Identification of DEEP Vein Thrombosis GENETIC RISK Variants by NEXT GENERATION Sequencing of Hemostatic Genes

Abstract 710

Next-generation DNA sequencing is opening new avenues for genetic association studies in complex diseases that, like deep vein thrombosis (DVT), have a strong genetic basis only partially accounted for by currently identified risk variants. Using next-generation DNA sequencing, we resequenced the entire protein-coding area and intron-exon boundaries of ≈200 hemostatic/pro-inflammatory genes (including all coagulation factor and anticoagulant protein genes) in 26 Italian individuals with idiopathic DVT of the lower limbs and 28 age-, gender- and ethnicity-matched healthy controls.

In order to enrich for genetic component, DVT patients were selected from a cohort of 730 cases of idiopathic DVT referred to the Angelo Bianchi Bonomi Hemophilia and Thrombosis Center (Milan, Italy) on the basis of the following criteria: (a) negative for anti-phospholipid autoantibodies, anticoagulant protein deficiencies, factor V Leiden and prothrombin G20210A, (b) age of onset <55 years. Genomic libraries with barcode sequence-tags were pooled (in pools of 8 or 16 samples) and enriched for target DNA sequences by capture on Roche NimbleGen HD2 2.1M-probe custom chips. Sequencing was performed on ABI SOLiD 4 platforms. After sequencing, reads with the barcodes were assigned to the corresponding sample and mapped to reference human genome, NCBI36/hg18, using BFAST software.

We produced more than 18 gigabases of raw sequence data to sequence at high depth of coverage (>40X after duplicate read removal) the 700-kilobase target region. A total of 2351 single nucleotide variants (SNVs) and 121 short indels were identified. We developed a dedicated software in order to be able to run association analyses in our dataset using PLINK. A plain association analysis was used to compare the frequencies of common SNVs, whereas a gene-based analysis of restricted non-synonymous mutations was used to reveal potential associations of rare variants.

Of 626 common SNVs tested for association, 32 SNVs from 13 genes were associated with p<0.01. These include 8 missense variants at 6 genes, 2 in genes already reported in association with DVT (FGA, ABO) and 6 in genes not previously reported (LPA, COL4A6, PLCG2 and KNG1). As a positive control we confirmed the association of rs6050 at FGA in ∼1400 Italian individuals (p=1.9 × 10⁻⁵, OR 1.45; 95% CI, 1.22–1.72). Replication of the other variants is ongoing. Gene-based analyses revealed an excess of private missense variants at anticoagulant protein genes (13 vs 3 variants; binomial probability, p=0.005) and at collagen 4 genes (14 vs 6 variants; binomial probability, p=0.02) in DVT cases compared to controls. We are currently expanding our investigations to 100 DVT cases and 100 controls.

Although obtained in a relatively small number of patients and controls, our results illustrate the potential of next-generation sequencing for the discovery of genetic variation predisposing to complex diseases like DVT.