwent a clinical and neuropsychological follow-up within a mean of 43 ± 19 months. Neuropsychological performances at baseline were compared with those of healthy controls matched for sex, age, and education.

Interventions:

Discontinuation of clonazepam at least 7 days before the follow-up evaluation.

Results:

At follow-up, the Wilcoxon test showed a significant worsening of scores on Raven Colored Matrices 47 (P = 0.01), Attentive matrices (P = 0.002), phonemic (P = 0.04) and sematic (P = 0.04) fluency. Thirteen patients (65%) showed cognitive deterioration involving multiple domains. Of these, four patients (20%) maintained a stable cognitive dysfunction and nine (45%) showed a progression of cognitive dysfunction: six (30%) in constructional abilities (P = 0.03), four (20%) in short-term memory (P = NS), three (15%) in executive functions and non-verbal logic (P = NS), one (5%) in verbal fluency (P = NS), and one (5%) in long-term memory (P = NS) (McNemar test). Seven patients (35%) retained a normal cognitive profile. Mild cognitive impairment (MCI) was diagnosed at baseline in seven patients (35%). At follow-up, three of these patients showed overt dementia that was accompanied by parkinsonism in all cases; one had worsened from non-amnesic single-domain to nonamnesic multiple-domain MCI, two were stable, and one patient no longer met the criteria for MCI. Four patients (20%) without MCI at baseline had MCI at follow-up. Patients who developed MCI/dementia had an older age at disease onset (65.8 ± 5.4 versus 56.8 ± 9.3; P = 0.01) compared with those who did not.

Conclusions:

Our findings corroborate evidence that visuospatial abilities constitute the area most affected in idiopathic rapid eye movement sleep behavior disorder (learning as a stable deficit and copying as an evolving deficit). Cognitive deterioration, involving mainly nonverbal logic, attention, and executive functions, can be observed in rapid eye movement sleep behavior disorder follow-up, suggesting an underlying evolving degenerative process. Our data confirm that mild cognitive impairment is frequent in idiopathic rapid eye movement sleep behavior disorder. The presence of mild cognitive impairment predicts the eventual risk of developing dementia, which seemed to be associated with parkinsonism.

Citation:

Terzaghi M; Zucchella C; Rustioni V; Sinforiani E; Manni R. Cognitive performances and mild cognitive impairment in idiopathic rapid eye movement sleep behavior disorder: results of a longitudinal follow-up study. SLEEP 2013;36(10):1527-1532.

INTRODUCTION

A considerable proportion of patients in whom idiopathic rapid eye movement sleep behavior disorder (iRBD) was initially diagnosed reportedly go on to develop a neurodegenerative disorder. Since the formal identification of rapid eye movement sleep behavior disorder (RBD) in 1986,¹ several lines of evidence have accumulated in the literature indicating a strong association with synucleinopathies^2–7: the risk of developing a synucleinopathy increases over time following an initial diagnosis of otherwise idiopathic RBD.^8–12 RBD may precede, even by years, the onset of a synucleinopathy.^9,13,14 The risk of developing a synucleinopathy was estimated to be more than 50% at 12 y after the initial diagnosis,¹³ whereas the risk of developing mild cognitive impairment (MCI) or parkinsonism has been found to be increased 2.2-fold in patients with RBD compared with patients without RBD¹⁵ in the general population.

The hypothesis of a strong link between synucleinopathies and RBD is further supported by the finding that some patients with iRBD show subtle clinical (including cognitive) as well as instrumental alterations similar to those found in synucleinopathies.¹⁶ Available literature data from cross-sectional and longitudinal studies show that RBD is associated with cognitive impairment in Parkinson's disease (PD),^17,18 and represents a predictor of progressive cognitive dysfunction, eventually evolving into a clinical picture of dementia.^12,19

Neuropsychological evaluations of patients with iRBD showed different domain-specific cognitive dysfunctions in patients with dementia, highlighting a cognitive profile closely resembling that observed in the α-synucleinopathies^18,20–22; this profile may, therefore, be a marker of an ongoing neurodegenerative process.¹⁶

However, it remains to be established whether these deficits are predictive of the development of synucleinopathies. Furthermore, the true neurobiological significance of cognitive deficits in iRBD is still unknown due to the absence of prolonged long-term follow-up data.

