Abstract
Genome-wide association studies (GWAS) allow an assessment of associations between single nucleotide polymorphisms (SNPs) and phenotypes or traits of interest in a non-hypothesis driven manner. Previously, based on limited candidate gene association analysis, we showed that survival in sickle cell anemia and exceptional longevity (EL) in the general population share common genetic modifiers (Blood, 52a, 2007). This preliminary result suggested that aging mechanisms and associated genes might play a role in the variability of sickle cell anemia. Using GWAS, we now report strong evidence supporting this conjecture. We conducted a GWAS using an Illumina platform that permits genotyping up to 1 million haplotype-tagging SNPs spread across the genome, as well as other types of genetic variation, in large populations. We used the Illumina 610K SNP array to discover SNPs that are associated with different degrees of severity of sickle cell anemia in 684 patients. Patients were assigned to either a severe or mild disease category based on an integrated measure of sickle cell anemia severity that was determined by a network model that assigns a score predicting the risk of death (Blood110: 272, 2007). In parallel, we used the Illumina 370K SNP and the Illumina 1M SNP arrays to discover SNPs associated with EL in 877 centenarians enrolled in the New England Centenarian Study and 1,850 younger controls. In both studies, each SNP was tested for association with the traits of severe or less severe sickle cell anemia and EL using Bayesian tests of general, dominant and recessive associations (BMC Genet.9, 2008). We then identified those SNPs satisfying these 3 criteria:
at least one model of association was 10 times more likely than no association in the GWAS of EL; the same model of association was at least 3 times more likely (because of the smaller sample size) than no association in the GWAS of sickle cell anemia severity, the same allele was more frequent in centenarians and in sickle cell anemia patients with milder disease.
This analysis identified 140 SNPs in more than 50 genes and some intergenic regions that showed robust and consistent associations. This number is more than twice the number that would be expected by chance. Among the most ‘significant’ genes with associated SNPs were ARFGEF2, ADAMTS12, DOK5, DPP10, FGF21, KCNQ1, IRF4, MYO3B NAIF1, TNNI3K; more than one SNP was found in ARFGEF2, NAIF1, DPP10, SORCS3, TNNI3K. KCNQ1 has a putative role in blood circulation and regulation of heart contraction. The frequency of the common genotype for SNP rs108961 increases by almost 60% in sickle cell anemia patients with severe disease (27% versus 43%). The same common genotype in random Caucasian controls has frequency 34% that decreases to 29% in centenarians. Mutations in this gene are associated with long and short QT syndrome, with familial atrial fibrillation, heart disease and sudden death. SNPs in 2 of the genes (HAO2, a peroxisome protein involved in fatty acid oxidation, and MAP2K1, a MAP kinase involved in multiple biochemical signals) that were significantly associated with both sickle cell disease severity and EL in our earlier candidate gene studies, were also associated in the GWAS. GWAS also revealed significant association with CDKN2A, a cyclin-dependent kinase that has been associated with Type 2 diabetes, risk of myocardial infarction and triglyceride levels in several GWAS, and with FGF21, the fibroblast growth factor 21 precursor that has been shown to regulate glucose metabolism. CDKN2A has been associated with disease free survival in other studies. Common metabolic pathways are likely to influence the chance of developing complications of Mendelian and multigenic diseases and the likelihood of achieving EL. This might explain the commonality of genes whose SNPs are associated with the vascular complications of sickle cell anemia, arteriosclerosis and diabetes. A new paradigm suggests that hitherto unexpected genetic differences modulate a limited number of pathways that form a common route toward determining good health and disease.
Using genetic information from candidate gene and genome-wide association studies in risk prediction for alcohol dependence