Interpretation of 10 years of Alzheimer’s disease genetic findings in the perspective of statistical heterogeneity

Common genetic variants and susceptibility loci associated with Alzheimer's disease (AD) have been discovered through large-scale genome-wide association studies (GWAS), GWAS by proxy (GWAX) and meta-analysis of GWAS and GWAX (GWAS+GWAX). However, due to the very low repeatability of AD susceptibility loci and the low heritability of AD, these AD genetic findings have been questioned. We summarize AD genetic findings from the past 10 years and provide a new interpretation of these findings in the context of statistical heterogeneity. We discovered that only 17% of AD risk loci demonstrated reproducibility with a genome-wide significance of P < 5.00E-08 across all AD GWAS and GWAS+GWAX datasets. We highlighted that the AD GWAS+GWAX with the largest sample size failed to identify the most significant signals, the maximum number of genome-wide significant genetic variants or maximum heritability. Additionally, we identified widespread statistical heterogeneity in AD GWAS+GWAX datasets, but not in AD GWAS datasets. We consider that statistical heterogeneity may have attenuated the statistical power in AD GWAS+GWAX and may contribute to explaining the low repeatability (17%) of genome-wide significant AD susceptibility loci and the decreased AD heritability (40-2%) as the sample size increased. Importantly, evidence supports the idea that a decrease in statistical heterogeneity facilitates the identification of genome-wide significant genetic loci and contributes to an increase in AD heritability. Collectively, current AD GWAX and GWAS+GWAX findings should be meticulously assessed and warrant additional investigation, and AD GWAS+GWAX should employ multiple meta-analysis methods, such as random-effects inverse variance-weighted meta-analysis, which is designed specifically for statistical heterogeneity.