Recently, a prospective 2-y follow-up study in iRBD reported a significant worsening of visuospatial learning over time; neuropsychological dysfunction did not predict the development of a neurodegenerative disorder.²³

To investigate the prognostic potential of neuropsychological deficits with regard to the development of dementia and/or parkinsonism, we report the results of a prospective, longitudinal follow-up study of patients with iRBD.

METHODS

Twenty patients in whom iRBD was diagnosed according to standard criteria²⁴ (mean age 66.1 ± 7.1 y; 19 males [95%], age at RBD onset 60 ± 9.1 y; mean RBD duration 7 ± 8.5 years; Mini-Mental State Examination [MMSE] 26.8 ± 1.7 [range 24.3-29.7]; education 7.2 ± 3 y) were enrolled in the study and underwent a clinical and neuropsychological follow-up within a mean of 43 ± 18.8 months.

Patients with a history of head trauma, cerebrovascular disease, neurological conditions, other sleep disorders (narcolepsy, sleep disordered breathing), or current/past treatment with a drug known to influence rapid eye movement (REM) sleep were not included.

None of the patients were on clonazepam at baseline, and clonazepam treatment was discontinued at least 7 days before the follow-up evaluation.

For comparison of neuropsychological performances at baseline, a control group of 20 healthy subjects matched for age, sex, and education was selected (MMSE score 27.3 ± 1.2, P = NS).

None of the patients presented with depressed mood, as assessed using the Beck Depression Inventory (cutoff score = 9).²⁵

No patient recorded a corrected MMSE score of less than 24 (the normative value for the Italian population).²⁶

No functional impairment was detected at baseline (Clinical Dementia Rating score 0 in all the patients included).²⁷

Neuropsychological Evaluation

Each patient underwent a neuropsychological evaluation that included²⁸:

• MMSE, to evaluate temporospatial orientation and to obtain a general index of cognitive functioning²⁶;

• Digit span forward, Word span, and Corsi tests, to evaluate short-term, verbal, and spatial memory²⁹;

• Rey 15-word test, both immediate and delayed recall, to investigate long-term verbal memory for nonstructured material³⁰;

• Logical memory, which evaluates long-term verbal memory for structured material²⁹;

• Wisconsin Card Sorting Test, which evaluates executive functions³¹;

• Attentive matrices, to investigate selective attention through visual search²⁹;

• Verbal fluency, both phonemic and semantic^29,30;

• Raven Colored Matrices 47, which evaluates nonverbal logic²⁹;

• Rey-Osterrieth complex figure (copy), which evaluates copying skills³²;

• Rey-Osterrieth complex figure (delayed recall), which evaluates visuoconstructional learning skills.³²

Statistical Analysis

The statistical analysis was performed using the Statistical Package for Social Sciences (SPSS, Baltimore, MD).

As in a previous study,²⁸ we grouped all the tests into new variables (“macrodomains”) on the basis of similarity of evaluated functions. The following macrodomains were identified: visuoconstructional learning skills, tested using the Rey-Osterrieth complex figure (delayed recall); short-term verbal and spatial memory, tested using the Digit span forward, Word span, and Corsi tests; long-term verbal memory, tested using the Logical memory and Rey 15-word immediate and delayed recall tests; executive functions and nonverbal logic, tested using the Wisconsin Card Sorting Test and Raven Colored Matrices 47; verbal fluency, based on evaluations of both phonemic and semantic fluency; visual search abilities, tested using the Attentive matrices; visuoconstructional abilities, tested using the Rey-Osterrieth complex figure (copy).

For each of these new variables, the patients were coded as nonpathological (= 0) or pathological (= 1) if they gave a pathological equivalent score³³ (equal to or less than 1) on all of the instruments used to test the macrodomain.

Because none of the study variables was normally distributed (the Kolmogorov-Smirnov test was used to test normality of distribution), comparison of continuous variables was performed using the Mann-Whitney U-test. Dichotomous variables were cross-tabulated in 2 × 2 tables and Fisher exact test (χ²) was applied.

Changes in cognitive performances at follow-up were assessed using the Wilcoxon test for paired samples or the McNemar test according to the nature (continuous or dichotomous) of the variable studied. A linear regression model including age, disease duration, and follow-up duration as covariates was run to test the relationship between clinical and cognitive variables.

