Diffusion Imaging of Gray Matter Microstructure in Alzheimer’s Disease

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by the accumulation of amyloid-β (Aβ) plaques, neurofibrillary tangles of tau, and neuroinflammation. These pathological changes lead to synapse loss, neuronal death, and macroscopic brain atrophy, manifesting clinically as cognitive decline. AD has a protracted pre-clinical phase during which these pathological processes unfold years before symptoms emerge. 1 Neuroimaging enables the quantification of these changes in vivo, which has aided in the characterization of AD pathophysiology across its various stages.

Magnetic resonance imaging (MRI)-based diffusion methods quantify water molecule diffusion within tissues and have been used to assess microstructural changes associated with neurodegeneration. While many diffusion MRI studies have focused on white matter tract changes using diffusion tensor imaging (DTI), DTI has also been used to investigate gray matter (GM) changes in AD with a focus on mean diffusivity (MD), the average magnitude of water diffusion in all directions within an MRI voxel. 2 Applying DTI models to cortical structures presents unique challenges. The thin cortex often results in partial volume effects with adjacent cerebrospinal fluid, the complex cellular architecture of GM makes it difficult to ascribe specific diffusion changes to underlying biological processes, and the lower anisotropy of GM often translates to a lower signal-to-noise ratio in diffusion measurements.3,4, 3,4 Despite these challenges, previous studies have shown that changes in cortical diffusivity are powerful imaging biomarkers that precede macroscopic volume loss and may even serve as better predictors of clinical disease progression than gross hippocampal atrophy. 5

Additional multi-compartment diffusion methods, such as free water imaging (FWI), have been developed to attempt to differentiate water molecules that are relatively unrestricted (free water) from those with more restricted motion due to interactions with cellular structures. FWI employs a bi-tensor model that separates the diffusion signal into components representing free water (FW) and water diffusing within more complex environments. 6 In a recent issue of the Journal of Alzheimer’s Disease, Nakaya et al. report on the “Assessment of Gray Matter Microstructural Alterations in Alzheimer’s Disease by Free Water Imaging.” 7 In this study, the authors measured macrostructural differences in gray matter volume using voxel-based morphometry, and microstructural differences using DTI measures and FWI in 31 participants who were Aβ-positive with clinical AD and 40 cognitively unimpaired (CU) individuals who were Aβ-negative. The participants with AD exhibited significantly higher FW, MD, and radial diffusivity (RD) values compared to the CU participants in regions commonly affected by AD, with the largest effect sizes observed in the amygdala, posterior cingulate, hippocampus, parahippocampus, and precuneus. These diffusion changes were of broader spatial extent than the macrostructural changes of volume loss which were focused in the hippocampus, providing insight into GM microstructural changes in AD and demonstrating the potential of diffusion imaging (including FWI and DTI) as a sensitive and early biomarker for AD.

The authors highlight that FW shows larger effect sizes compared to DTI metrics (MD or RD) for detecting differences between AD and CU groups. They suggest this is because FWI more specifically assesses extracellular fluid alterations in AD, as AD-related microstructural changes like neurodegeneration and inflammation increase extracellular water with unrestricted diffusion. In contrast, they argue, DTI metrics like MD measure average diffusion within a voxel, including both intra- and extracellular space water diffusion, making them less specific to AD-related changes. This concept will need further investigation as the magnitude and spatial extent of group differences in both FW and MD were similar, suggesting that they are highly correlated values with overlapping contributions to their signal.

Multi-modal imaging studies with inclusion of diffusion MRI outcomes give clues to the microstructural features underlying FW and MD signal changes in AD. In their previous study, Nakaya et al. found a significant positive correlation between FW and [¹⁸F]THK5351, a PET marker of tau deposition and possibly astrogliosis. 8 In our previous work, we demonstrated strong inverse correlations between MD and synaptic density in many regions associated with AD pathology. 9 Such studies support the idea that AD pathology can alter tissue microstructure and increase MD and FW. Furthermore, comparing diffusion MRI methods incorporating single (e.g., DTI) and multi-compartment (e.g., FWI and Neurite Orientation Dispersion and Density Imaging) 10 models can further elucidate the biological mechanisms behind these changes, clarifying the respective contributions of neurite density, orientation dispersion, and extracellular water to the observed diffusion signal alterations. These techniques are a valuable tool in future multi-modal and longitudinal imaging studies that can offer insights into AD neurobiology.