Shared and Distinct Genomics of Chronic Thromboembolic Pulmonary Hypertension and Pulmonary Embolism

Abstract

Rationale: Chronic thromboembolic pulmonary hypertension involves the formation and nonresolution of thrombus, dysregulated inflammation, angiogenesis, and the development of a small-vessel vasculopathy. Objectives: We aimed to establish the genetic basis of chronic thromboembolic pulmonary hypertension to gain insight into its pathophysiological contributors. Methods: We conducted a genome-wide association study on 1,907 European cases and 10,363 European control subjects. We coanalyzed our results with existing results from genome-wide association studies on deep vein thrombosis, pulmonary embolism, and idiopathic pulmonary arterial hypertension. Measurements and Main Results: Our primary association study revealed genetic associations at the ABO, FGG, F11, MYH7B, and HLA-DRA loci. Through our coanalysis, we demonstrate further associations with chronic thromboembolic pulmonary hypertension at the F2, TSPAN15, SLC44A2, and F5 loci but find no statistically significant associations shared with idiopathic pulmonary arterial hypertension. Conclusions: Chronic thromboembolic pulmonary hypertension is a partially heritable polygenic disease, with related though distinct genetic associations with pulmonary embolism and deep vein thrombosis.

Keywords: genome-wide association study; pulmonary arterial hypertension; venous thromboembolism.

Figure 1.

Flowchart of the study design. Principal component analysis (PCA): excluded due to inferred ancestry based on PCA. Rel/dup: excluded due to being closely related to another sample, or a duplicate of another sample. Heterozygosity: excluded due to abnormal heterozygosity rate. Missingness: excluded due to high missingness in genotype or otherwise unusable genotype. Full details are provided in the Methods section of the online supplement. In short, we recruited cases and control subjects from a variety of centers around the United Kingdom and Europe, including existing control subjects. Our discovery cohort consisted of U.K. and U.S. samples; our replication cohort consisted of European samples. Exclusions were applied sequentially: For instance, some samples that would have been excluded for abnormal heterozygosity rate in the replication cohort were already excluded by PCA. EGPA = eosinophilic granulomatosis with polyangiitis; GWAS = genome-wide association study; HWE = Hardy-Weinberg equilibrium; IPAH = idiopathic pulmonary arterial hypertension; MAF = minor allele frequency; VTE = venous thromboembolism.

Figure 2.

Manhattan plot of −log₁₀(P) values derived from meta-analysis of discovery and replication cohorts. Points higher up correspond to variants more strongly associated with chronic thromboembolic pulmonary hypertension (CTEPH). Variants reaching genome-wide significance (P_CTEPH < 5 × 10⁻⁸) are marked in black, and variants discovered using coanalysis with pulmonary embolism are marked in blue, both labeled with the likely associated gene. The black horizontal line denotes genome-wide significance. Values of −log₁₀(P) larger than 16 are truncated to 16.

Figure 3.

Back-to-back Manhattan plots for chronic thromboembolic pulmonary hypertension (CTEPH) and pulmonary embolism (PE). The distance from the middle line corresponds to −log₁₀(P) values; points farther from the middle line correspond to variants more strongly associated with CTEPH (upward) or PE (downward). Values of −log₁₀(P) larger than 16 are truncated to 16. Peak variants as in Table 1 are marked with the likely corresponding gene. There is substantial sharing between associations with CTEPH and with PE. Genome-wide associations (P < 5 × 10⁻⁸) are marked in red. Additional associations discovered through leverage (conditional false discovery rate) are marked in blue.

Figure 4.

(A and B) Z-scores for chronic thromboembolic pulmonary hypertension (CTEPH) against Z-scores for deep vein thrombosis (DVT) (left) and pulmonary embolism (PE) (right). Each point corresponds to an SNP, with color and shape corresponding to chromosome as per the legend. Z-score pairs close to the origin are excluded. Points higher up correspond to variants more associated with DVT/PE, and points farther to the right correspond to variants more associated with CTEPH. Potential genes (F11, F5, HLA-DRA, etc.) are labeled for some SNPs. The area to the right of the dotted black line is a rejection region based on a CTEPH genome-wide significance threshold of P_CTEPH < 5 × 10⁻⁸. The area to the right of the solid black line is a rejection region based on the leveraged analysis using conditional false discovery rates, equivalent to a V value <5 × 10⁻⁸. The solid red line shows the expected position of Z-score pairs if SNP effect sizes for CTEPH and DVT/PE were identical. If effect sizes were identical for all SNPs, the probability of any of the points corresponding to the ∼200 SNPs reaching genome-wide significance for CTEPH or DVT/PE falling outside the dashed red lines is <0.05. We see that peak SNPs for F5 and HLA-DRA fall outside the dashed lines in both plots.