Abstract
Diffusion tensor imaging has traditionally been used to assess white matter (WM) integrity in Alzheimer's disease (AD). However, the tensor model is limited in modeling complex WM structure. Neurite Orientation Dispersion and Density Imaging (NODDI), a cutting-edge technique applied to multishell diffusion MRI, can offer more precise insights into microstructural features of WM integrity. We assessed whether NODDI more sensitively detects AD-related changes in medial temporal lobe (MTL) WM than traditional tensor metrics. In total, 199 older adults with multishell diffusion MRI from ADNI3 (mean age = 75 years; 60% female; cognitively unimpaired, n = 121; cognitively impaired MCI/dementia, n = 77) were analyzed. Tensor metrics of fractional anisotropy (FA) and mean diffusivity (MD), as well as NODDI metrics of Neurite Density Index (NDI) and orientation dispersion Index (ODI), were calculated for MTL WM tracts (JHU Atlas: hippocampal cingulum, fornix column/body, fornix/stria terminalis, and uncinate). A subset of participants received 18F-florbetapir or 18F-florbetaben to measure Aβ (n = 146; converted to Centiloids), 18F-flortaucipir to measure tau (n = 135), and neuropsychological testing including the Clinical Dementia Rating Sum of Boxes (CDR-SB) and memory composite score (ADNI-MEM). NODDI measures in MTL tracts were more strongly correlated with cognitive performance and AD pathology than standard tensor measures. For example, entorhinal tau was strongly associated with NDI in the cingulum hippocampus and the uncinate, and with ODI in the fornix ST. Both ODI and NDI across the majority of tracts were associated with CDR-SB and ADNI-MEM. In contrast, FA in any MTL tract was not significantly correlated with either tau or global amyloid-beta, while MD in MTL tracts showed limited correlations with pathology or cognition. NDI partially mediated the relationship between AD pathology (entorhinal tau, meta temporal tau, or Aβ) and the memory composite score. Random forest modeling showed that a combination of NODDI metrics and DTI had the strongest estimates of memory performance and clinical impairment. NODDI metrics offer more sensitive insights about MTL WM integrity in AD that could have been previously missed due to the limitations of DTI analyses. Additionally, combining NODDI with DTI yielded the strongest predictive performance for memory and clinical impairment, suggesting that the two approaches are complementary. The use of advanced diffusion acquisitions such as multishell which allows for analyses such as NODDI is crucial for the future development of disease identification and structural understanding.