Validate association of gene loci and establish genetic risk prediction models for late-onset Alzheimer's disease in Chinese populations

Background More than 60 independent single-nucleotide polymorphisms (SNPs) have been associated with Alzheimer's disease risk by genome-wide association studies in European. Objective We aimed to confirm these SNPs in Chinese Han populations and investigate the utility of these genetic markers. Methods Altogether 1595 late-onset Alzheimer's disease (LOAD) patients and 2474 controls from Chinese population were recruited. We replicated the association of 68 SNPs with LOAD and established polygenetic risk score (PRS) prediction model using significant SNPs. Meta-analysis for MS4A6A rs610932 and PICALM rs3851179 were performed. Results According to our findings, 14 out of 68 SNPs are validated significantly associated with LOAD (adjusted p < 0.05) after adjusting age and sex in the Chinese population. Besides, after stratification by APOE ε4 status, almost all SNPs retain markedly relationship with LOAD in APOE ε4 noncarriers. However, few loci retain correlation in APOE ε4 carriers. Furthermore, the area under the receiver operating characteristic curve prediction model for distinguishing LOAD patients from normal subjects were 0.614 for PRS and 0.689 for PRS and APOE . In addition, meta-analysis including this study of East Asian populations confirmed that rs610932 and rs3851179 were dramatically related to the LOAD (OR = 0.85, 95% CI = 0.74-0.97; OR = 0.87, 95% CI = 0.83-0.91). Conclusions Despite genetic heterogeneity, there are still common loci among different races. PRS based on AD risk-associated SNPs may supplement APOE for better assessing individual risk for AD in Chinese. Besides, interactions between genes and gene environment affect the impact of risk allele on diverse populations.