Non-compacted, PET-insensitive amyloid states increase after systemic inflammation and predict neuritic damage across Aβ pathology models and Alzheimer patients.

Neuroinflammation is a key modulator of Alzheimer's disease (AD) risk, yet the impact of non-genetic inflammatory risk factors - such as systemic inflammation - remains poorly defined. Building on our previous work, here we show that 9 months after systemic lipopolysaccharide (LPS) challenge in APP23 mice, microglia-plaque interaction is disturbed and shifts Aβ aggregates toward a less compacted state, as revealed by conformation-sensitive amyloid dyes. Importantly, these structural changes are associated with increased plaque-associated neuritic dystrophy, phenocopying the effects of microglial risk genes. Generalising these findings, we show that across aging in APP23 and APPPS1 mice, and in AD patient tissue, non-compacted amyloid and microgliosis - but not compacted amyloid - are consistent predictors of neuritic damage. Notably, both in mouse and human tissue, ex vivo amyloid-PET signal largely reflects compacted but not non-compacted amyloid load. Our findings suggest that genetic and environmental risk factors converge on shared mechanisms of impaired microglial-plaque interaction and amyloid restructuring, and that commonly used amyloid-PET measures insufficiently capture amyloid states that define the severity of neuritic damage, with important implications for clinical trials in AD.