Fang Ji1, Ofer Pasternak2, Eric Kwun Kei Ng1, Joanna Su Xian Chong1, Siwei Liu1, Liwen Zhang1, Hee Youn Shim1, Yng Miin Loke1, Boon Yeow Tan3, Narayanaswamy Venketasubramanian4, Christopher Li-Hsian Chen5, Juan Zhou1
1Duke-NUS Medical School, Singapore, Singapore, 2Harvard Medical School, Boston, USA, 3St Luke's Hospital, Singapore, Singapore, 4Raffles Hospital, Singapore, Singapore, 5National University of Singapore, Singapore, Singapore
Memory deficits are typically the most salient cognitive symptom of Alzheimer's disease (AD) and its prodromal stage, amnestic mild cognitive impairment (aMCI) (Scheltens, et al., 2016). Works from our group and others identified patterns of abnormalities in volumetric, microstructural white matter, and functional connectivity in the different stages (Berlot, et al., 2014; Ji, et al., 2017; Maier-Hein, et al., 2015; Pasternak, et al., 2012). However, little is known about the relation of these brain abnormalities with memory, especially in the prodromal stage of AD. Previous studies assumed a constant relationship between brain structural and functional abnormalities and memory impairment (Mielke, et al., 2012; Moodley and Chan, 2014). Yet, such brain-behaviour relationship may vary as the disease severity increases.
To cover these gaps, we assessed data collected from 116 memory clinic participants (41 aMCI, 36 AD, and 39 age-matched controls). We calculated the association of memory deficits with microstructural white matter measures (free-water (FW) and corrected fractional anisotropy (FAT)), with grey matter volume, with functional connectivity of the default mode and hippocampal-cortical memory networks, and with the volume of white-matter hyperintensities. Moreover, to probe for variation in the association between abnormal brain measures and memory deficits along the disease continuum (measured by clinical dementia rating sum-of-boxes (CDR-SB)), we used a novel sparse varying coefficient model (Hong, et al., 2015).
The WM microstructural abnormalities (widespread FW increases and reduced FAT in the fornix) were associated with verbal memory impairment in the aMCI patients; yet this association was not significant in clinical stage (Fig. 1A & B). In contrast, medial temporal lobe atrophy was associated with memory deficits only in AD (Fig. 1C). Notably, the association between functional disconnections and memory impairment was significant in both the prodromal and clinical stages of AD (Fig. 1D & E). More importantly, rather than assuming constant effect within the aMCI and AD groups separately, the novel SVC model revealed differential dementia severity-dependent effects of multimodal brain measures on memory deterioration: reduced FAT was more dominating in the early aMCI stage; the influence of FW increases peaked at the late aMCI stage; atrophy was more strongly associated with reduced memory performance in the clinical phase. An association between functional disconnections (in the HCN and DMN) and memory deficits was present and continued during AD progression (Fig. 2).
Together, we demonstrated differential severity-dependent associations between multimodal brain abnormalities and memory performance: the effect of the focal fornix tissue deterioration and widespread mild vascular and/or inflammation preceded those of neural loss in the medial temporal regions; in parallel, functional network degeneration was associated with memory deficits in both the prodromal and clinical AD. Our findings may guide future early intervention strategies to slow down memory decline.
Disorders of the Nervous System:
Alzheimer's Disease and Other Dementias 1
Multi-Modal Imaging 2
Modeling and Analysis Methods:
Diffusion MRI Modeling and Analysis
WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC
1|2Indicates the priority used for review