Intracellular fluid accumulation underlies brain volume increases in early Alzheimer's disease

In the preclinical stages of Alzheimer's disease, increased brain volume has been associated with amyloid-beta pathology, particularly in regions that undergo volume reductions as the disease progresses. Glial reactivity and water diffusion alterations have been linked to such macroscopic volumetric changes. Brain volume reductions have also been reported following amyloid-beta removal with anti-amyloid therapies with beneficial clinical effects, but it remains unclear whether these changes result from resolving amyloid-triggered neuroinflammation or neurodegeneration. Intravoxel incoherent motion modelling based on multi-shell diffusion-weighted imaging may provide a better understanding of the processes underlying these paradoxical changes. This study used intravoxel incoherent motion diffusion MRI to examine how alterations in cerebral water pools contribute to increased brain volume linked to amyloid-beta deposition and neuroinflammation in cognitively unimpaired individuals. We developed a three-compartment diffusion MRI model with four parameters of cerebral water diffusion: slow diffusion coefficient, fast diffusion coefficient, slow signal portion, and perfusion fraction. We computed these diffusion parameters in 297 cognitively unimpaired late middle-aged adults, 35% of whom showed evidence of amyloid deposition. We examined their correlation with demographic factors (age, sex, apolipoprotein E status), markers of Alzheimer's disease pathology, neurodegeneration, neuroinflammation, and mean diffusivity. Then, we identified regions showing grey matter volume increases related to amyloid burden and examined the association between grey matter volume and diffusion parameters within these regions. We did not find evidence of associations between diffusion parameters and amyloid-related biomarkers, whether assessed by PET or cerebrospinal fluid measures. In contrast, the four diffusion parameters showed strong and widespread associations with biomarkers of neuroinflammation and neurodegeneration, particularly in frontoparietal and cingulate regions. Additionally, in grey matter regions where volume increases were related to amyloid levels, volumes were negatively correlated with the slow diffusion coefficient (P = 0.001), perfusion fraction (P = 0.036) and mean diffusivity (P = 0.047). These findings indicate that diffusion-derived measures are more sensitive to neuroinflammatory and neurodegenerative processes than to amyloid pathology in cognitively unimpaired individuals. Furthermore, the observed negative association between grey matter volume and slow diffusion coefficient in amyloid-related regions may reflect increased cellular complexity rather than intracellular water accumulation. This interpretation suggests that glial remodelling or microstructural changes could underlie brain volume increases in amyloid-positive individuals without overt neurodegeneration. These results underscore the value of intravoxel incoherent motion-derived metrics for gaining deeper insights into the pathophysiological mechanisms of Alzheimer's disease, influencing brain volume changes as well as those resulting from the response to anti-amyloid therapies.