Neurobiological correlates of longitudinal grey matter volume changes in preclinical Alzheimer’s disease

Abstract

Structural brain changes during the earliest asymptomatic stages of Alzheimer’s disease (AD) remain poorly understood. Previous research in preclinical AD shows heterogeneous findings, reporting both subtle neuronal loss and paradoxical increases in grey matter (GM) volume. This study applies an extensive cerebrospinal fluid (CSF) biomarker panel to better understand the biological processes underlying longitudinal GM changes in cognitively unimpaired (CU) adults, spanning the amyloid/tau (AT) continuum. We analysed data from 627 CU individuals from three longitudinal cohorts (ALFA+, Wisconsin ADRC, WRAP), with repeated MRI (3.5 ± 0.9 years) and baseline CSF biomarkers from the NeuroToolKit panel (Roche Diagnostics). Using non-negative matrix factorization, we decomposed the CSF biomarker levels into six latent components, reflecting amyloid-β (Aβ) pathology, tau-related pathophysiology with synaptic injury, neuroaxonal injury, microglial reactivity, astrocytic reactivity, and cytokine signalling. We tested associations between component weights and voxel-wise longitudinal GM volume changes using single-component and a joint-all components model. Analyses were performed across the full sample and stratified by AT status. Associations with longitudinal cognitive performance (PACC) were assessed using linear mixed-effects models. The Aβ pathology component was the strongest and most widespread predictor of longitudinal GM atrophy, predominantly in temporal and frontal regions, also when controlling for tau pathophysiology, neuroaxonal injury, or neuroinflammatory components. Higher Aβ pathology scores were also associated with cognitive decline. The component capturing tau-related pathophysiology and synaptic injury initially associated with GM loss but lost significance after accounting for other biomarker components. In contrast, components reflecting microglial reactivity, astrocytic reactivity, and cytokine signalling were associated with longitudinal GM volume increases, with effects varying by AT stage. In this large longitudinal sample of asymptomatic individuals, the Aβ-dominant biomarker component showed the strongest association with longitudinal GM atrophy and cognitive decline, beyond the effects of tau pathophysiology and neuroaxonal injury. While glial and inflammatory processes may contribute to transient GM increases in preclinical AD. A better understanding of these dynamic relationships between structural brain changes and various biological pathways at the earliest stages of AD is crucial to inform the development of interventions before irreversible neurodegeneration occurs.