Abstract / Summary
NTRODUCTION: Alzheimer's disease (AD) shows sex differences in risk, clinical progression, and biomarker profiles, but whether inherited genetic liability to modifiable risk factors contributes differently to AD susceptibility in females and males remains unclear. We evaluated whether genetic relationships between several modifiable dementia risk factors and AD differ between females and males. METHODS: We integrated polygenic risk scores (PRSs) for modifiable risk factors, genome-wide genetic correlation, and two-sample Mendelian randomization (MR) analyses. PRS associations with AD diagnosis were evaluated in the Alzheimer's Disease Genetics Consortium (n=27,751) and with AD plasma biomarkers in the Health and Aging Brain Study-Health Disparities (n=2,307-2,737, depending on the biomarker). Primary analyses evaluated female- and male-specific associations and compared sex-specific estimates. RESULTS: Educational attainment and physical activity PRSs were associated with AD in both sexes. PRS sex differences were nominal for educational attainment, total cholesterol, and LDL cholesterol, but none remained significant after FDR correction. Genetic correlation showed FDR-significant sex differences for social isolation, physical activity, LDL cholesterol, and pulse pressure. MR identified FDR-significant associations within sex strata; higher HDL cholesterol was associated with lower AD risk in females, whereas higher total cholesterol and type 2 diabetes were associated with lower and higher AD risk, respectively, in males. Sensitivity analyses provided inconsistent support, and no sex difference in MR estimates remained significant after FDR correction. In HABS-HD, type 2 diabetes PRS was associated with higher NfL, BMI PRS with higher pTau181, and the composite PRS with higher NfL among males after FDR correction. DISCUSSION: Evidence that modifiable dementia risk factors relate differently to AD by sex was limited and depended on the genetic measure examined. Sex differences in genome-wide genetic overlap were not consistently reflected in PRS or MR, indicating that shared genetic architecture does not necessarily translate into sex-specific inherited liability or causal effects