Multimodal topographic properties of white matter functional segregation in Alzheimer's disease: A simultaneous PET/MR study.

OBJECTIVE: White Matter (WM) functional networks offer insight into the network processes underlying progression along the Alzheimer's disease (AD) continuum. Amyloid (Aβ) retention in WM regions, as measured by Aβ-PET, reflects WM integrity and is associated with cognition. This study used simultaneous PET/MR imaging to investigate topographic alterations in WM functional networks and Aβ-burden networks, aiming to provide objective neuroimaging evidence of white matter-related pathology in Alzheimer's disease. METHODS: A total of 85 subjects were included in this study, including 42 patients with AD, 24 patients with Prodromal AD (PAD), and 19 age- and sex-matched healthy controls (HC). Cognitive assessments were collected along with clinical information, including the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA). Every enrolled subject underwent an integrated 18F-Florbetapir (18F-AV45) PET/MR scan. High-resolution T1-weighted structural images, blood oxygen level dependent fMRI (BOLD-fMRI) and diffusion tensor imaging (DTI) data were acquired for all subjects. K-means clustering was utilized to obtain white matter functional networks (BWM) from BOLD-fMRI. Normalized Aβ standardized uptake value ratio (SUVR) was calculated within each BWM, with the whole cerebellum as the reference region. The group covariance matrix was generated by Pearson correlation to construct functional, structural, and Aβ retention networks. Topographic properties including the clustering coefficient, characteristic path length and global efficiency were calculated. RESULTS: AD patients showed lower mean MMSE and MoCA scores than HC and PAD subjects (pcorrected < 0.001). In the functional network, higher characteristic path length and lower global efficiency were found in AD patients, compared with the PAD and HC groups (both p < 0.05). Higher characteristic path length and lower global efficiency were observed in the Aβ retention network of the AD group (all p < 0.05). Furthermore, the clustering coefficient was lower in AD patients (both p < 0.05). The AD group exhibited a significant difference in functional-Aβ retention coupling compared with the PAD group (p = 0.002) and the HC group (p < 0.001). No significant difference was found in the structural network. CONCLUSION: We observed cognitive impairment in AD patients as compared with participants of the HC or PAD group. AD patients also showed higher characteristic path length and lower global efficiency in both functional and Aβ retention networks. These results may highlight the importance of the multimodal topographic properties of white matter functional segregation for early diagnosis and personalized management of AD.