The causal association between smoking, alcohol consumption and risk of bladder cancer: A univariable and multivariable Mendelian randomization study

Smoking and alcohol consumption are associated with bladder cancer risk in observational studies. We conducted a two-sample univariable and multivariable Mendelian randomization (MR) analysis to determine whether those associations are causal. We used 21, 126, 360, 39 single nucleotide polymorphisms (SNPs) as instrumental variables for number of cigarettes per day, lifetime smoking index, smoking initiation, and drinks per week, respectively. A total of 1115 cases with bladder cancer and 174 006 noncases from FinnGen consortium and 2883 cases with bladder cancer and 417 955 noncases from UK Biobank study were obtained. Genetic predisposition to cigarettes per day, lifetime smoking index and smoking initiation were positively associated with an increased risk of bladder cancer in both the FinnGen and UK Biobank consortium. The summary odds ratio (OR) of bladder cancer was 1.79 (95% confidence interval [CI], 1.31-2.45; P = .0002), 2.38 (95% CI, 1.45-3.88; P = .0005) and 1.91 (95% CI, 1.46-2.50; P = 1.59 x 10(-06)) for one SD increase in the number of cigarettes per day, lifetime smoking index and smoking initiation, respectively. The genetically instrumented number of drinks per week was not associated with bladder cancer (OR = 0.69; 95% CI, 0.44-1.10; P = .1237). Estimates were consistent in multivariable MR analyses by the adjustments of body mass index and education. Our study suggests a causal potential of the association of smoking but not alcohol consumption with bladder cancer according to current evidence.