Advertisement

GRAPH-THEORY ANALYSIS SHOWS A HIGHLY EFFICIENT BUT REDUNDANT NETWORK IN MCI TAU PROPAGATION

  • Sulantha S. Mathotaarachchi
    Affiliations
    McGill University, Montreal, QC, Canada

    McGill University Research Centre for Studies in Aging, Verdun, QC, Canada

    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada
    Search for articles by this author
  • Tharick A. Pascoal
    Affiliations
    McGill University, Montreal, QC, Canada

    McGill University Research Centre for Studies in Aging, Verdun, QC, Canada

    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada

    Douglas Hospital Research Centre, Verdun, QC, Canada

    Centre for Studies on Prevention of Alzheimer’s Disease (StoP-AD Centre), Douglas Mental Health Institute, Verdun, QC, Canada
    Search for articles by this author
  • Monica Shin
    Affiliations
    McGill University Research Centre for Studies in Aging, Verdun, QC, Canada

    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada

    Cerebral Imaging Centre - Douglas Research Centre, Verdun, QC, Canada
    Search for articles by this author
  • Andrea Lessa Benedet
    Affiliations
    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada
    Search for articles by this author
  • Min-Su Kang
    Affiliations
    Translational Neuroimaging Laboratory, McGill University Research Centre for Studies in Aging- McGill University, Montreal, QC, Canada
    Search for articles by this author
  • Hanne Struyfs
    Affiliations
    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada

    University of Antwerp, Antwerp, Belgium

    icometrix, Leuven, Belgium

    Reference Center for Biological Markers of Dementia (BIODEM), Laboratory of Neurochemistry and Behavior, Institute Born-Bunge, University of Antwerp, Antwerp, Belgium
    Search for articles by this author
  • Kok Pin Ng
    Affiliations
    McGill University Research Centre for Studies in Aging, Verdun, QC, Canada

    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada

    National Neuroscience Institute, Singapore, Singapore
    Search for articles by this author
  • Joseph Therriault
    Affiliations
    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada
    Search for articles by this author
  • Vladimir S. Fonov
    Affiliations
    Montreal Neurological Institute, McGill University, Montreal, QC, Canada
    Search for articles by this author
  • Serge Gauthier
    Affiliations
    McGill University Research Centre for Studies in Aging, Verdun, QC, Canada

    Douglas Hospital Research Centre, Verdun, QC, Canada
    Search for articles by this author
  • Misic Bratislav
    Affiliations
    Montreal Neurological Institute, Montreal, QC, Canada
    Search for articles by this author
  • Pedro Rosa-Neto
    Affiliations
    McGill University, Montreal, QC, Canada

    McGill University Research Centre for Studies in Aging, Verdun, QC, Canada

    Translational Neuroimaging Laboratory- McGill University, Verdun, QC, Canada

    Centre for Studies on Prevention of Alzheimer’s Disease (StoP-AD Centre), Douglas Mental Health Institute, Verdun, QC, Canada

    Douglas Mental Health University Institute, Montreal, QC, Canada

    McConnell Brain Imaging Centre - McGill University, Montréal, QC, Canada
    Search for articles by this author

      Background

      The stages of neurofibrillary tangles (NFT) during the course of Alzheimer's disease is well understood through the Braak stages; however, the underlying mechanism behind the NFT propagation remains unclear. Here, we propose a graph theory based method to identify the patterns of NFT deposit propagation using [18F]AV1451 Tau positron emission tomography (PET) images.

      Methods

      [18F]AV1451 images of 38 cognitively normal (CN) and 34 mild cognitive impairment (MCI) individuals were acquired from the ADNI cohort and the standardized uptake value ratio (SUVr) maps were generated using the cerebellar grey matter as the reference region. Group-based Tau networks of CN and MCI were then constructed from 201 nodes distributed across the cerebral cortex and using Pearson correlation coefficients based on 100 bootstrap samples. The networks were corrected for multiple comparisons using False Discovery Rate (FDR) and thresholded at r >= 0.5. Network measures such as density, average path length, global efficiency, clustering coefficient, and small worldness were calculated for each of the bootstrap sample. Welch two sample t-test was used to compare each measure across both subject groups.

      Results

      The density of the MCI Tau network was significantly higher compared to CN [Figure 1]. Furthermore, the average path length was significantly lower in MCI when compared to the CN network, resulting in a significantly higher global efficiency. The clustering coefficient of the MCI network was significantly higher than the CN network. Lastly, the small worldness parameter for the MCI network was significantly lower compared to the CN [Figure 2].

      Conclusions

      Deposition of Tau proteins in the MCI [Figure 3] stage involves an increased number of long range and short range network connections (redundancy) [Figure 1] indicated by the increased density, clustering coefficient, and decreased average path length. Nonetheless, the small world property decreases mainly due to the increase in long range connections in the MCI group. These results indicate that in the CN stage, local Tau deposition patterns are confined within the local structural boundaries and as the disease progresses, deposition patterns expand beyond their structural boundaries.
      Figure thumbnail fx1
      Figure thumbnail fx2
      Figure thumbnail fx3