The accepted level of significance was set at P ≤ 0.05.

Mild Cognitive Impairment

Data on subjective cognitive complaints, reported by the patient or by a relative/caregiver, were systematically collected at baseline and at follow-up. MCI was defined as the simultaneous presence of the following conditions: subjective cognitive complaint reported by the patient or by a relative/ caregiver; objective evidence of cognitive decline, defined as pathological equivalent score on at least two of the instruments used to test a given cognitive domain; preserved activities of daily living. The cognitive domains considered were¹⁸: learning and memory (Rey-Osterrieth complex figure – delayed recall; Logical memory; Rey 15-word immediate and delayed recall), executive functions and attention (Wisconsin Card Sorting Test; Raven Colored Matrices 47; phonemic and semantic fluency; Attentive matrices; Digit span forward; Word span; Corsi tests), and visuoconstructional abilities (Rey-Osterrieth complex figure – copy; in this case only one instrument was considered).

RESULTS

Neuropsychological Testing

Baseline

At baseline (Table 1) the RBD group, compared with the control group, recorded lower scores (Mann-Whitney U test) on the Rey-Osterrieth complex figure (delayed recall) (P = 0.007), Digit span forward (P = 0.006), Word span (P = 0.04), Raven Colored Matrices 47 (P = 0.02), Logical memory (P = 0.02), Rey 15-word test delayed recall (P = 0.03), Wisconsin Card Sorting Test (P = 0.004), and Attentive matrices (P = 0.03).

Table 1.

Results of neuropsychological tests in the patient series at baseline compared with paired controls, mean values ± standard deviation

Considering the pathological equivalent scores calculated from the raw scores (Table 2), impaired performances were found (Fisher exact test) in visuoconstructional learning skills (11 of 20 patients [55%], P = 0.006) and visual search abilities (5 of 20 patients [25%], P = 0.04).

Table 2.

Number of patients with pathological equivalent scores in the various cognitive domains tested at baseline compared with paired controls

Follow-up

At follow-up, the Wilcoxon test showed a significant worsening of scores on the Raven Colored Matrices 47 (nonverbal logic, P = 0.03) and Attentive matrices (selective attention, P = 0.03, Table 3).

Table 3.

Results of neuropsychological tests in the patient series at baseline compared with follow-up, mean values ± standard deviation

On the basis of the pathological equivalent scores recorded at follow-up³³ (Table 4), nine patients (45%) showed further cognitive deterioration: the decline in cognitive performances involved visuoconstructional abilities in six cases (30%; P = 0.03), short-term memory in four (20%; P = NS), executive functions and nonverbal logic in three (15%; P = NS), and long-term memory in one (5%; P = NS) (McNemar test).

Table 4.

Number of patients with pathological equivalent scores in the various cognitive domains tested at baseline compared with follow-up

Four patients (20%) maintained a stable cognitive dysfunction (two in visuoconstructional learning, one in visuoconstructional learning and visual search, one in multiple domains).

Seven patients (35%) maintained a cognitive profile within normal ranges.

A linear regression model including age, disease duration, and follow-up duration as covariates, showed that a worse performance on Word span (P = 0.036), phonemic fluency (P = 0.02), semantic fluency (P = 0.04), and the Wisconsin Card Sorting Test (P = 0.038) was associated with a longer follow-up.

Mild Cognitive Impairment

At baseline MCI was diagnosed in seven patients with RBD (35%, Table 5).

Table 5.

Cognitive profile at baseline and follow-up and evolution toward dementia

At follow-up, three of these patients (15%, two with nonamnesic single-domain and one with amnesic multiple-domain MCI) presented with overt dementia, all also exhibiting parkinsonism; one patient had worsened from nonamnesic single-domain to nonamnesic multiple-domain MCI; two patients were stable (one with amnesic multiple domain MCI, one with nonamnesic multiple-domain MCI), and one patient no longer fulfilled the criteria for MCI.

Four patients (20%) without MCI at baseline had newly diagnosed MCI at follow-up (two amnesic multiple-domain, two nonamnesic single-domain–executive).

Patients who developed MCI/dementia had an older age at disease onset (65.8 ± 5.4 versus 56.8 ± 9.3; P = 0.01) compared with those who did not.

