Altered network efficiency in isolated REM sleep behavior disorder: A multicentric study

Abstract

Introduction: Isolated rapid eye movement (REM) sleep behavior disorder (iRBD), characterized by abnormal movements during REM sleep, is a prodromal stage of dementia with Lewy bodies (DLB) and Parkinson's disease (PD). While iRBD shows emerging brain changes, their impact on structural connectivity and network efficiency, and their predictive value, remain poorly characterized.

Methods: In this international prospective study, 198 polysomnography-confirmed iRBD patients and 174 controls underwent diffusion magnetic resonance imaging and were analyzed. Cutting-edge diffusion tractography and network-based statistics were applied to reconstruct individual connectomes and assess network properties predicting DLB or PD.

Results: Structural architecture was already disrupted in iRBD, with both reduced and compensatory increased connections. Global efficiency was decreased. Local efficiency in motor regions was altered and associated with early clinical symptoms. Altered local efficiency in the supramarginal gyrus predicted DLB only.

Discussion: Early disruption of brain architecture in iRBD predicts progression to synucleinopathy-related dementia, offering a novel potential prognostic biomarker.

Highlights: Isolated rapid eye movement sleep behavior disorder (iRBD) patients show significant alterations in inter-regional structural connectivity. Global efficiency is reduced in iRBD compared to controls. Areas with increased local efficiency contribute to decreased global efficiency. Altered network efficiency is associated with emerging Parkinsonian features. Higher supramarginal efficiency predicts dementia with Lewy bodies in iRBD.

Keywords: dementia with Lewy bodies; diffusion magnetic resonance imaging; graph theory; parasomnias; sleep; structural connectivity; synucleinopathies.

FIGURE 1

Main steps of the processing pipelines. The diffusion‐weighted images were first acquired from different cohorts with the T1w images. The T1w processed with FreeSurfer v.7.1.1 and DWI were used in TractoFlow‐ABS v.2.4.4 to build a tractogram for each subject in the iRBD and healthy control groups. The Cammoun atlas with 448 cortical and 14 subcortical regions was registered in the T1w native space (ANTs registration) and quality control (dMRIQcpy) was performed for each main step of TractoFlow‐ABS. Subsequently, the Connectoflow pipeline (v.1.1.0) was used with COMMIT2 to build a connectome for each subject (representing the number of streamlines, adjusted for the connection length, between each region of the atlas). The iRBD group average and standard deviation is shown for visualization. DWI, diffusion‐weighted imaging; iRBD, isolated rapid eye movement sleep behavior disorder; T1w, T1‐weighted.

FIGURE 2

Flowchart showing the number of subjects in the iRBD and HC groups at each main step of the method and exclusion criteria. DWI, diffusion‐weighted imaging; HC, healthy control; iRBD, isolated rapid eye movement sleep behavior disorder.

FIGURE 3

Structural connectivity changes in patients with iRBD. A, Circular plot showing the structural connections with significantly lower (red) and higher (blue) connection density in iRBD compared to controls (P value_FDR < 0.05 using NBS with age, sex, and scanner as covariates). Subregions of the Cammoun atlas have been combined into 78 regions for visualization. B, Brain map representations of the 27 subregions showing significantly lower connectivity (red areas) and the 6 subregions with stronger connectivity (blue areas) in iRBD compared to controls. FDR, false discovery rate; iRBD, isolated rapid eye movement sleep behavior disorder; LH, left hemisphere; NBS, network based statistics; RH, right hemisphere.

FIGURE 4

Local efficiency alterations in iRBD. A, Cortical and subcortical regions showing significant decreases (red) and increases (blue) in local efficiency in the iRBD group compared to controls (P value_FDR < 0.05). B, Decreases and increases in local efficiency in specific cortical and subcortical regions are associated with the reduced global efficiency observed in iRBD. FDR, false discovery rate; iRBD, isolated rapid eye movement sleep behavior disorder; LH, left hemisphere; RH, right hemisphere.

FIGURE 5

Relationships between local efficiency alterations in the iRBD group and graph theory metrics. A, Subcortical and cortical brain maps illustrating significant regional differences in iRBD compared to controls (W scores) for clustering coefficient, connection strength, and degree (number of connections). B, Significant positive spatial correlations were found between local efficiency alterations in iRBD and the clustering coefficients, while significant negative correlations were observed between local efficiency and both connection strength and degree in iRBD. All correlations (Pearson r) were compared against null coefficient distributions using a model that preserves spatial autocorrelation between regions, with FDR correction (significance: P value_spin‐FDR  < 0.05). FDR, false discovery rate; iRBD, isolated rapid eye movement sleep behavior disorder.

FIGURE 6

Kaplan–Meier survival curves showing the association between local efficiency in the right supramarginal gyrus (W score) and phenoconversion in iRBD. A, Time to phenoconversion in DLB. Individuals with iRBD with a positive W score (blue line) in the right supramarginal gyrus showed a significantly lower survival probability (i.e., higher risk of conversion), compared to those with a negative W score (pink line); B, Time to phenoconversion in PD. No significant difference in survival was observed between individuals with iRBD with positive (blue) and negative (pink) W scores. P value (p) indicates statistical significance from the log‐rank test. DLB, dementia with Lewy bodies; iRBD, isolated rapid eye movement sleep behavior disorder; MRI, magnetic resonance imaging; PD, Parkinson's disease.

FIGURE 7

Schematic representation of potential network alterations in iRBD. A, Example of network disruption in iRBD compared to controls, where neighboring regions (A, B, E) lose connections (black lines) between them, resulting in reduced local efficiency and clustering coefficient in the central region. B, Example of network reorganization leading to increased local efficiency and clustering coefficient along with lower degree (number of direct connections) and average connection strength (thinner red lines) in a region. Increased strength (or density) in specific connection (thicker red line) can happen among more isolated neighbor regions (G and F) in the context of loss of connections and inputs, due to potential competition reduction. iRBD, isolated rapid eye movement sleep behavior disorder.