Comprehensive Mendelian Randomization Analysis of Smoking and Its Effects on Venous Thromboembolism

An increasing number of Mendelian randomization (MR) studies have evaluated the causal link between smoking and venous thromboembolism (VTE). However, previous studies often rely on single genetic variants related to smoking quantity and exhibit various other shortcomings, making them prone to pleiotropy and potentially leading to imprecise causal estimates. Thus, the deeper causal mechanisms remain largely unexplored. This MR study reassessed the causal relationship between smoking and VTE, including its subtypes-deep vein thrombosis (DVT) and pulmonary embolism (PE). Data on VTE were sourced from the FinnGen consortium with nonoverlapping sample sizes. The smoking phenotypes analyzed included smoking initiation, lifetime smoking, the number of cigarettes smoked per day by both current and former smokers (CigDay), and total pack-years of smoking in adulthood. The primary analytical method was inverse-variance-weighted (IVW), supplemented by multiple verification methods to ensure robust results. Statistical rigor was ensured through LDtrait pruning and Steiger filtering for reverse causation, with comprehensive sensitivity analyses including RadialMR confirming the findings' robustness. After Bonferroni correction, this study demonstrates significant causal evidence linking lifetime smoking with the incidence of VTE (odds ratio [OR]_IVW = 1.50, 95% confidence interval [CI] 1.21-1.85, p = 1.75 × 10^-4) and PE (OR_IVW = 1.69, 95% CI 1.25-2.28, p = 6.55 × 10^-4), and suggestive evidence with DVT, consistent in direction with previous studies but showing considerable differences in effect sizes and significance. Additionally, CigDay (past and current) increases the risks of VTE and DVT, while no causal link was found between smoking initiation and VTE or its subtypes (p < 0.05), both directly contradicting previous conclusions. Furthermore, our study is the first to suggest a causal link between pack-years and an increased risk of VTE. This MR study employed rigorous statistical pruning of its instrumental variables, using the most comprehensive smoking phenotype to date. It successfully mitigated biases such as winner's curse, yielding causal effect results distinct from previous studies.