Tracking a decade of structural changes in preclinical and prodromal Alzheimer's disease: insights from amyloid-β pathology.

Alzheimer's disease (AD) pathology is typically associated with reduced brain volume and cortical thickness, interpreted as neurodegeneration. However, several cross-sectional studies in cognitively unimpaired individuals have paradoxically reported larger brain volume and thicker cortex in the presence of amyloid-β (Aβ), suggesting that Aβ-brain structure associations may not follow a simple linear pattern early in the disease. We leveraged over a decade of longitudinal data from the PREVENT-AD cohort (N=367, mean follow-up 7.17 years, range 0-11.27 years) to investigate how Aβ burden related to both cross-sectional levels and longitudinal changes in brain volume, cortical thickness, and cortical tissue mean diffusivity (MD _T ) derived from free-water corrected diffusion tensor imaging. We examined associations separately in individuals below (Aβ-) and above (Aβ+) the Aβ-positivity threshold, tested for nonlinearity across all participants, and aligned structural trajectories to the estimated years from Aβ-positivity onset. We found that higher Aβ was associated with larger brain volume, higher cortical thickness and lower MD _T in regions such as the fusiform gyrus, supramarginal gyrus, hippocampal volume, inferior parietal and middle temporal cortex in the Aβ- group. The opposite associations were found in the Aβ+ group. Across all participants, volume and thickness showed an inverse U-shaped relationship with Aβ, while MD _T followed a U-shaped pattern. Longitudinally, the rate of cortical thickness change showed an inverse U-shaped association with Aβ, whereas Aβ burden was associated with faster volume loss and greater MD _T increase. When structural measures were aligned to the estimated time of Aβ positivity onset, volume and thickness increased years before the expected Aβ positivity onset and declined thereafter, while MD _T showed no association with time relative to Aβ positivity onset. Our results help reconcile inconsistencies across prior studies and suggest that brain structural changes related to Aβ pathology start years before Aβ positivity onset and follow nonlinear trajectories. These early structural changes might be due to pathological processes such as neuroinflammatory swelling early in the course of the disease.