파킨슨 병 전구 증상 - 잠꼬대 90%는 퇴행성 뇌질환 파킨슨병, 루이체 치매로 진행

[문형철]

https://www.youtube.com/watch?v=6PCp8ZRFDpo

1. nature  
2. npj parkinson's disease  
3. articles  
4. article

Clinical subtypes in patients with isolated REM sleep behaviour disorder

Download PDF

• Article
• Open access
• Published: 17 November 2023

Clinical subtypes in patients with isolated REM sleep behaviour disorder

• Aline Seger, 
• Anja Ophey, 
• Christopher E. J. Doppler, 
• Johanna Kickartz, 
• Marie-Sophie Lindner, 
• Maximilian Hommelsen, 
• Gereon R. Fink & 
• Michael Sommerauer 

npj Parkinson's Disease volume 9, Article number: 155 (2023) Cite this article

•  
•  
•  
•  

Abstract

Patients with Parkinson’s disease (PD) show a broad heterogeneity in clinical presentation, and subtypes may already arise in prodromal disease stages. Isolated REM sleep behaviour disorder (iRBD) is the most specific marker of prodromal PD, but data on clinical subtyping of patients with iRBD remain scarce. Therefore, this study aimed to identify iRBD subtypes. We conducted comprehensive clinical assessments in 66 patients with polysomnography-proven iRBD, including motor and non-motor eval‎uations, and applied a two-step cluster analysis. Besides, we compared iRBD clusters to matched healthy controls and related the resulting cluster solution to cortical and subcortical grey matter volumes by voxel-based morphometry analysis. We identified two distinct subtypes of patients based on olfactory function, dominant electroencephalography frequency, amount of REM sleep without atonia, depressive symptoms, disease duration, and motor functions. One iRBD cluster (Cluster I, late onset—aggressive) was characterised by higher non-motor symptom burden despite shorter disease duration than the more benign subtype (Cluster II, early onset—benign). Motor functions were comparable between the clusters. Patients from Cluster I were significantly older at iRBD onset and exhibited a widespread reduction of cortical grey matter volume compared to patients from Cluster II. In conclusion, our findings suggest the existence of clinical subtypes already in the prodromal stage of PD. Future longitudinal studies are warranted that replicate these findings and investigate the risk of the more aggressive phenotype for earlier phenoconversion and dementia development.

초록

파킨슨병(PD) 환자는

임상 양상에서 광범위한 이질성을 보이며,

전구 질환 단계에서 이미 하위 유형이 발생할 수 있습니다.

고립 렘수면 행동 장애(iRBD)는

전구기 PD의 가장 구체적인 마커이지만,

iRBD 환자의 임상적 하위 유형에 대한 데이터는 아직 부족합니다.

따라서

이 연구는

iRBD 하위 유형을 식별하는 것을 목표로 했습니다.

수면다원검사를 통해

iRBD가 입증된 66명의 환자를 대상으로

운동 및 비운동 평가를 포함한 종합적인 임상 평가를 실시하고

2단계 군집 분석을 적용했습니다.

또한

복셀 기반 형태 측정 분석을 통해

iRBD 클러스터를 건강한 대조군과 비교하고

결과 클러스터 솔루션을 피질 및 피질하 회백질 부피와 연관시켰습니다.

후각 기능,

우세한 뇌파 검사 빈도,

운동 무력증이 없는 렘수면의 양,

우울 증상,

유병 기간,

운동 기능에 따라 두 가지 하위 유형의 환자를 식별했습니다.

한 iRBD 클러스터(클러스터 I, 후기 발병-공격성)는

양성 하위 유형(클러스터 II, 조기 발병-양성)보다 질병 기간이 더 짧았음에도 불구하고

비운동 증상 부담이 더 높은 것이 특징이었습니다.

운동 기능은 두 군집 간에 비슷했습니다.

클러스터 I에 속하는 환자들은

클러스터 II에 속하는 환자들에 비해

iRBD 발병 시기가 훨씬 더 빨랐고

대뇌피질 회백질 부피가 광범위하게 감소한 것으로 나타났습니다.

결론적으로,

우리의 연구 결과는

이미 PD의 전구 단계에 있는

임상적 아형이 존재한다는 것을 시사합니다.

향후

이러한 연구 결과를 반복하고

더 공격적인 표현형의 조기 현상 전환 및 치매 발병 위험을 조사하는

종단 연구가 필요합니다.

Similar content being viewed by others

Reorganization of intrinsic functional connectivity in early-stage Parkinson’s disease patients with probable REM sleep behavior disorder

Article Open access03 January 2024

Evolution patterns of probable REM sleep behavior disorder predicts Parkinson’s disease progression

Article Open access05 April 2022

Probable REM sleep behavior disorder is associated with longitudinal cortical thinning in Parkinson’s disease

Article Open access02 March 2021

참고) secondary RBD는 기면발작질환이 있거나, 항우울제 등 약물을 복용시 발생...

