Premium
Relationships of time‐varying resting state network stability and cognitive function along the Alzheimer’s disease spectrum
Author(s) -
Chumin Evgeny J.,
Risacher Shan L.,
West John D.,
Apostolova Liana G.,
Farlow Martin R.,
McDonald Brenna C.,
Saykin Andrew J.,
Sporns Olaf
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.040993
Subject(s) - resting state fmri , cognition , stability (learning theory) , neuroimaging , alzheimer's disease , psychology , neuroscience , medicine , disease , computer science , machine learning
Background Resting‐state functional connectivity (rsFC) neuroimaging studies of Alzheimer’s Disease (AD) have reported alterations in network community structure and time‐varying rsFC (tvFC). However, the temporal stability of community organization in tvFC has not been investigated. Therefore, the purpose of this work was to characterize the relationship of tvFC and cognitive function (CF) along the spectrum of AD progression. Methods Data were part of the Indiana Alzheimer Disease Center and Indiana Memory and Aging study cohorts, randomly split into discovery and validation samples (Table1). After standard preprocessing, rsFC (10min acquisition) and tvFC (∼60sec tapered, partially overlapping windows) were estimated for two cortical brain region parcellations (200 and 300 nodes). Modularity was estimated for rsFC and for all tvFC windows, across a range of community scales, with final network stability metric calculated as area under the curve of the mean temporal agreement across scales, within and between seven canonical resting‐state networks (RSNs). The Montreal Cognitive Assessment score was used as an index of overall CF. Spearman partial correlation (adjusted for age, education, and sex) was used to investigate relationships of tvFC stability with CF. Results No significant relationships were identified between CF and rsFC or tvFC community structure metrics (quality and number of communities). For tvFC, temporal stability of two network blocks, between visual‐frontoparietal and within ventral attention, showed significant associations with CF across both samples and both parcellations. Additionally, in the 300‐node parcellation, visual network metrics correlated with CF (Figure1). CF was negatively associated with tvFC community stability of regions belonging to different RSNs (Figure1A) and positively with regions within RSNs (Figure1B,C). Conclusion Overall, these findings demonstrate a loss in the stability of RSNs in tvFC with decreasing CF. The networks identified here are generally implicated in cognitive control as well as orientation and attention to salient stimuli. These findings demonstrate that tvFC stability (independent of community and parcellation size) is related to CF along a spectrum of AD risk. Temporal dynamics at rest have potential as a biomarker to characterize progression in prodromal AD. Longitudinal studies are needed to assess the predictive validity of RSN dynamics.