Considering the APOE locus in Alzheimer’s disease polygenic scores in the Health and Retirement Study: a longitudinal panel study

BMC Medical Genomics

Kelly M. Bakulski¹  Colter M. Mitchell²  Erin B. Ware²  Jessica D. Faul² 

[1] Department of Epidemiology, School of Public Health, University of Michigan, 48109, Ann Arbor, MI, USA;Survey Research Center, Institute for Social Research, University of Michigan, 426 Thompson St., Rm. 3320 ISR-Thompson, 48104, Ann Arbor, MI, USA;

关键词: Polygenic score;    Alzheimer’s disease;    Dementia;    Apolipoprotein E;    P-value;    Thresholding;   

DOI  :  10.1186/s12920-020-00815-9

来源: Springer

【 摘 要 】

BackgroundPolygenic scores are a strategy to aggregate the small, additive effects of single nucleotide polymorphisms across the genome. With phenotypes like Alzheimer’s disease, which have a strong and well-established genomic locus (APOE), the cumulative effect of genetic variants outside of this area has not been well established in a population-representative sample.MethodsHere we examine the association between polygenic scores for Alzheimer’s disease both with and without the APOE region (chr19: 45,384,477 to 45,432,606, build 37/hg 19) at different P value thresholds and dementia. We also investigate the addition of APOE-ε4 carrier status and its effect on the polygenic score—dementia association in the Health and Retirement Study using generalized linear models accounting for repeated measures by individual and use a binomial distribution, logit link, and unstructured correlation structure.ResultsIn a large sample of European ancestry participants of the Health and Retirement Study (n = 9872) with an average of 5.2 (standard deviation 1.8) visit spaced two years apart, we found that including the APOE region through weighted variants in a polygenic score was insufficient to capture the large amount of risk attributed to this region. We also found that a polygenic score with a P value threshold of 0.01 had the strongest association with the odds of dementia in this sample (odds ratio = 1.10 95%CI 1.0 to 1.2).ConclusionWe recommend removing the APOE region from polygenic score calculation and treating the APOE locus as an independent covariate when modeling dementia. We also recommend using a moderately conservative P value threshold (e.g. 0.01) when creating polygenic scores for Alzheimer’s disease on dementia. These recommendations may help elucidate relationships between polygenic scores and regions of strong significance for phenotypes similar to Alzheimer’s disease.

【 授权许可】

CC BY