Introduction

Cumulating evidence revealed a broad heterogeneity in clinical presentation and disease progression in patients with Parkinson’s disease (PD)1. Many attempts to identify PD subtypes have been carried out in the past, primarily focussing on motor symptoms2. However, PD heterogeneity may not start at disease onset—formally based on the presence of motor symptoms—but already at its prodromal stage during the incipient spread of α-synuclein aggregates3. This stage is characterised by the occurrence of NMS but no or only mild motor symptoms4,5. Hence, analysis of variation in NMS, including sleep disturbances, hyposmia, autonomic dysfunction, and cognitive impairment, might be essential for early identification of subtypes3.

Isolated rapid eye movement (REM) sleep behaviour disorder (iRBD) is presumed to be the most specific marker of prodromal PD6,7. iRBD is characterised by dream-enacting behaviours and the loss of physiological muscle atonia during REM sleep8,9. Longitudinal studies have demonstrated that patients with iRBD not only convert to classic ‘motor-dominant’ PD (~50% of converters), but a nearly similar proportion of patients (~45%) converts to a ‘dementia phenotype’ with relevant cognitive impairment10,11. In stark contrast, ~5% of patients with iRBD convert to multiple system atrophy (MSA)10. Thus, iRBD indicates the stage of an emerging α-synucleinopathy and represents a heterogeneous population. Additionally, considerable variability of the temporal sequence and preval‎ence of NMS as well as time to phenoconversion have been observed in patients with iRBD12.

There have been great efforts in identifying clinical biomarkers of the prodromal stage and predictors for the risk of phenoconversion: advanced age, olfactory loss, abnormal colour vision, pronounced motor symptoms, and non-use of antidepressants were identified as markers for a higher risk for short-term phenoconversion13. It has been shown that higher amounts of REM sleep without atonia (RWSA) point to a faster phenoconversion and to a greater risk of developing mild cognitive impairment (MCI)11,14. Likewise, electroencephalographic (EEG) slowing in iRBD is associated with a higher risk of developing MCI15. However, most studies have focused on univariate analysis using single predictors, disregarding that multiple symptoms might co-exist and yield clusters of distinct phenotypes in iRBD.

A more integrative characterisation of patients with iRBD may advance subtyping in prodromal disease stages and improve our understanding of varying disease development, helping to predict distinct disease courses, which might also be necessary for patient stratification for future clinical trials, e.g., on novel neuroprotective drugs.

We assessed motor and non-motor features, polysomnography data, and grey matter volumes in 66 patients with iRBD and conducted a two-step cluster analysis. To our knowledge, this is the first study subtyping patients with iRBD using comprehensive clinical data and cluster analysis for integrative analysis of multiple biomarkers.

축적된 증거에 따르면

파킨슨병(PD) 환자의 임상 증상과 질병 진행에

광범위한 이질성이 있는 것으로나타났습니다1.

과거에는

주로 운동 증상에 초점을 맞춰

PD 아형을 식별하려는 시도가 많이 이루어졌습니다2.

그러나

PD 이질성은

공식적으로 운동 증상이 나타나는 발병 시점이 아니라

α-시누클레인 응집체가 확산되는 전구 단계에서3 이미 시작될 수 있습니다.

이 단계에서는

NMS(non motor symptom)가 발생하지만

운동 증상이 없거나 경미한 증상만 나타나는 것이 특징입니다4,5.

Hence, analysis of variation in NMS, including

sleep disturbances,

hyposmia,

autonomic dysfunction, and

cognitive impairment,

might be essential for early identification of subtypes3.

따라서

수면 장애,

후각저하 hyposomia,

자율 기능 장애 및 인지 장애를 포함한

non motor symptom의 변화를 분석하는 것이

하위 유형을 조기에 식별하는 데 필수적일 수 있습니다3.

램 수면 행동 장애(iRBD)는

전구기 PD의

가장 구체적인 마커로 추정됩니다6,7.

iRBD는

꿈을 꾸는 행동과 REM 수면 중

생리적 근육 이완의 상실이 특징입니다8,9.

종단 연구에 따르면

iRBD 환자는

전형적인 '운동 우세형' PD로 전환될 뿐만 아니라(전환자의 ~50%),

거의 비슷한 비율의 환자(~45%)가 관련 인지 장애를 동반하는 '치매 표현형'으로 전환되는 것으로

나타났습니다10,11.

이와는 대조적으로,

iRBD 환자의 ~5%는

다계통 위축(MSA)10으로 전환됩니다.

따라서

iRBD는

새로운 α-시누클레인 병증의 단계를 나타내며

이질적인 집단을 나타냅니다.

또한,

iRBD 환자에서

NMS의 시간적 순서와 유병률, 증상전환 phenoconversion까지의 시간에서

상당한 변동성이 관찰되었습니다12.

