A genome-wide association study of venous thromboembolism identifies risk variants in chromosomes 1q24.2 and 9q

Abstract

Objectives: To identify venous thromboembolism (VTE) disease-susceptibility genes.

Patients and methods: We performed in silico genome wide association scan (GWAS) analyses using genotype data imputed to approximately 2.5 million single-nucleotide polymorphisms (SNPs) from adults with objectively-diagnosed VTE (n=1503), and controls frequency matched on age and gender (n=1459; discovery population). Single-nucleotide polymorphisms exceeding genome-wide significance were replicated in a separate population (VTE cases, n=1407; controls, n=1418). Genes associated with VTE were re-sequenced.

Results: Seven SNPs exceeded genome-wide significance (P<5×10(-8)): four on chromosome 1q24.2 (F5 rs6025 [factor V Leiden], BLZF1 rs7538157, NME7 rs16861990 and SLC19A2 rs2038024) and three on chromosome 9q34.2 (ABO rs2519093 [ABO intron 1], rs495828, rs8176719 [ABO blood type O allele]). The replication study confirmed a significant association of F5, NME7 and ABO with VTE. However, F5 was the main signal on 1q24.2 as only ABO SNPs remained significantly associated with VTE after adjusting for F5 rs6025. This 1q24.2 region was shown to be inherited as a haplotype block. ABO re-sequencing identified 15 novel single nucleotide variations (SNV) in ABO intron 6 and the ABO 3' UTR that were strongly associated with VTE (P<10(-4)) and belonged to three distinct linkage disequilibrium (LD) blocks; none were in LD with ABO rs8176719 or rs2519093. Our sample size provided 80% power to detect odds ratios (ORs)=2.0 and 1.51 for minor allele frequencies=0.05 and 0.5, respectively (α=1×10(-8); 1% VTE prevalence).

Conclusions: Apart from F5 rs6025, ABO rs8176719, rs2519093 and F2 rs1799963, additional common and high VTE-risk SNPs among whites are unlikely.

Figure 1

A. Manhattan plot of –log₁₀(P-values) from the case-control association analysis on the merged/imputed VTE data assuming an additive genetic model. The horizontal line at 1.99 × 10^-08 represents the Bonferroni correction. Figure 1B. Location and linkage disequilibrium for SNPs on chromosome 1q24.2 surrounding F5. The black diamond corresponds to F5 6025 (Factor V Leiden) the most significant SNP. Figure 1C. Location and linkage disequilibrium for SNPs on chromosome 9q surrounding ABO. The black diamond corresponds to the most significant SNP, rs495828. For Figures 1B and 1C, the linkage disequilibrium measure, r², between the most significant SNP and any other SNP in the region is symbol coded; the square, star, triangle, circle, and cross symbols correspond to r² > 0.95, > 0.8 and ≤ 0.95, > 0.5 and ≤ 0.8, > 0.2 and ≤ 0.5, and ≤ 0.2, respectively.

Figure 2

Receiver Operator Characteristic (ROC) Curve Analyses of Chromosome 1q24.2 and ABO SNPs

Figure 3

Linkage disequilibrium (LD) between novel ABO SNVs identified by resequencing, and ABO rs8176474 (blood type A allele), rs8176746 (blood type B allele), rs8176719 (blood type O allele) and rs2519093 (ABO intron 1), among 82 VTE cases and 14 controls selected based on ABO rs8176719 and rs2519093 genotype (see Methods). The numbers inside each square and the square shading indicate the degree of LD between SNPs and/or SNVs (i.e., black squares with no number indicate complete LD [r²=1.0]; intensity of the gray shading indicates intermediate values between complete LD [r²=1.0] and no LD [r²=0.0], and white squares with 0 number indicate no LD).

Figure 4

Potential functional roles of ABO SNVs identified on resequencing, by LD block or deletion.