Abstract
PURPOSE: To explore neurodynamic bases underlying subjective cognitive decline (SCD) based on edge-centric functional network. METHODS: 211 SCD patients and 210 healthy controls (HC) were recruited from the Alzheimer's Disease Neuroimaging Initiative. Edge time series (ETS) were obtained based on resting-state functional magnetic resonance data. The top 10% co-fluctuation signals of all time points in ETS were extracted to construct the high-amplitude frame networks, and the co-fluctuation signals from the remaining time points were used to construct the low-amplitude frame networks. In both network states, the graph theory and network-based statistics (NBS) analyses were used to compare SCD and HC. The correlation of the imaging indicators with cognitive scores and apolipoprotein E (APOE) ε4 genes was performed by Spearman correlation analysis. RESULTS: SCD exhibited lower peak amplitude and longer trough-to-trough duration (TTD) compared to HC. In both network states, the normalized clustering coefficient, normalized characteristic path length, small-worldness, and global efficiency of SCD were significantly reduced, and the altered nodal centralities of SCD predominantly exhibited a decreasing trend. However, the high-amplitude frame network identified more altered brain regions compared to the low-amplitude frame network. Furthermore, a SCD-related subnetwork was found in the high-amplitude frame network, which was composed of 11 brain regions and 13 edges. TTD was positively related to the number of APOE ε4 genes; the normalized characteristic path length, the betweenness centrality of right postcentral gyrus, and the connection between bilateral angular gyrus were correlated with cognitive scores. CONCLUSION: Our findings demonstrate that the edge-centric network framework reveals details of brain network alterations in SCD through different perspectives, and these alterations hold potential as novel biomarkers for SCD.