Premium
Network‐level relationships between cortical neurodegeneration and FDG‐PET hypometabolism across clinical and A/T/N subgroups in AD
Author(s) -
Stocks Jane,
Karteek Popuri,
Martersteck Adam,
Beg Mirza Faisal,
Wang Lei
Publication year - 2020
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 - 10.1002/alz.041618
Subject(s) - neuroimaging , atrophy , disease , psychology , connectome , concordance , alzheimer's disease neuroimaging initiative , white matter , pathological , neuroscience , neurodegeneration , medicine , alzheimer's disease , pathology , functional connectivity , magnetic resonance imaging , radiology
Background Converging evidence suggests that Alzheimer’s disease (AD) affects multiple large‐scale brain networks. Determining the relationship between network‐level alterations in neuroimaging‐derived metrics of brain structure and function and whether concordance between metrics varies by clinical presentation or pathological status can increase our understanding of the disease process in AD. To this aim, we evaluated the relationship between cortical thinning (neurodegeneration) and cortical glucose hypometabolism within distributed brain networks in individuals across clinical (AD, MCI who later progress to AD, MCI who remain stable, and normal controls) and CSF‐defined A/T/N subgroups. Method T1‐MPRAGE and FDG‐PET scans were downloaded from the Alzheimer’s Disease Neuroimaging Initiative website (cohorts ADNI‐1&2 at baseline only). Regional W‐score maps were computed for each patient and each imaging modality, to adjust for the effect of normal aging. Using a well‐validated atlas from the Human Connectome Project/Glasser et al. (2016) and the corresponding cortical network solution by Ji et al. (2019), network‐level Pearson correlations were computed between cortical thickness and FDG metabolism, for each participant , reflecting individual consistency in the degree of atrophy and hypometabolism at each network. Multivariate analysis of variance models assessing the effects of clinic diagnostic category and, separately, A/T/N status, on correlations while accounting for age, education and APOE‐4 status were examined. Results Among clinical diagnostic groups, atrophy‐hypometabolism correlations showed significant differences in 11 of 12 networks (Table 1), with correlations increasing by disease severity specifically within networks known to be affected by AD (e.g., frontoparietal, default mode, ventral multimodal; Figures 1 and 2). Among A/T/N subgroups, significant differences in correlation scores were found only within select networks including frontoparietal and default mode networks (Table 2; Figure 3). Conclusion With increased disease severity, cortical atrophy is increasingly related to cortical FDG hypometabolism across brain networks. In the A/T/N framework, our findings suggest that both amyloid and tau increase this structure‐function relationship. However, considerable heterogeneity was present among all A/T/N subgroups, warranting further exploration. Multimodal neuroimaging analyses can unravel the structure‐function relationships that contribute to clinical outcomes and diagnostic uncertainty in AD.