전구 단계의 임상 바이오마커와

증상전환 위험에 대한 예측 인자를 확인하기 위해

많은 노력을 기울여 왔으며,

고령,

후각 상실,

색각 이상,

뚜렷한 운동 증상,

항우울제 미사용이

단기 증상전환 위험이 높은 마커로 확인되었습니다13.

무운동성 렘수면 REM sleep without atonia (RWSA)이

많을수록 현상 전환이 빨라지고

경도 인지 장애(MCI)11,14가 발생할 위험이 커지는 것으로 나타났습니다.

마찬가지로,

iRBD에서 뇌파(EEG)가 느려지는 것은

MCI 발병 위험이 더 높다는 것과 관련이 있습니다15.

그러나 대부분의 연구는

단일 예측 인자를 사용한 단변량 분석에 초점을 맞추었기 때문에

여러 증상이 공존하여 뚜렷한 표현형 군집을 형성할 수 있다는 점을 간과했습니다.

iRBD 환자를 보다 통합적으로 특성화하면

전구 질환 단계에서 하위 유형화를 발전시키고

다양한 질병 진행에 대한 이해를 개선하여

뚜렷한 질병 경과를 예측할 수 있으며,

이는 향후 새로운 신경 보호 약물 등의

임상시험을 위한 환자 계층화에도 필요할 수 있습니다.

저희는

66명의 iRBD 환자의

운동 및 비운동 특징,

수면다원검사 데이터,

회백질 용적을 평가하고

2단계 클러스터 분석을 실시했습니다.

종합적인 임상 데이터와 여러 바이오마커의 통합 분석을 위한 군집 분석을 사용하여

iRBD 환자를

하위 유형화한 연구는

저희가 아는 한 이번이 처음입니다.

Results

We included 66 patients with iRBD (8 females) and 25 healthy control (HC) subjects (five females). The mean age of patients was 66.8 ± 6.4 years, with an average iRBD duration of 7.9 ± 6.0 years. HC subjects did not differ significantly regarding sex and age. Clinical baseline data of all patients with iRBD and HC subjects are summarised in Supplementary Table 1.

Model characteristics of cluster analysis

Two iRBD clusters were identified by the two-step cluster analysis. Correctly identified items at Sniffin’ Sticks (variable importance index (VII): 1), dominant EEG peak frequency (VII: 0.41), disease duration (VII: 0.34), amount of RSWA (VII: 0.18), depression (VII: 0.11), and motor symptoms (VII: 0.09) were the variables with highest informative value to discriminate two clusters based on the given quality indicators. The silhouette measure of cohesion and separation was 0.3, indicating a fair structure16.

66명의 iRBD 환자(여성 8명)와

25명의 건강한 대조군(HC) 피험자(여성 5명)를 포함했습니다.

환자의 평균 연령은

66.8±6.4세였으며,

평균 iRBD 기간은 7.9±6.0년이었습니다.

HC 피험자는 성별과 연령에 따라 큰 차이가 없었습니다.

모든 iRBD 및 HC 환자의 임상 기준 데이터는 보충 표 1에 요약되어 있습니다.

클러스터 분석의 모델 특성

2단계 클러스터 분석을 통해

두 개의 iRBD 클러스터가 식별되었습니다.

스니핑 스틱에서 올바르게 식별된 항목(가변 중요도 지수(VII): 1), 우세한 뇌파 피크 주파수(VII: 0.41), 질병 지속 기간(VII: 0.34), RSWA의 양(VII: 0.18), 우울증(VII: 0.11), 운동 증상(VII: 0.09)이 주어진 품질 지표를 기반으로 두 클러스터를 구분하는 데 가장 정보적 가치가 높은 변수였습니다. 응집력과 분리의 실루엣 측정값은 0.3으로, 공정한 구조를 나타냅니다16.

Description of iRBD clusters and post-hoc comparison

Detailed characteristics of both iRBD clusters are given in Table 1 and Supplementary Table 2. The first cluster (Cluster I, late onset—aggressive) encompassed 22 patients with iRBD and was characterised by a higher age of onset of RBD symptoms but a shorter disease duration. The second cluster (Cluster II, early onset—benign) encompassed 44 patients with an earlier onset of RBD yet a longer disease duration (age at onset, Cluster I: 62.4 ± 6.2 years vs. Cluster II: 56.9 ± 7.7 years, t(64) = −2.935, p = 0.005; disease duration, Cluster I: 4.4 ± 2.6 years vs. Cluster II: 9.7 ± 6.5 years, U = 238.500, z = −3.339, p < 0.001). Still, both clusters of patients had comparable ages at iRBD diagnosis.

Table 1 Characteristics of iRBD clusters and healthy controls.

