Basic Science and Pathogenesis

Two-thirds of Alzheimer's disease (AD) patients are women, and sex-specific APOE-related cognitive impairment is well-documented. However, the interplay among APOE genotype, sex, and common genetic variants (SNPs) remains poorly understood, particularly for specific cognitive domains. We performed a GWAS meta-analysis to investigate the SNP×APOE-ε4×sex interaction on memory, leveraging harmonized data from multiple cohorts. Future analyses will address executive function and language.

We aggregated data from six cohorts (ACT, MAP, NACC, NIAADFBS, ROS, and WRAP), with plans to incorporate additional cohorts in the final analysis. All participants had genome-wide SNP data and harmonized memory scores derived via confirmatory factor analysis. Our current sample includes 33,440 European-ancestry individuals, with final analyses expanding to African and other ancestries. APOE status was determined by genotyping rs7412 and rs429358. Our meta-analysis focused on the three-way interaction (SNP×APOE-ε4×sex) for memory. In addition, we ran four main-effects GWAS meta-analyses stratified by sex and APOE-ε4 status (female carriers, female non-carriers, male carriers, male non-carriers), adjusting for age and principal components. Two-way interaction analyses (SNP×APOE-ε4, SNP×sex) are planned for future analyses.

Preliminary analyses identified 112 SNPs with suggestive associations (p <5×10^-6) for the three-way interaction on memory. Notably, one locus is intronic to HGSNAT, a gene involved in lysosomal function and implicated in cognition. Because APOE-ε4 is also linked to lysosomal dysfunction and neuroinflammation, these findings may suggest a novel connection between HGSNAT and APOE-ε4 in cognitive decline. Stratified GWAS revealed four genome-wide significant main-effect signals in male APOE-ε4 carriers for memory. The most significant signal was rs76307224 (β_{male-carriers}=-2.2, P_{male-carriers}=2.18×10^-10), which approached nominal significance in male non-carriers (β_{male-noncarriers}=-0.01, P_{male-noncarriers}=0.07) but was insignificant in females regardless of APOE genotype (p's >0.5).

By modeling these interactions, our study aims to uncover novel biological pathways underlying cognitive decline. Our preliminary findings highlight the potential for identifying SNP associations with cognitive performance that are modulated by sex and APOE status. Expanding the dataset with additional cohorts, diverse ancestries, and other cognitive domains will enhance statistical power and generalizability. Future work will assess whether these interaction effects persist longitudinally and vary by ancestry, offering deeper insights into AD risk.