ABSTRACT

We conducted a Mendelian randomization study to assess whether alcohol and coffee consumption and smoking are causally associated with risk of developing migraine. Independent single-nucleotide polymorphisms associated with the potential risk factors at P < 5 × 10-8 in large-scale genome-wide association studies were selected as instrumental variables. Summary-level data for the associations of the selected single-nucleotide polymorphisms with migraine were obtained from the FinnGen consortium comprising 6687 cases and 144,780 noncases and the UK Biobank study comprising 1072 cases and 360,122 noncases. Estimates derived from the FinnGen and UK Biobank cohorts were combined using fixed-effects meta-analysis. We found evidence for associations of genetically predicted alcohol consumption (odds ratio [OR] 0.54 per SD increase in log-transformed alcoholic drinks per week, 95% confidence interval [CI], 0.35-0.82; P = 0.004), coffee consumption (OR 0.56 per 50% increase in coffee consumption, 95% CI, 0.45-0.70; P < 0.001), and smoking initiation (OR 1.15 for one SD increase in the prevalence of smoking initiation, 95% CI, 1.01-1.31; P = 0.038). These associations persisted in sensitivity analyses, including mutual adjustment in multivariable Mendelian randomization analyses. In reverse Mendelian randomization analyses, genetic liability to migraine was inversely associated with alcohol consumption but was not associated with coffee consumption or smoking initiation. This study provides genetic evidence in support of a protective role of moderate coffee consumption and a detrimental role of cigarette smoking in the etiology of migraine. The inverse association between alcohol consumption and migraine risk may be attributable to reverse causality.