Sex‐specific genetic predictors of memory, executive function, and language performance

Background Alzheimer’s disease (AD) is more prevalent in women than men, and robust evidence shows sex differences in the biological response to the AD neuropathological cascade. However, there is a lack of large‐scale genetic studies on sex‐specific genetic predictors of AD‐related cognitive outcomes. Thus, we sought to elucidate the sex‐specific genetic etiology of memory, executive function, and language performance. Method This study included six cohorts of cognitive aging (Nmales=7,267, Nfemales=9,518). We applied psychometric approaches to build harmonized memory, executive function, and language composite scores. Next, for all domains, we calculated slopes from the cognitive scores (two or more timepoints) with linear mixed effects models. Then we performed sex‐stratified and sex‐interaction GWAS on these phenotypes, covarying for baseline age and the first three genetic principal components. We meta‐analyzed across cohorts with a fixed‐effects model. Sensitivity analyses for all models restricted the sample to cognitively unimpaired individuals. Result In addition to well‐established associations with cognition at the APOE locus, we identified three genetic loci that showed sex‐specific effects with cognition. A chromosome 16 locus (rs114106271), a splicing‐quantitative trait locus for RP11‐152O14.4 and LINC02180 in the testis (GTEx), associated with baseline memory performance in men (β=0.13, P=2.40×10‐8; PInteraction=8.96×10‐6; Figures 1‐2) but not in women (β=‐0.01, P=0.76). A chromosome 14 locus (rs34074573), an expression‐quantitative trait locus (GTEx) for HOMEZ (a homeobox gene), and for BCL2L2 (a previously reported AD risk gene), associated with longitudinal memory performance in men (β=‐0.01, P=4.15×10‐8; PInteraction=5.83×10‐7; Figures 3‐4) but not in women (β=0.001, P=0.09). Finally, a chromosome 6 locus (rs9382966) associated with longitudinal language performance in men with near genome‐wide significance (β=‐0.004, P=6.29×10‐8; PInteraction=2.01×10‐4) but not in women (β=‐0.0003, P=0.61). Conclusion Our results highlight some key sex differences in the genetic architecture of cognitive outcomes. Findings further suggest that some sex‐specific genetic predictors have domain‐specific associations, providing an exciting opportunity to better understand the molecular basis of memory, executive function, and language through genomic analysis. Although our findings need to be replicated, our GWAS analyses highlight the contribution of sex‐s