Imputation Accuracy of Apolipoprotein E ε Alleles in Genome-wide arrays and real-time SNP Genotyping assays




Subasinghe, Kumudu
Garlotte, Isabelle
O'Bryant, Sid E.
Barber, Robert C.
Phillips, Nicole


0000-0001-9429-1993 (Subasinghe, Kumudu)

Journal Title

Journal ISSN

Volume Title



Purpose: The vast majority of the established genetic-based risk for late-onset Alzheimer's disease (AD) is attributable to variation within the apolipoprotein E (APOE) gene. This gene, which encodes a protein implicated in various aspects of AD pathology, is characterized by two single nucleotide polymorphisms (SNPs; rs429358 and rs7412) that result in three distinct isoforms (ε4, ε3 and ε2). Most population-based genome-wide association studies to date have identified the APOE ε4 and ε2 alleles as the strongest genetic-based risk and protective factors for AD, respectively. APOE genotype is not only critical for determining disease risk and diagnosis, but also for developing individualized therapeutic strategies. Genotyping via real-time quantitative PCR (qPCR) is the gold standard for APOE isoform determination; however, if genome wide SNP data is available, imputation of APOE (i.e., probabilistic genotyping through inference) may eliminate the need for qPCR genotyping. In this project, we evaluate the concordance of APOE genotypes obtained via qPCR and a genome-wide SNP chip in non-Hispanic White and Mexican American individuals from the Health & Aging Brain among Latino Elders (HABLE) cohort. Method: DNA was extracted from buffy coat samples (n = 1650) on the Hamilton robotic system with the Mag-Bind Blood & Tissue DNA HDQ 96 Kit. qPCR was then performed using the TaqMan Genotyping Kit as per manufacturer's protocol. Results produced via qPCR were then compared to those imputed for rs429358 and directly typed for rs7412 on the Illumina Infinium Global Screening Arrays (GSA) and analyzed with Genome Studio 2.0. Samples with call rates less than 98% were repeated or excluded. Results: Concordance between the APOE genotypes obtained from qPCR and Infinium GSA was 99.32%. Discordance was likely due to poor sample quality and low-frequency imputation errors of rs429358, which may be corrected with more conservative thresholding of the imputed genotype confidence statistics. Conclusion: Genotype imputation from SNPs commonly typed in the APOE region is an effective method for APOE isoform determination, even in Mexican Americans who are more genetically heterogenous due to ancestral admixture; this method may be effectively implemented in large population-based studies of aging and AD.