Full size table

Further, patients from Cluster I showed a higher amount of RSWA (Cluster I: 44.4 ± 10.6% vs. Cluster II: 35.6 ± 15.3%, t(64) = −2.431, p = 0.018) and slowing of EEG peak frequency (Cluster I: 8.4 ± 0.7 Hz vs. Cluster II: 9.5 ± 1.1 Hz, U = 221.500, z = − 3.576, p < 0.001). Besides, patients of Cluster I experienced subjective cognitive decline (SCD) in more cognitive domains (Cluster I: 2.1 ± 1.6 vs. Cluster II: 0.7 ± 0.9, H(2) = 15.378, z = 3.731, p = 0.001), although no difference in the Montreal Cognitive Assessment (MoCA) testing was observed (Cluster I: 27.2 ± 1.7 vs. Cluster II: 27.5 ± 2.0). They showed a higher burden of depressive symptoms (Beck Depression Inventory-II (BDI-II), Cluster I: 8.1 ± 7.1 vs. Cluster II: 4.8 ± 7.1, H(2) = 9.095, z = -2.439, p = 0.044) and hyposmia (Sniffin’ Sticks, Cluster I: 4.1 ± 2.1 vs. Cluster II: 7.9 ± 1.8, H(2) = 46.587, z = 4.772, p < 0.001) (Fig. 1). Within Cluster I, three subjects took antidepressive medication, in Cluster II, this was the case for two subjects. The clusters did not significantly differ regarding motor symptoms, orthostatic blood pressure dysregulation, subjective sleep disturbances, anxious symptoms, and non-motor symptoms as assessed with the non-motor symptom questionnaire (NMSQ).

Fig. 1: Comparison of patients with iRBD from Cluster I (late onset—aggressive) and Cluster II (early onset—benign).

A Key clinical markers of clusters (blue = Cluster I, red = Cluster II). MDS-UPDRS III, disease duration and EEG Peak frequency were compared using Mann-Whitney U test. Age at onset and RSWA were compared using Student’s t-test. BDI-II and Sniffin’ Sticks were compared using Kruskal-Wallis test with post-hoc Dunn-Bonferroni. *p < 0.05, **p < 0.01, ***p < 0.001. B Voxel-based morphometry analysis of grey matter volume. Blue indicates reduced grey matter volume in patients from Cluster I compared to Cluster II at p < 0.05 (FDR-corrected). Abbreviations: BDI-II Beck Depression Inventory, EEG electroencephalography, MDS-UPDRS III Movement Disorders Society—Unified Parkinson’s disease rating scale part III, REM rapid eye movement, RSWA REM sleep without atonia.

Full size image

Voxel-based morphometry (VBM) analysis revealed a more widespread grey matter volume loss in patients from Cluster I compared to patients from Cluster II with significantly lower grey matter volume in the occipital, frontal and temporal lobes, the caudate nucleus, and the cerebellum (Fig. 1).

Similar to the between-cluster analysis, patients from Cluster I showed a higher burden of depressive symptoms and declared SCD more frequently than HC subjects. Additionally, Cluster I patients suffered more frequently from anxiety symptoms and subjective sleep disturbances than HC subjects. Both clusters of patients with iRBD significantly differed from HC subjects in RBD screening questionnaire (RBDSQ), Apathy Eval‎uation Scale (AES), and NMSQ scores as well as in olfactory function.

Discussion

In this study, using comprehensive clinical phenotyping and cluster analysis, we could identify two distinct subtypes of patients with iRBD. One subgroup of patients (Cluster I) was characterised by a late onset—aggressive phenotype: despite comparable age at diagnosis and motor symptom burden, these patients had a shorter disease duration, older age at RBD symptom onset, higher RSWA amount, EEG slowing, pronounced hyposmia, and accentuated neuropsychiatric symptoms including SCD and depressive symptoms compared to patients with iRBD from the early onset—benign subtype (Cluster II). Additionally, patients from Cluster I exhibited a more widespread decrease in grey matter volume compared to Cluster II.

Notably, a nearly equal proportion of patients with iRBD convert to a dementia-dominant phenotype and classic, motor-predominant PD. Even though, we cannot predict the outcome of phenoconversion in our sample due to the cross-sectional design of our study, the different clusters might represent corresponding different prodromal disease subtypes7,17. Recently, operationalised prodromal criteria for DLB were published and proposed biomarkers, besides iRBD, were EEG slowing, cortical grey matter volume loss, and the occurrence of neuropsychiatric symptoms—all of these biomarkers were features of Cluster I17. Moreover, a recent study revealed that cortical grey matter loss in patients with iRBD was linked to a greater risk of developing MCI18. Patients from Cluster I reported SCD in more cognitive domains than Cluster II. SCD is presumed to be an intermediate state between age-appropriate cognition and MCI. Thus, SCD is considered a risk factor for developing dementia19,20,21. Hence, patients from Cluster I are likely to be at higher risk of developing DLB or PD-D. Depressive symptoms may impact the experience of SCD22; however, mean BDI-II total scores of patients from both clusters did not reach the proposed cut-off of 9 as an indicator of mild depression23.

