Rare Germline Loss-of-Function Variants in C1-Inhibitor (SERPING1) Are Associated with Venous Thromboembolism in Population-Scale Sequencing Datasets

Background: Hereditary angioedema (HAE) is a rare condition caused by autosomal dominant mutations in SERPING1 that result in deficiency or dysfunction of its gene product, C1 inhibitor (C1INH). C1INH is the main physiologic inhibitor of several contact pathway constituents, including factors XI, XII, and plasma kallikrein. Although accepted wisdom holds that HAE is not associated with venous thromboembolism (VTE), recent studies have suggested an increased risk of VTE in HAE patients, likely due to impaired regulation of contact pathway-mediated coagulation. Population-scale next-generation sequencing datasets offer an unprecedented opportunity to assess the clinical impacts of genetically-defined SERPING1 haploinsufficiency while minimizing ascertainment bias and maintaining case/control homogeneity. Here, we assessed the impact of deleterious SERPING1 variants on VTE risk in over 600,000 individuals.

Methods: We used two biorepositories with paired genetic and clinical data: the UK Biobank (UKB, N=452,209) and the NIH All of Us program (AoU, N=204,006). Rare (MAF ≤0.001%) germline variants in C1INH (SERPING1) were each assigned an in silico-derived functional impact score (FIS) ranging from 0.0 to 1.0. Variants with higher scores are more likely to result in loss of protein function. An FIS of 1.0 signifies mutations that cause stop-gained (nonsense), essential splice site disruption, or frameshift. We assessed the risk of VTE in carriers of rare damaging SERPING1 variants (defined as FIS ≥0.7) across the UKB and AoU using Firth’s penalized-likelihood Cox proportional hazards modeling, and results were meta-analyzed using a random-effects model. Selected sensitivity analyses were performed using Firth’s logistic regression. Circulating C1INH levels were assessed in a subset of UKB participants (N=41,483) using the Olink® Explore 3072 plasma proteomics panel and reported as linearized normalized protein expression (L-NPX) units.

Results: We identified 28 carriers of SERPING1 variants with an FIS ≥0.7 in the UKB and 27 in AoU, for a total of 55 carriers (crude prevalence: 8.534 x 10^-5 or approximately 1:12,000 ). Positive SERPING1 variant carrier status was significantly associated with increased risk of VTE (OR=4.641, 95% CI: 2.084-10.336, P=0.0002). We next validated in silico predictions of variant effects on protein function using UKB plasma proteomics data. Circulating C1INH levels were significantly lower in carriers of variants with FIS ≥0.4 (P<0.0001) and ≥0.6 (P=0.003) compared to non-carriers. Effect size estimates for VTE also rose steadily with increasing FIS cutoffs, indicating that greater biological impact corresponded with the likelihood that variants cause C1INH loss-of-function. An extended Firth’s model limited to UKB data and adjusting for known VTE risk factors, including sex, age, BMI, blood type, CRP, and a history of smoking, reproduced the expected associations, confirming the accuracy of clinical annotation in the UKB. An additional Firth’s model in the UKB adjusting for polygenic risk of VTE and factor V Leiden and prothrombin gene mutation G20210A carrier status also failed to meaningfully change effect size estimates for SERPING1, indicating that the risk conferred by SERPING1 variant carrier status functions independently of known common genetic risk factors for VTE.

Conclusions: Leveraging a population-scale multiomic dataset, we have identified a significantly increased lifetime risk of VTE among individuals with rare, germline loss-of-function variants in the gene encoding C1INH. These data support the ongoing shift in our understanding of the thrombotic risk associated with C1INH deficiency.