Premium
O3‐08‐01: PHENOTYPES OF CARDIOVASCULAR DAMAGE AND THE ASSOCIATION WITH COGNITIVE DECLINE AMONG OLDER ADULTS: A LATENT CLASS ANALYSIS APPROACH
Author(s) -
Miller Lindsay M.,
Wu Chenkai,
Hirsch Calvin,
Lopez Oscar L.,
Cushman Mary,
Odden Michelle
Publication year - 2019
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 - 10.1016/j.jalz.2019.06.4666
Subject(s) - medicine , phenotype , subclinical infection , disease , cognitive decline , brain natriuretic peptide , cardiology , dementia , heart failure , gene , biochemistry , chemistry
Background: Emerging in vivo models indicate that the spread of tau pathology in the human brain occurs through neuroanatomically and functionally connected regions. However, it remains to be determined how in vivo models can show evidence of sex-specific differences in brain-wide tau propagation. Methods: We have utilized graph theoretical analysis applied to [F]flortaucipir PET data to explore patterns of transcortical tau spread in cognitively normal (CN; 123 male and 178 female) and mild cognitive impairment (MCI; 101 male and 60 female) ADNI subjects. We extracted SUVR values from 70 [F]flortaucipir PET brain regions and visualized networks of tau-connected brain regions (nodes) by calculating correlations between regional PET SUVRs across subjects (edges). The Louvain algorithm was used for modularity detection in each group and relevant network information (e.g., clustering coefficient and betweenness centrality) was calculated. For comparison, a support vector machine (SVM) algorithm was implemented for the sex-specific diagnostic group classification based on regional PET SUVRs as features. Results: Three main communities (modularity classes) were detected in tau networks of CN and MCI male groups (figure 1). The majority of nodes within class 1 were Braak stages 1-4 regions while class 2 nodes were Braak 4-5 regions and class 3 nodes were Braak regions 5 and 6 associated with late-stage tauopathy. Several nodes with high betweenness centrality (BC) were identified. In CN-male, these included the (predominately left-hemispheric) parahippocampus, supramarginal gyrus, and precuneus, which served as major bridges between early, middle, and late-stage tau communities. The CN-female group had a higher graph density, less modularity, and higher BC (p1⁄4 0.03) compared to CN-males. Conversely, their network clustering coefficient was lower (p<0.0001) than CNmale. The MCI-female group had the highest graph density and lowest modularity. The SVM accuracy was modest (59.7% in CN and 63% in MCI group) (figure 2). Conclusions: The [F]flortaucipir PET network architecture closely resembles the temporal progression of tauopathy across (early-mid-late) Braak stages and highlights sex-specific differences. Tau networks in women indicate highly interconnected regions and less modularity than men, which may confer particular vulnerability of women to brainwide tau propagation.