The prevalence of cannabis use is rising in many regions because of legalization and decriminalization. Observational epidemiologic studies on cannabis and cancer risk have produced inconsistent and sometimes conflicting results. To investigate potential causal relationships while mitigating confounding and reverse causation, the authors performed a two-sample Mendelian randomization (MR) analysis to test whether genetic liability to cannabis use influences site-specific cancer risk in individuals of European ancestry.
The investigators identified 22 independent genetic variants strongly associated with cannabis use (threshold p < 5 × 10⁻⁸) from a large meta-analysis of genome-wide association studies (GWAS) of people of European ancestry. These variants served as instrumental variables representing lifelong genetic liability to cannabis use. No additional instrument details or per-variant information were reported in the abstract.
Genome-wide association summary-level data for site-specific cancers were obtained from FinnGen, a Finnish population resource comprising individuals of European ancestry. The abstract does not report per-cancer case counts or overall sample sizes; users are referred to the full manuscript or FinnGen resources for detailed sample-size and phenotype definitions. The authors note that the original GWAS sources obtained informed consent and institutional approvals; this MR analysis used publicly available summary statistics and required no new ethical approvals.
Primary MR analyses were conducted using the inverse-variance weighted (IVW) approach. The authors also implemented a multivariable IVW method adjusting for cigarette smoking, given smoking is a known confounder for several cancers. Sensitivity analyses included the simple median, weighted median, MR-Egger, and MR pleiotropy residual sum and outlier (MR-PRESSO) methods to assess robustness and horizontal pleiotropy. According to the abstract, these sensitivity analyses found no evidence of horizontal pleiotropy between genetic liability to cannabis use and the site-specific cancers considered.
In multivariable IVW analyses that adjusted for cigarette smoking, genetic liability to cannabis use was associated with increased odds of esophageal cancer (odds ratio [OR] = 1.74, 95% confidence interval [CI] = 1.29–2.15; p = 0.013) and lung cancer (OR = 1.35, 95% CI = 1.13–1.58; p = 0.009) in individuals of European ancestry in FinnGen. These estimates represent the MR-derived causal effect of genetically predicted cannabis use on these site-specific cancer outcomes after accounting for smoking in multivariable models.
Unexpectedly, the multivariable IVW analysis reported a protective association between genetic liability to cannabis use and pancreatic cancer (OR = 0.77, 95% CI = 0.57–0.91; p = 0.032) in the FinnGen sample. The abstract does not provide mechanistic explanations for this finding; readers should consult the full text for any exploration of biological plausibility or further sensitivity testing specific to pancreatic cancer.
The authors conclude that genetic liability to cannabis use was associated with higher risk of esophageal and lung cancers and a lower risk of pancreatic cancer in individuals of European ancestry, based on FinnGen summary data. Sensitivity analyses reported no evidence of horizontal pleiotropy. The analysis used publicly available GWAS summary statistics; the authors indicate that supporting GWAS data are available via FinnGen (https://www.finngen.fi/en).
This report is a medRxiv preprint and has not been peer reviewed; the authors explicitly state it should not guide clinical practice. The abstract provides effect estimates and p-values but does not report sample sizes, detailed instrument strength metrics, or a dedicated limitations section in the abstract. For full methodological details, per-variant instrument data, sample sizes, and the authors' discussion of limitations and implications, consult the full preprint.