Genetic liability to bipolar disorder increased cannabis use risk, but not the reverse

Genetic liability to bipolar disorder was significantly associated with increased risk of lifetime cannabis use across all three MR methods (inverse variance weighted, weighted median, and Egger regression). In contrast, genetic liability to lifetime cannabis use showed no association with bipolar disorder risk. Sensitivity analyses found no evidence of confounding through pleiotropic effects.

Two-sample bidirectional Mendelian randomization using genome-wide significant SNPs from large GWAS of bipolar disorder and lifetime cannabis use. Three complementary MR methods applied with sensitivity analyses for pleiotropy.

Previous MR studies examined cannabis and schizophrenia, depression, and ADHD, but not bipolar disorder. This study fills that gap and finds a unidirectional relationship: bipolar vulnerability drives cannabis use, not the other way around. This has implications for understanding self-medication in bipolar disorder.

The finding that bipolar genetics drives cannabis use (but not vice versa) contrasts with the cannabis-schizophrenia relationship, where evidence suggests bidirectional effects. This means the cannabis-bipolar association may primarily reflect self-medication rather than cannabis causing bipolar disorder.

Mendelian randomization assumes genetic instruments only affect the outcome through the exposure (exclusion restriction). Lifetime cannabis use (ever vs. never) is a crude measure. Cannot examine heavy use or CUD specifically. Population-level finding may not apply to individuals.