Mapping sex differences in brain and cognition in relation to APOE4 and amyloid burden: A longitudinal normative modelling study

Sex differences in Alzheimer’s disease (AD) risk and progression are increasingly recognized, with females exhibiting higher global prevalence rates. Yet it remains unclear how genetic and biomarker indicators of Alzheimer’s risk, such as the apolipoprotein E-ɛ4 ( APOE4 ) allele and amyloid burden, relate to sex differences in brain and cognitive health during the preclinical stage. Using established normative models trained on ~58,000 healthy participants, we computed regional z-scores from T1-weighted MRI scans in 372 cognitively normal participants from the Insight 46 cohort. Scans were acquired at two timepoints, approximately three years apart, beginning at age 70. Regions with z-scores below −1.96 were classified as brain-structure outliers and summarized as total outlier count (tOC). We used linear mixed-effects models to examine how sex, age, and AD risk ( APOE4 status and amyloid burden) relate to tOC and cognitive outcomes measured by Preclinical Alzheimer Cognitive Composite (PACC) scores. Both cross-sectional associations and longitudinal changes in tOC and PACC scores were examined, and we tested whether the effects of APOE4 status and amyloid burden on brain and cognitive measures differed by sex. Cross-sectional analyses showed that males had greater tOC than females at younger ages. At timepoint 1, spatial maps showed more regions with outliers in males, though high outlier proportions were limited to occipital areas. By timepoint 2, group differences became more spatially distinct, with males and females showing deviations in different regions. Longitudinally, older males exhibited steeper increases in tOC over time compared to females. Females showed higher PACC scores overall, while no sex differences were observed in cognitive change over time. Greater tOC and amyloid burden were both associated with poorer cognitive outcomes, with the strongest association observed in female APOE4 carriers. However, we found no evidence that AD risk influenced age-related changes in tOC or cognition over time. These findings highlight the complex interplay between sex, age, and AD risk in shaping brain structure and cognition in later life. Some of the observed patterns may reflect emerging vulnerability not yet captured by short-term longitudinal change, underscoring the importance of continued observation. In conclusion, normative modelling provides a valuable approach for detecting subtle variation in brain and cognitive aging across risk groups.