Using genotyping and whole-exome sequencing data to improve genetic risk prediction in deep venous thrombosis

BackgroundDeep Vein Thrombosis (DVT) is a common disease that can lead to serious complications such as pulmonary embolism and in-hospital mortality. More than 60% of DVT risk is influenced by genetic factors, such as Factor V Leiden (FVL) and prothrombin G20210A mutations (PTM). Characterising the genetic contribution and stratifying participants based on their genetic makeup can favourably impact risk prediction. Therefore, we aimed to develop and evaluate a genetic-based prediction model for DVT based on polygenic risk score (PRS) in the UK Biobank cohort. MethodsWe performed a genome-wide association study (GWAS) and constructed a PRS in the 60% (N=284,591) of the UK Biobank cohort. The remaining 40% (N=147,164) was employed to evaluate the PRS and to perform gene-based tests on exome-sequencing data to identify effects by rare variants. ResultsIn the GWAS, we discovered and replicated a novel variant (rs11604583) near TRIM51 gene and in the exome-sequencing data, and we identified a novel rare variant (rs187725533) located near CREB3L1, associated with 2.2-fold higher risk of DVT. In our PRS model, the top decile is associated with 3.4-fold increased risk of DVT, an effect that is 2.3-fold, when excluding FVL carriers. In the top PRS decile, cumulative risk of DVT at age of 80 years is 10% for FVL carriers, contraposed to 5% for FVL non-carriers. ConclusionWe showed that common and rare variants influence DVT risk and that the PRS improves risk prediction on top of FVL. This suggests that individuals classified with high PRS scores could benefit from early genetic screening.