Glucose metabolism in hyper-connected regions predicts neurodegeneration and speed of conversion in Alzheimer's disease.

Abstract

Here, we combined a longitudinal design to assess whole-brain hyper- and hypo-connectivity in the different clinical phases of Alzheimer's disease (AD) with a multimodal approach to understand how such connectivity changes were related to glucose hypometabolism. We selected a longitudinal cohort of N = 66 subjects with clinical, cerebrospinal fluid and FDG-PET assessments, from Alzheimer's Disease Neuroimaging Initiative (ADNI) database. N = 31 AD individuals were assessed at three stages: mild cognitive impairment (AD-MCI, T0), early phase of dementia (mild-AD, T1) and dementia (AD-D, T2). We included N = 35 age/sex-matched healthy controls. We assessed longitudinal metabolic connectivity using Pearson's correlation, clustering analysis and graph theory metrics. In the MCI-AD stages, hypo- and hyper-connectivity coexisted. Data-driven, longitudinal clustering analysis identified specific pathological clusters: a default mode network cluster, with prevalent hypo-connectivity and severe, persistent hypometabolism; a limbic cluster showing hyper-connectivity and steeper metabolic decline. Metabolism in hyper-connected limbic regions showed a mediation effect on worsening of AD-like parieto-temporal hypometabolism and predicted faster conversion to dementia. Hypo- and hyper-connectivity, especially in early stages, may have different roles in AD neurodegenerative processes, with metabolism in hyper-connected regions acting as a mediator on the neurodegeneration of core regions of AD pathology.