Background:

Current biomarkers enable accurate detection of amyloid-β and tau pathology years before symptoms; however, it remains unclear which individuals will subsequently develop downstream neurodegenerative changes, and how early such changes can be detected. Cortical diffusion measures may capture microstructural tissue alterations prior to overt atrophy, but their relationship to other pathophysiological events in Alzheimer’s disease (AD) is not fully understood. Here, we investigated how longitudinal change in cortical mean diffusivity (MD), a marker of cortical microstructural integrity, relates to AD pathology and biomarkers of glial reactivity.

Methods: We included 299 cognitively unimpaired participants from the ALFA+ study with baseline biomarkers of astrocytic (CSF YKL-40, CSF s100B, plasma GFAP) and microglial reactivity (CSF sTREM2), and longitudinal diffusion MRI (n=77 with two scans; n=222 with three scans; mean follow-up: 6.2 years, range: 2.7-8.8). Cortical MD was estimated using DONSURF and summarised within a previously validated early AD signature meta-ROI. AT₁ status was defined using CSF Aβ42/40 and CSF p-tau181. Linear mixed-effects models tested whether AT₁ stage and glial biomarkers were associated with longitudinal MD change. Mediation analyses assessed whether glial biomarkers mediated associations between AD pathology and MD change.

Results: A+T₁+ participants showed steeper annual MD increases compared to both A+T₁- (β=0.099, SE=0.025, p<0.001) and A-T₁- (β=0.133, SE=0.023, p<0.0001); A+T₁- also differed from A-T₁- (β=0.034, SE=0.014, p<0.05). Higher baseline CSF YKL-40 and plasma GFAP, but not CSF s100B or CSF sTREM2, were associated with faster MD increases (both p<0.0001), independent of A and T₁ status and cortical thickness. Plasma GFAP partially mediated the association of Aβ (24.3%) and p-tau (14.2%) with MD change, while CSF YKL-40 partially mediated the association of p-tau with MD change (14.5%) (all p<0.01).

Conclusions: Different aspects of astroglial reactivity, measured by GFAP and YKL-40, are differentially associated with longitudinal cortical microstructural alterations in preclinical AD. Their partial mediation of the effects of Aβ and tau pathology on increased cortical MD, i.e. less restricted diffusion, suggests a pathway linking core AD pathology and astroglial reactivity to early downstream tissue changes. These findings support exploring astrocyte-related interventions alongside anti-amyloid and anti-tau strategies in prevention trials.