Parkinsonism

A total of five patients developed parkinsonism. Of them, two had PD and one multiple-system atrophy (MSA-P). Two patients had extrapyramidal signs not fulfilling diagnostic criteria for PD.

One patient with PD, one with MSA-P, and one patient with extrapyramidal signs not fulfilling diagnostic criteria for PD also had overt dementia.

One patient with extrapyramidal signs not fulfilling diagnostic criteria for PD developed MCI (amnesic multiple-domain–all domains); one patient with PD showed complete absence of deficits at baseline and follow-up.

Pathological scores at baseline as well as at follow-up were recorded in visuoconstructional learning skills in four patients, in visual search abilities in two, and in verbal fluency in one. At baseline versus follow up, pathological scores were recorded by, respectively, one and four patients in visuoconstructional abilities, one and three patients in executive functions and nonverbal logic, zero and two patients in short-term memory, and zero and one patient in long-term memory (mean values ± standard deviations at each test performed at baseline/follow-up: Rey-Osterrieth complex figure delayed recall 10.7 ± 2.8/9.5 ± 3.1; Rey-Osterrieth complex figure (copy) 32.0 ± 2.7/18.6 ± 12.0; Word span 4.0 ± 0.6/3.3 ± 0.9; Digit span forward 4.0 ± 0.8/4.5 ± 1.6; Logical memory 12.2 ± 1.8/6.0 ± 3.8; Rey 15-word test immediate recall 34.0 ± 6.3/28.6 ± 11.6; Rey 15-word test delayed recall 7.6 ± 2.3/6.1 ± 3.8; Raven Colored Matrices 47 22.9 ± 4.4/20.5 ± 5.5; Wisconsin Card Sorting Test 87.0 ± 28.7/93.5 ± 19.6; Corsi test 4.3 ± 0.4/3.9 ± 0.6; phonemic fluency 26.9 ± 10.6/25.7 ± 13.2; semantic fluency 12.7 ± 3.9/14.3 ± 4.5; Attentive matrices 38.8 ± 10.7/40.3 ± 7.2).

DISCUSSION

Both subclinical cognitive deficits and so-called MCI have been documented in iRBD.^8,18,34

The prevalence and type of subclinical cognitive deficits occurring in iRBD, even though these are frequent, are not well established on account of differences across the various studies.^{18,20–23,28} Performances in visuospatial learning, short-and long-term verbal memory, attention and executive functions and, more recently, decision making^35,36 are reportedly the ones most frequently impaired in iRBD. The significance of poor visuospatial/visuoperceptual abilities in iRBD is debated, with some authors suggesting that these deficits are related to the degree of cognitive decline.^{9,18,20,23,34,37} Impaired visual-perceptual-organizational skills, constructional praxis, and verbal fluency have also been described in RBD plus dementia.³⁸

It is unclear whether cognitive dysfunction in iRBD is an associated feature or an evolving condition possibly predicting subsequent development of a synucleinopathy.

Data on the longitudinal evolution of cognitive performances in RBD could help to determine the risk of developing dementia, but unfortunately no such data are available in the literature.

To date, most of the available data come from cross-sectional studies. Recently, a prospective 2-y follow-up study in iRBD reported a significant worsening of visuospatial learning over time; neuropsychological dysfunction did not predict the development of a neurodegenerative disorder.²³

The current study confirms the presence of impaired performance on multiple cognitive tests, with visuoconstructional learning skills and visual search abilities emerging as the cognitive domains most affected in iRBD.

Our follow-up findings corroborate evidence that visuo-spatial abilities are the domain most affected in iRBD^9,21–23,28 (learning as a stable deficit and copying as an evolving deficit), and that dysfunction in this area may constitute a cognitive marker of the disease. Further cognitive deterioration, involving mainly nonverbal logic, attention, and executive functions, can be observed in iRBD follow-up with no evidence of motor abnormalities, thus suggesting the presence of an underlying evolving degenerative process.

MCI is an intermediate stage between the cognitive decline that is expected in normal aging and the more serious decline of dementia. The frequency of MCI has been shown to be significantly higher in patients with iRBD than in age- and sex-matched healthy controls (50% versus 8%).¹⁸ The type of MCI most frequently encountered in iRBD is nonamnesic single-domain MCI, with attention and executive functions the domains predominantly affected. A recent 4-y follow-up study in patients with PD showed that RBD is associated with an increased risk of dementia. All the patients who developed dementia had MCI at the baseline evaluation.¹²

The data from our sample confirm that MCI is frequent in iRBD with nonamnesic single-domain MCI (impaired attention and executive functions) and amnesic multiple-domain MCI (all domains affected) the subtypes most frequently encountered.

