Within‐ and across‐network relationships between cortical atrophy and hypometabolism across A/T/N subgroups of the Alzheimer's disease continuum.

Background: Neurodegeneration caused by pathological protein aggregation in Alzheimer’s disease (AD) is reflected in cortical atrophy and glucose hypometabolism, measured with structural MRI and FDG‐PET, but how these disease processes relate to each other within and across brain networks remains unclear. We examine multimodal neuroimaging relationships within and between multiple large‐scale brain networks among CSF‐defined (i.e., amyloid, total tau, and p‐tau) “A/T/N” subgroups of AD. Method: Alzheimer’s Disease Neuroimaging Initiative subjects with CSF analysis, MRI and FDG‐PET: AD‐Continuum (A+(T or N ‐/+), N = 559), Suspected Non‐AD Pathologic Change (SNAP) (A‐(T or N+), N = 149), Normal (A‐T‐N‐, N = 179). T1‐MPRAGE scans underwent FreeSurfer processing to estimate cortical thickness, normalized 18FDG‐PET images were co‐registered and uptake values were projected onto corresponding cortical surface. Using a well‐validated atlas, MRI and FDG‐PET images were parcellated into 360 cortical “patches” corresponding to 12 cortical networks. Partial Pearson correlations were computed between the 360 patches of cortical thickness and FDG metabolism and displayed visually as heatmaps, organized by network. Bonferroni correction was used to account for multiple comparisons. Result: Limited significant correlations were seen in Normal subjects (Figure 1). Among SNAP subjects, positive within‐network correlations between atrophy and hypometabolism were found within the dorsal attention, frontoparietal and default‐mode networks (Figure 2). Across‐network relationships were also present with hypometabolism in the dorsal attention and frontoparietal networks associated with widespread cortical atrophy. In AD‐Continuum subjects, positive within‐network correlations between cortical atrophy and glucose hypometabolism were most frequent within the dorsal attention, default‐mode and posterior multimodal networks (Figure 3). Across‐network relationships were frequent with atrophy within the dorsal attention and default‐mode networks associated with widespread hypometabolism and hypometabolism within the dorsal attention, frontoparietal and default‐mode networks associated with widespread atrophy. Conclusion: These findings suggest that, in both SNAP and AD‐Continuum subjects, local and distant relationships between cortical atrophy and hypometabolism are common, especially within regions known to be impacted by AD neuropathology (e.g., frontoparietal and default‐mode networks). SNAP subjects show fewer across‐network correlations as compared to AD‐Continuum, supporting the conclusion that tau aggregation drives local neurodegeneration while the relationships between amyloid and neurodegeneration are less region‐specific. [ABSTRACT FROM AUTHOR]