Interestingly, post-mortem studies have suggested that the extent of olfactory impairment is not correlated to α-synuclein pathology in the olfactory bulb but with a more general and widespread cortical and subcortical α-synuclein pathology24,25. This observation fits well with our findings that patients of Cluster I had pronounced olfactory impairment, which was the variable with the highest informative value to discriminate the two clusters of patients. Moreover, patients from Cluster I also exhibited reduced cortical grey matter volume. Secondly, our finding on age-related subtypes aligns with previous reports in patients with PD, proposing a higher age at disease onset to predict a more aggressive PD phenotype26,27. Ageing, in general, is one of the most significant risk factors for developing PD28, and its influence on disease-related factors, such as genetic variants, is well established29. Despite this epidemiological evidence, the interplay between ageing effects and neurodegenerative processes is still poorly understood. In general, the manifold clinical presentation and disease course of PD may rely on multiple individual factors independent of age30,31. Yet, the cause of a potential interindividual or tissue-specific vulnerability remains unclear. Aside from host-specific factors, specific α-synuclein strains may contribute to the diverse clinical phenotypes32. To add more complexity, additional neuropathological changes, i.e., amyloid aggregates, might add to the α-synuclein pathology, and the molecular structures of α-synuclein aggregates might not only differ between individuals but also across brain areas32,33,34.

Surprisingly, emerging motor symptoms were the least relevant factor within our cluster solution. This finding may be attributed to the circumstances that patients with iRBD express only mild motor symptoms, which are hardly differentiated by the Movement Disorder Society—Unified Parkinson’s Disease Rating Scale Part III (MDS-UPDRS III) due to a floor effect of the scale, potentially hampering the detection of subtle differences between groups. More sensitive motor assessments might have elucidated differences between the subgroups35. Conversely, our findings emphasise the importance of including NMS in subtyping α-synucleinopathies, particularly in the early stages.

Our study has several limitations and strengths. Most importantly, we only used cross-sectional data as longitudinal follow-up was yet unavailable. Hence, the impact of the detected subtypes on the future disease course remains speculative and it is possible that our assumptions may not be confirmed. It should be taken into consideration that the clusters may to some point represent different stages of the disease, e.g. subjects of cluster I may be closer to phenoconversion36. Additionally, some patients with iRBD might not convert at all due to misdiagnosis or an extraordinary benign course of the disease. However, this proportion might be rather small as a recent meta-analysis estimated that >95% of patients with iRBD phenoconvert eventually. We also included subjects with younger age at anticipated iRBD onset. Although patients with iRBD with younger age of onset may have a lower likelihood of having a neurodegenerative aetiology of RBD, additional investigations, e.g. olfactory performance, DaTSCAN imaging or skin biopsies, strengthened the likelihood of an underlying synucleinopathy in these subjects (Supplementary Table 3). As soon as longitudinal data is available in our cohort, we will eval‎uate the impact of the subtypes identified by this study on disease progression. However, despite the lack of longitudinal data, we observed a significantly shorter self-reported symptom duration in the late onset—aggressive cluster in the present analysis. Nevertheless, self-reported symptom duration might be biased by the awareness of a subject as there are no objective markers available to collect information about age at disease onset and disease duration at this point. One strength of the applied cluster analysis is that we used data from multiple clinically relevant domains. Applying a similar cluster analysis approach in existing, deeply-phenotyped, longitudinal iRBD cohorts would be of high value to validate our findings. It must be noted that the outcome of each cluster solution highly depends on the variables included in the model. Therefore, we aimed to include one objectively assessed biomarker for a variety of motor and non-motor categories to avoid highly correlating markers within the same category, and followed Mestre et al.’s recommendations on reporting the subtyping approach37. Still, neuropsychiatric symptoms mainly were assessed through questionnaires which depend on the individuals’ subjective perception and the extend of neuropsychiatric symptoms may have been biased by the recruitment via newspaper advertisement (e.g., subjects with major depression or pronounced cognitive decline are less likely to actively respond to an advertisement).

In conclusion, this study demonstrated distinct clinical subtypes in patients with iRBD, elucidating relevant differences in the expression‎ of symptoms and potential disease trajectories primarily based on non-motor assessments. We are convinced that subtyping patients within the prodromal stage will improve the understanding of the underlying pathological pathways and hopefully help guide therapeutic decisions in the near future.

