PREDICTING FUTURE DEMENTIA USING COGNITION, APOE, AND STRUCTURAL MAGNETIC RESONANCE IMAGING: THE ROLE OF WHITE MATTER INTEGRITY

Abstract

Background: The topographical relationships between tau, amyloid-beta (Ab) and neurodegeneration in Alzheimer’s Disease (AD) are incompletely understood. Methods: Eighty-one amyloidpositive patients (age 64.469.5, N1⁄440/81 Females, MMSE 21.4666.05, CDR-sb 4.362.27) meeting clinical criteria for AD or MCI due to AD and available F-Flortaucipir (FTP), C-PiB (PIB), F-FDG (neurodegeneration/hypometabolism) PET and 3T MRI (neurodegeneration/atrophy) were included. W-score maps were created and binarized for each imaging-modality (thresholded W>1.64, p<0.05), adjusting for age, sex and total Intracranial Volume (sMRI) based on amyloid-negative cognitively normal adults (PIB/FTP/MRI controls: N1⁄446, age 69.35618.7; FDG controls: N1⁄438, age 69.2616.2). FDG-PET and atrophy W-maps were combined into neurodegeneration maps (ND), with a voxel classified as ND+ when positive on either FDG or MRI. Spatial extent was estimated as the proportion of supra-threshold voxels in the cortical gray matter (GM) for each imaging-modality or PIB6/FTP6/ ND6 combinations. Results: Amyloid accumulation was the most extensive (w90% of the GM involved, Fig.1A), followed by tau (w71%) and neurodegeneration (w40%) (p<0.001 Friedman Test). The most frequent hierarchies of spatial extents were PIB>FTP>ND (found in 79% of the patients), FTP>PIB>ND (13%) andPIB>ND>FTP (6%). The different hierarchieswere associated with age (p1⁄40.03 Fisher Exact Test), with FTP>PIB>ND over-represented in younger patients and PIB>ND>FTP over-represented in older patients (Fig.1B). At the group-level, the majority of abnormal voxels in the whole cortical GM were PIB+/FTP+/ND+ (w34%), followed by PIB+/FTP+/ND(w31%) and PIB+/FTP-/ ND(w20%), with other combinations showing small percentages (range 1-5%) (Fig.2A). PIB+/FTP+/ND+ voxels weremostly located in bilateral temporo-parietal regions (Fig.2B) and their proportion strongly correlatedwith global cognition (i.e. MiniMental Score Examination MMSE, r1⁄4-0.688, p<0.001), outperforming all other PIB6/FTP6/ND6 voxel classifications (Fig.2C). Conclusions: On average, amyloid-PET changes were more extensive than tau and neurodegeneration imaging changes, matching the expected temporal relationships between biomarkers, but notably with some agerelated heterogeneity. Neurodegeneration frequently co-localized with both amyloid/tau pathology accumulation, suggesting a possible shared vulnerability or synergistic toxic effects. Global cognition correlated with the proportion of PIB+/FTP+/ND+ cortical voxels suggesting that cognitive declinemay be explained by the confluence of molecular pathology and neurodegeneration.