Follow-up data in our sample indicate that patients with iRBD and MCI are at risk of developing dementia associated with parkinsonism, the conversion rate at 3.5 y being 43% (three of seven patients). Our study provides evidence that MCI can progress over time, a suggestion supported by the fact that new cases were diagnosed at the 3.5-y follow-up, with a 20% rate of incidental cases.

Even though our findings indicate a close link between parkinsonism and cognitive deterioration (four of five patients with parkinsonism had overt dementia or multiple-domain MCI), one patient who did not show neuropsychological deficits either at baseline or at follow-up went on to develop PD.

A nonnegligible proportion of patients with iRBD continue to show normal cognitive performances at follow-up. This finding, in accordance with the observation that signs of neuro-degeneration are not universal in RBD,¹⁰ suggests that not all patients with iRBD are destined to develop cognitive dysfunction, even when parkinsonism is detected. A younger age at disease onset characterized this group of patients versus the cognitively impaired patients.

Our data, indicating the presence of neuropsychological alterations in the form of MCI or more subtle cognitive deficits (“pre-MCI”) at baseline in otherwise iRBD patients, confirm that cognitive dysfunction can be found in patients with RBD without parkinsonism. On the basis of the assumption that cognitive deficits reflect cortical damage, this could indicate an early involvement of corticolimbic areas without significant extension to nigrostriatal dopaminergic neurons. This is compatible with the concept of a descending progression of degenerative changes from the neocortex/limbic system to the nigrostriatal system^34,39 and also with the hypothesis of a patchy and discontinuous vulnerability of different populations of neurons,⁴⁰ wherein the system responsible for cognitive functioning and the areas implied in RBD genesis⁴¹ reach a critical threshold of degenerative changes prior to the nigrostriatal neuronal networks, as in the Lewy body pathology of the dementia with Lewy bodies phenotype.^34,42

However, it cannot be excluded that subtle cognitive deficits might be due not to cortical pathology, but rather, in accordance with the Braak staging scheme⁴³ denoting a bottom-up progression of neurodegeneration from the brainstem to the cortex, are related to progressive alterations in subcortical circuitry paralleled by cell loss in the substantia nigra that is not sufficient to produce extrapyramidal signs. Were this the case, cognitive dysfunctions could be present early in extrapyramidal disorders even in the absence of other signs/symptoms.

In summary, the results of our study show that cognitive deterioration can be observed in iRBD follow-up, suggesting an underlying evolving degenerative process, and that the presence of MCI predicts the eventual risk of developing dementia, which appears to be closely linked to parkinsonism.

Our data suggest that patients showing evolution toward dementia/parkinsonism are characterized by an older age at disease onset, but this preliminary finding needs to be tested in larger case series (the small number of patients developing dementia/parkinsonism, due to the small size of the sample studied, constitutes a limitation of the current study). More prolonged long-term follow-up data are needed to confirm our preliminary finding about the neurobiological significance of cognitive dysfunction in iRBD. Although our data failed to establish a correlation between follow-up duration and performance on the various tests, we suggest that patients with iRBD should be monitored regularly.

The fact that the evolution of cognitive performances in the patients with RBD was not compared to that of the baseline control group constitutes a major limitation of our study: data on the evolution of cognitive functions in the controls over the years were not collected. This omission could mean that the extent of the cognitive decline in patients with RBD was over-accentuated. Furthermore, our findings refer to a small patient group. Finally, the battery of tests adopted did not allow us to replicate the grouping of neuropsychological tests used in order to define MCI in previous authoritative literature.¹⁸ In particular, visuoconstructional abilities were assessed using only the Rey-Osterrieth complex figure (copy). Differences in testing procedures for MCI diagnosis could limit the comparability of our data to those of previous reports. As new criteria have been proposed and validated for definition of MCI in PD,^44,45 these should be taken into consideration in further studies seeking to define MCI in RBD.

DISCLOSURE STATEMENT

This was not an industry supported study. The authors have indicated no financial conflicts of interest.