Directed graph‐based longitudinal model for spatiotemporal dynamics of amyloid, tau, and neurodegeneration in the Alzheimer’s disease spectrum

Background β amyloid deposition, pathological tau, and neurodegeneration are regarded as critical biomarkers to biologically define Alzheimer’s disease (AD) in living people. Development of tracers for positron emission tomography (PET) enables extracellular amyloid plaques and intraneuronal neurofibrillary tangles to be detected in vivo. Identifying in vivo spatiotemporal dynamics of those three biomarkers could be a key to understand the disease progression in Alzheimer’s continuum. Methods We recruited 74 healthy controls, 61 amnestic mild cognitive impairment patients, and 38 AD patients with T1‐weighted magnetic resonance image (MRI) and two PET scans ([18F]Flortaucipir for tau and [18F]Florbetaben for Aβ) with about two‐year follow‐up data. Longitudinal W‐scores were calculated from standardized uptake value ratios (SUVRs) for two PET data and cortical thicknesses for MRI data. All participants were aligned in the order of standardized baseline tau SUVRs using the frequency‐based tau spreading model. Partial correlations (PC) were then computed between every regional pair of baseline and follow‐up data for each biomarker within a time window involving a part of participants. Directed graphs were constructed based on the PCs for each biomarker by moving the time window through the temporal participant‐orders. Results Amyloid seemed to follow the Thal stage. With the early window most connections existed within neocortex, and the connections moved to allocortex. Tau also seemed to follow the Braak stage showing the connections advanced from entorhinal regions to neocortex as the subject window moves to late order. However, temporal domain of amyloid and tau connections within the same area were distinct each other. Especially for tau, we found that the network hubs varied temporarily and diameters of the temporal graphs seemed to increase substantially as the hub moves out of the medial temporal lobe. Neurodegeneration showed similar trends to tau with later temporal domain. Conclusions In this study, we suggested the in vivo temporal alterations of interregional affections for amyloid, tau, and neurodegeneration from asymptomatic to clinical AD. All biomarkers exhibited analogous trends to previous researches and the moments of appearances for each biomarker were different. To detect these peculiar spatiotemporal characteristics between biomarkers might be valuable to figure out AD progression.