Effects of Alzheimer's and Vascular Pathologies on Structural Connectivity in Early- and Late-Onset Alzheimer's Disease

Abstract

Early- and late-onset Alzheimer's disease (AD) patients often exhibit distinct features. We sought to compare overall white matter connectivity and evaluate the pathological factors (amyloid, tau, and vascular pathologies) that affect the disruption of connectivity in these two groups. A total of 50 early- and 38 late-onset AD patients, as well as age-matched cognitively normal participants, were enrolled and underwent diffusion-weighted magnetic resonance imaging to construct fractional anisotropy-weighted white matter connectivity maps. [¹⁸F]-THK5351 PET, [¹⁸F]-Flutemetamol PET, and magnetic resonance imaging were used for the evaluation of tau and related astrogliosis, amyloid, and small vessel disease markers (lacunes and white matter hyperintensities). Cluster-based statistics was performed for connectivity comparisons and correlation analysis between connectivity disruption and the pathological markers. Both patient groups exhibited significantly disrupted connectivity compared to their control counterparts with distinct patterns. Only THK retention was related to connectivity disruption in early-onset AD patients, and this disruption showed correlations with most cognitive scores, while late-onset AD patients had disrupted connectivity correlated with amyloid deposition, white matter hyperintensities, and lacunes in which only a few cognitive scores showed associations. These findings suggest that the pathogenesis of connectivity disruption and its effects on cognition are distinct between EOAD and LOAD.

Keywords: amyloid; early-onset AD; late-onset AD; positron emission tomography; small vessel disease; tau; white matter connectivity.

FIGURE 1

Study overview. A connectivity matrix for each group was created using T1- and diffusion-weighted images. We compared the networks between the AD groups and age-matched control groups, before W-score networks were generated from the AD groups with the respective reference control group. W-score networks were also compared and analyzed for the correlations with topological pathology maps comprising multiple factors drawn from T1 and PET images. Identified subnetworks from the correlation analyses were used to assess associations with cognitive scores.

FIGURE 2

Subnetworks identified by connectivity comparisons in EOAD and LOAD with respective age-matched controls through network-based statistics. The red circle areas are representative regions, with nodal degree exceeding the mean plus standard deviation of all nodal degrees.

FIGURE 3

Subnetworks identified by correlation analyses between W-score networks and pathologies in (A) EOAD and (B) LOAD through cluster-based statistics. The factors used include global and regional THK retention in EOAD and global FLUTE retention, white matter hyperintensity volume, and lacunes in LOAD. Regional retention indicates the averaged retention between regions at both ends of an edge. The red circles are representative regions, with nodal degree exceeding the mean plus standard deviation of all nodal degrees. Edges within different clusters were displayed with different colors.