토론

이 연구에서는 종합적인 임상 표현형과 군집 분석을 사용하여 두 가지 뚜렷한 하위 유형의 iRBD 환자를 확인할 수 있었습니다. 한 하위 환자군(클러스터 I)은 후기발병-공격적 표현형의 특징을 보였는데, 진단 시 연령과 운동 증상 부담은 비슷했지만 이 환자들은 조기발병-양성 하위 유형(클러스터 II)의 iRBD 환자에 비해 유병 기간이 짧고, RBD 증상 발병 연령이 높았으며, RSWA 양이 많고, EEG 둔화, 뚜렷한 후각저하, SCD 및 우울 증상을 포함한 신경정신과적 증상이 강조된 것이 특징이었습니다. 또한 클러스터 I의 환자들은 클러스터 II에 비해 회백질 부피가 더 광범위하게 감소한 것으로 나타났습니다.

특히, 거의 동일한 비율의 iRBD 환자가 치매 우세 표현형과 전형적인 운동 우세 PD로 전환되는 것으로 나타났습니다. 연구의 단면 설계로 인해 표본의 표현형 전환 결과를 예측할 수는 없지만, 서로 다른 클러스터는 그에 상응하는 서로 다른 전구 질환 아형을 나타낼 수 있습니다7,17. 최근에 DLB에 대한 전구 질환 기준이 발표되었으며, iRBD 외에 제안된 바이오마커는 뇌파 둔화, 피질 회백질 용적 손실, 신경정신과적 증상 발생으로, 이 모든 바이오마커는 클러스터 I17의 특징이었습니다. 또한, 최근 연구에 따르면 iRBD 환자의 피질 회백질 손실이 MCI 발병 위험과 관련이 있는 것으로 나타났습니다18. 클러스터 I의 환자들은 클러스터 II보다 더 많은 인지 영역에서 SCD가 보고되었습니다. SCD는 연령에 적합한 인지 능력과 MCI 사이의 중간 상태로 추정됩니다. 따라서 SCD는 치매 발병의 위험 요인으로 간주됩니다19,20,21. 따라서 클러스터 I에 속하는 환자는 DLB 또는 PD-D 발병 위험이 더 높을 가능성이 높습니다. 우울 증상은 SCD 경험에 영향을 미칠 수 있지만22, 두 클러스터 환자의 평균 BDI-II 총점은 경미한 우울증의 지표로 제안된 컷오프인 9점에 도달하지 못했습니다23.

흥미롭게도, 사후 연구에 따르면 후각 장애의 정도는 후각구의 α-시누클린 병리보다는 더 일반적이고 광범위한 피질 및 피질하 α-시누클린 병리24,25와 상관관계가 있는 것으로 나타났습니다. 이 관찰은 두 군집의 환자를 구별하는 데 가장 높은 정보 가치를 가진 변수였던 군집 I의 환자들이 뚜렷한 후각 장애를 가지고 있다는 연구 결과와 잘 맞아떨어집니다. 또한 클러스터 I의 환자들은 대뇌 피질 회백질 부피도 감소한 것으로 나타났습니다. 둘째, 연령 관련 하위 유형에 대한 우리의 발견은 PD 환자의 이전 보고와 일치하며, 질병 발병 연령이 높을수록 더 공격적인 PD 표현형을 예측할 수 있다고 제안합니다26,27. 일반적으로 노화는 PD 발병의 가장 중요한 위험 요인 중 하나이며28, 유전적 변이와 같은 질병 관련 요인에 미치는 영향은 잘 알려져 있습니다29. 이러한 역학적 증거에도 불구하고 노화 영향과 신경 퇴행성 과정 사이의 상호 작용은 아직 제대로 이해되지 않고 있습니다. 일반적으로 파킨슨병의 다양한 임상 증상과 질병 경과는 나이와 무관하게 여러 개별 요인에 따라 달라질 수 있습니다30,31. 그러나 잠재적인 개인 간 또는 조직별 취약성의 원인은 여전히 불분명합니다. 숙주 특이적 요인 외에도 특정 α-시누클레인 균주가 다양한 임상 표현형에 기여할 수 있습니다32. 더 복잡하게는 아밀로이드 응집체와 같은 추가적인 신경 병리학적인 변화가 α-시누클린 병리에 추가될 수 있으며 α-시누클린 응집체의 분자 구조는 개인마다 다를 뿐만 아니라 뇌 영역에 따라 다를 수 있습니다32,33,34.

놀랍게도, 새로운 운동 증상은 클러스터 솔루션에서 가장 관련성이 낮은 요인이었습니다. 이 결과는 iRBD 환자가 경미한 운동 증상만 표현하는 상황에서 비롯된 것으로 보이며, 이는 척도의 바닥 효과로 인해 운동 장애 학회 통합 파킨슨병 평가 척도 파트 III(MDS-UPDRS III)로는 거의 구분되지 않아 그룹 간의 미묘한 차이를 감지하는 데 방해가 될 수 있습니다. 보다 민감한 운동 평가를 통해 하위 그룹 간의 차이를 설명할 수 있었을 것입니다35. 반대로, 우리의 연구 결과는 특히 초기 단계에서 α- 시누클레인 병증의 하위 유형 분류에 NMS를 포함하는 것이 중요하다는 점을 강조합니다.

