Functional eigenvector centrality dynamics are related to amyloid deposition in preclinical Alzheimer’s Disease

Luigi Lorenzini*, Silvia Ingala, Lyduine E. Collij, Viktor Wottschel, Sven Haller, Kaj Blennow, Giovanni B. Frisoni, Gael Chetelat, Pierre Payoux, Pablo Martinez-Lage, Michael Ewers, Adam Waldman, Joanna M. Wardlaw, Craig W. Ritchie, Juan Domingo Gispert, Henk Jan Mutsaerts, Pieter Jelle Visser, Philip Scheltens, Betty M. Tijms, Frederik BarkhofAlle Meije Wink

*Corresponding author for this work

Research output: Contribution to journalComment/Letter to the editorAcademicpeer-review

Abstract

Background: In preclinical Alzheimer’s disease (AD), amyloid accumulates in highly-functionally connected brain regions. This selective vulnerability is related to the high neuronal fluctuations, typical of these regions. Dynamic functional connectivity (FC) was introduced to investigate network organization over time, with high network variations indicating regional flexibility, hence promoting functional integration. The relation of early amyloid deposition with FC dynamics remains unclear. Eigenvector centrality (EC) evaluates a node's importance in functional networks, both for the whole functional MRI time series (“static” EC) or within sliding-windows (“dynamic” EC). We studied the association of cerebrospinal fluid (CSF) amyloid load with static and dynamic EC in non-demented individuals from the European Prevention of Alzheimer’s Dementia (EPAD) cohort. Methods: Data for 701 non-demented participants were available. CSF Aß1-42 levels <1000 pg/mL were defined as amyloid positive (A+) (Elecsys assay). Both static and dynamic voxel-wise EC were computed from rs-fMRI time series. Static EC differences between A+ and A- participants were assessed using linear models. Standard deviation and range of dynamic EC were compared between A+ and A- participants within significant clusters of static EC differences, and within 10 resting-state networks (RSN). Linear models were used to determine interaction between amyloid and cognitive performance on dynamic EC variability (Figure 1). Results: Demographics and clinical characteristics are shown in Table 1. A+ participants showed lower static EC values in parietal and occipital clusters, and higher static EC values in a medio-frontal cluster (Figure 2). The medio-frontal cluster had also lower dynamic EC variability in A+ (Figure 3). The default mode (Figure 4) and the visual networks of A+ participants also showed lower dynamic EC variability (p < 0.001). Dynamic properties in the DMN and visual networks were differentially associated with cognitive domains in the A- and A+ groups, with lower variability found in A+ participants with higher cognitive scores (Figure 5). Conclusion: We demonstrate that early amyloid deposition affects static and dynamic EC, possibly by reducing involvement of functional hubs in different network dynamics and functional integration, thus relating to cognitive dysfunctions. Our data suggest dynamic EC as an early biomarker of the interplay between early amyloid deposition and cognitive decline.
Original languageEnglish
Article numbere064631
JournalAlzheimer's & Dementia
Volume18
Issue numberS1
DOIs
Publication statusPublished - 1 Dec 2022

Fingerprint

Dive into the research topics of 'Functional eigenvector centrality dynamics are related to amyloid deposition in preclinical Alzheimer’s Disease'. Together they form a unique fingerprint.

Cite this