Advertisement

COGNITIVE COMPLAINTS IN OLDER ADULTS AT RISK FOR ALZHEIMER’S DISEASE ARE ASSOCIATED WITH ALTERED RESTING STATE NETWORKS

      Background

      Pathophysiological changes that may have a disruptive influence on brain networks accompany or anticipate early clinical symptoms in prodromal Alzheimer’s disease (AD). Resting state fMRI (rsfMRI) combined with brain connectomics permits assessment of changes in whole-brain functional connectivity (FC) including downstream effects on neuronal processes. FC patterns can be grouped into resting-state networks (RSNs) allowing analysis of higher-order changes within and between networks. Here we investigate the relationship of self-perceived cognitive changes to rsfMRI FC changes in a sample from the Indiana Memory and Aging Study.

      Methods

      Participants included 58 older adults classified as cognitively normal (CN, 13), subjective cognitive decline (SCD, 16), early amnestic MCI (EMCI, 5), late MCI (LMCI, 16), and mild AD dementia (AD, 8) who underwent baseline rsfMRI processed with an in-house pipeline after Power et al, [1] to extract FC matrices based on a functional parcellation including 278 regions. An independent component analysis (ICA) connectivity data-driven approach (connICA) was used to extract FC independent patterns (FC traits). FastICA decomposition (15 independent components) was performed over a matrix of all subjects FC connectivity profiles (Figure 1). Each component signal was then used as a response in a multilinear regression model with cognitive variables (Cognitive Complaint Index (CCI) [2] scores, episodic memory and executive function domain scores) serving as the predictors and nuisance variables (age, gender, and education) included as covariates.

      Results

      Two connICA components were strongly associated with CCI scores (FC-traits 1, 2). FC trait 1 involves a decrease in FC within each RSN whereas FC trait 2 involves an increased FC within specific somatomotor regions, and increased inter-RSN FC between somatomotor and dorsal attention networks and a general FC decrease between somatomotor and other RSNs. In both cases CCI is shown to be the best predictor of FC traits 1 and 2. FC trait 5 could not be attributed to a specific variable but is worth noting due to its robust finding (Figure 2).

      Conclusions

      Self-reported cognitive complaints are strongly associated with a pattern of specific rsFC network changes as the disease progresses. Further examination of psychometric performance and FC patterns is required. [1] Power et al (2014) Neuroimage; [2] Saykin et al. (2006)Neurology.
      Figure thumbnail fx1
      Figure thumbnail fx2