이 연구에는 몇 가지 한계와 강점이 있습니다. 가장 중요한 것은 아직 종단적 추적 관찰이 불가능했기 때문에 단면 데이터만 사용했다는 점입니다. 따라서 발견된 하위 유형이 향후 질병 경과에 미치는 영향은 아직 추측에 불과하며, 우리의 가정이 확인되지 않을 수도 있습니다. 클러스터가 어느 시점에서는 질병의 다른 단계를 나타낼 수 있다는 점을 고려해야 합니다(예: 클러스터 I의 대상자는 페노컨버전에 가까울 수 있음36). 또한 일부 iRBD 환자는 오진 또는 비정상적인 양성 경과로 인해 전혀 전환되지 않을 수도 있습니다. 그러나 최근 메타분석에 따르면 iRBD 환자의 95% 이상이 결국에는 페노컨버전하는 것으로 추정되므로 이 비율은 다소 적을 수 있습니다. 또한 iRBD 발병이 예상되는 연령이 더 어린 피험자도 포함했습니다. 발병 연령이 어린 iRBD 환자는 RBD의 신경 퇴행성 병인이 있을 가능성이 낮을 수 있지만, 후각 기능, DaTSCAN 영상 또는 피부 생검과 같은 추가 조사를 통해 이러한 피험자의 기저 시누클레인 병증이 있을 가능성이 강화되었습니다(보충 표 3). 코호트에서 종단 데이터가 확보되는 대로 이 연구에서 확인된 하위 유형이 질병 진행에 미치는 영향을 평가할 예정입니다. 그러나 종단 데이터가 부족함에도 불구하고, 본 분석에서는 발병 후기공격적 집단에서 자가 보고 증상 기간이 현저히 짧은 것으로 관찰되었습니다. 그럼에도 불구하고 현재 시점에서 질병 발병 연령과 질병 기간에 대한 정보를 수집할 수 있는 객관적인 마커가 없기 때문에 자가 보고 증상 기간은 대상자의 인식에 의해 편향될 수 있습니다. 적용된 클러스터 분석의 한 가지 강점은 임상적으로 관련된 여러 영역의 데이터를 사용했다는 점입니다. 기존의 심층 표현형 종단형 iRBD 코호트에서 유사한 클러스터 분석 접근법을 적용하면 연구 결과를 검증하는 데 큰 도움이 될 것입니다. 각 클러스터 솔루션의 결과는 모델에 포함된 변수에 따라 크게 달라진다는 점에 유의해야 합니다. 따라서 동일한 범주 내에서 높은 상관관계가 있는 마커를 피하기 위해 다양한 운동 및 비운동 범주에 대해 객관적으로 평가된 하나의 바이오마커를 포함하고자 했으며, 하위 유형화 접근법 보고에 대한 Mestre 등의 권장 사항을 따랐습니다37.하지만 신경정신과적 증상은 주로 개인의 주관적 인식에 의존하는 설문지를 통해 평가되었으며, 신문 광고를 통한 모집으로 인해 신경정신과적 증상의 범위가 편향되었을 수 있습니다(예: 주요 우울증이나 뚜렷한 인지 저하가 있는 대상자는 광고에 적극적으로 응답하지 않을 가능성이 높음).

결론적으로, 이 연구는 주로 비운동 평가를 기반으로 증상 발현과 잠재적 질병 궤적의 관련 차이를 규명하여 iRBD 환자에서 뚜렷한 임상적 하위 유형을 보여주었습니다. 전구기 단계의 환자를 하위 유형화하면 근본적인 병리 경로에 대한 이해가 향상되고 가까운 미래에 치료 결정을 내리는 데 도움이 될 것으로 확신합니다.

Methods

Participants

Participating patients with iRBD were recruited from our local iRBD cohort at the University Hospital Cologne38. The cohort was consecutively recruited from the general population by newspaper advertisements including the German version of the single-question screen for RBD (RBD1Q) followed by a structured telephone screening. Inclusion criteria for full screening was the answer “yes” to the RBD1Q. The structured telephone screening included demographic data, medical history and sleep questionnaires (RBD screening questionnaire (RBDSQ), Pittsburgh Sleep Quality Index (PSQI), STOP-Bang questionnaire, Epworth Sleepiness Scale (ESS), Regensburg Insomnia Scale (RIS), and screening for Restless-Legs-Syndrom). Exclusion criteria for full screening were any known neurological disorder, age <35 years or >80 years, early age of symptom onset (<35 years), alcohol or drug abuse, and having a pacemaker38. Based on information of the telephone screening selected subjects were invited to video-polysomnography (PSG). All candidates were asked to stop antidepressive medication 2 weeks before PSG (n = 5). Subjects diagnosed with iRBD according to the International Classification of Sleep Disorders III criteria for RBD39 were invited for a clinical assessment and underwent MRI scan.

