Cannabis use is increasing annually but the relationship between cannabis use and cancer incidence is not uniform because of confounding factors. We aimed to assess the effect of cannabis use on cancer risk using a two-sample Mendelian randomization (MR) approach.
Secondary data analyses were performed on pooled data based on Genome-Wide Association Study (GWAS), selecting data from the ICC and UK-Biobank and 23andMeInc lifetime cannabis use and cannabis use disorder related to the substance use disorders working group from the Psychiatric Genomics Consortium, then selecting highly correlated SNPs as instrumental variables. The substance use disorders working group, iPSYCH, and deCODE GWAS data, and then highly correlated SNPs were selected as instrumental variables for two-sample Mendelian randomization analyses using inverse variance weighting, MR-Egger regression, and weighted median, respectively, to evaluate the relationship between lifetime cannabis use and nine tumors, and subsequently analyzed these results in the same way using cannabis use disorders.
The risk of all cancers except breast cancer was not associated with lifetime cannabis use. Our inverse variance weighting method found that lifetime marijuana use reduced the breast cancer risk (P = 0.016, odds ratio [OR] = 0.981), and we subsequently conducted analyses of cannabis use disorders and cancer risk, which showed that cannabis use disorders elevated the risk of breast cancer (P = 0.007, OR = 1.007) as well as the risk of lung cancer (P = 0.014, OR = 1.122).
Large MR analyses suggest that lifetime cannabis use may reduce breast cancer risk, but cannabis use disorder exacerbates the risk of breast and lung cancer. The mechanisms responsible for this outcome remain to be investigated.