Abstract / Summary
INTRODUCTION: Sex and APOE {epsilon}4 influence Alzheimer's disease risk, but whether they are associated with longitudinal structural brain aging and whether these associations depend on amyloid status is unclear. We tested these relationships using brain age gap (BAG), an MRI-derived marker of age-disproportionate brain atrophy. METHODS: We trained a deep-learning brain-age model on 26,512 neurologically healthy UK Biobank participants and adapted it in 3,332 amyloid-negative cognitively normal MCSA and OASIS-3 scans. Without ADNI training, the model was applied to 4,399 T1-weighted MRI visits from 1,337 ADNI participants. Linear mixed-effects and survival models tested associations among sex, APOE {epsilon}4 allele dose, amyloid status, BAG, cognition, atrophy, and amyloid conversion; equivalence tests evaluated null effects. RESULTS: Among amyloid-positive participants, women and men had similar baseline BAG, but BAG increased faster in women. APOE {epsilon}4 was associated with faster BAG increase in a dose-dependent manner. Higher and increasing BAG tracked hippocampal, fusiform, and whole-brain tissue loss in both amyloid groups. In amyloid-negative participants, BAG change was unrelated to four of five cognitive measures, with effects equivalent to zero within {+/-}0.1 SD, whereas within-person BAG-cognition associations were evident across all five measures in amyloid-positive participants. Higher baseline BAG predicted progression from MCI to dementia (HR 1.67) and from cognitively normal to impaired (HR 1.40). DISCUSSION: Sex and APOE {epsilon}4 allele dose were associated with longitudinal brain aging primarily among amyloid-positive participants. BAG tracked structural neurodegeneration and showed stronger cognitive associations among amyloid-positive participants, supporting its interpretation as a stage-dependent marker of Alzheimer-related neurodegeneration. Independent replication is needed.