For the current cluster analysis, we only included patients with (self-reported) age at onset over 40 years. Furthermore, the current analysis only included patients with a completed clinical assessment. In self-eval‎uation questionnaire data, this was defined as at least 80% valid data within a questionnaire. If an entire questionnaire was missing, subjects were excluded. For comparison, we included clinical datasets of 25 matched HC subjects without a known movement or sleep disorder who participated in independent studies at the Department of Neurology of the University Hospital Cologne. The local ethic committee of the Medical Faculty of the University of Cologne approved the study. All participants gave written informed consent before participation.

Clinical assessment

Assessment of patients with iRBD included the collection of disease-related features, PSG data, non-motor and motor testing, and self-eval‎uation questionnaires from different categories:

Demographic data

•  
•  

RBD-related features

•  
•  
•  
•  

PD motor symptom severity

•  

Autonomic function

•  
•  

Cognition

•  
•  
•  

Neuropsychiatric symptoms

•  
•  
•  
•  

Sleep

•  
•  
•  

Olfaction

•  

General non-motor symptom burden

•  

Assessments of HC subjects included self-eval‎uation questionnaires (BDI-II, BAI, FSMC, AES, NMSQ, SCOPA-AUT, ESS, PDSS), olfactory testing (Sniffin‘ Sticks) and cognitive assessment (MoCA). HC subjects did not undergo PSG, orthostatic testing, motor examination, or magnetic resonance imaging (MRI).

Image acquisition and preprocessing

MRI measurements were obtained during clinical routine using a 1.5 T Siemens MR scanner (Siemens, Erlangen, Germany) at the Department of Radiology, University Hospital Cologne. T1-weighted brain images of 57 iRBD patients were collected and acquired using a magnetisation-prepared rapid acquisition with gradient-echo (MP-RAGE) sequence with the following parameters: 7.6 ms repetition time, 3.5 ms echo time, 8 degree flip angle, 150 slices, 266 × 246 × 142 mm field of view, 280 × 216 matrix resolution (voxel size: 0.95 × 0.95 × 0.95 mm3). Data were preprocessed and analysed using the CAT12 toolbox (https://neuro-jena.github.io/cat/). Images were reoriented and aligned to the anterior commissure, followed by segmentation into grey matter (GM), white matter (WM), and cerebrospinal fluid (CSF). The resulting images were normalised to the Montreal Neurological Institute (MNI) space, modulated using the Jacobian determinant, and smoothed using a Gaussian kernel with a value of 8 mm full width at half maximum.

Statistical analysis

Statistical analyses were performed using SPSS 28.0. Single missing values were imputed by group-wise (iRBD vs. HC) means. We carried out a two-step cluster analysis allowing a balanced inclusion of categorical and continuous variables. Continuous variables were z-standardised (M = 0, SD = 1) to unify the range of their values. Cluster solutions based on 1 to 15 clusters were compared and the most suitable solution (i.e., the number of clusters) was selected according to the lowest Akaike’s Information Criterion (AIC) eval‎uated across cluster solutions. A variable importance index as implemented in SPSS with a range from 0 to 1, with 1 indicating the highest importance, was reported for each variable contributing to the final cluster solution. As the resulting clustering is highly dependent on the entered variables, we aimed to include one objective biomarker (e.g., RSWA instead of RBDSQ) of each category to avoid highly correlating variables of the same category (for categories, see “Clinical Assessment”). Different combinations were applied to the two-step clustering algorithm, and finally, the best solution was chosen based on quality indicators (i.e., silhouette measure of cohesion and separation, and AIC). Additional cluster solutions are presented in Supplementary Fig. 1.

Data were checked for normal distribution using the Shapiro Wilk test and the Kolmogorov Smirnov test. Chi-square-tests, one-way ANOVA, Kruskal-Wallis tests with post hoc Dunn-Bonferroni, and Student’s t- or Mann–Whitney U-tests were performed for comparisons between resulting clusters and HC subjects, as appropriate. If not stated otherwise, data are presented as mean value ± standard deviation. P-values < 0.05 were considered significant.

To analyse differences in VBM, we compared the smoothed GM images of the two groups resulting from the cluster analysis (Cluster I, n = 18; Cluster II, n = 39) using a two-sample t-test. Total intracranial volume, calculated using the CAT12 toolbox, was included as a covariate to correct for differences in brain sizes. The resulting second-level model was analysed using a non-parametric permutation test with 5000 permutations performed by the TFCE (threshold-free cluster enhancement) toolbox included in CAT12. The statistical significance threshold was set to p < 0.05 (FDR-corrected).

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

The datasets used during the current study are available from the corresponding author on request. The data are not publicly available due to the inclusion of information that could compromise the participants’ privacy.

References

1.  
2.  
3.