Aims: Observational studies show metformin use associated with lower cancer risk, although experimental evidence is inconsistent. To provide genetic validation for repositioning of metformin in cancer prevention, we assessed genetically proxied effects of putative metformin targets on cancer outcomes using a drug-target Mendelian randomization (MR) design.
Materials and methods: We identified genetic proxies of 11 metformin targets (PRKAA1, PRKAA2, PRKAB1, PRKAB2, PRKAG1, PRKAG2, PRKAG3, ETFDH, GPD1, SLC47A1 and ACACB) based on their associations with tissue-specific gene expression, overall/sex-specific HbA1c and type 2 diabetes. We then evaluated genetically proxied effects of these targets on five major cancers using MR. We also employed a conventional MR design to assess the relationship of HbA1c with cancer using the inverse variance method, with sensitivity analyses. Associations were corrected for multiple comparisons using false discovery rates.
Results: We identified two genetic proxies of putative metformin targets (PRKAG1 and GPD1) as valid instrumental variables (F statistics > 10). PRKAG1 was associated with a reduced risk of colorectal cancer (OR: 0.74 per mmol/mol reduction in overall HbA1c, 95% CI: 0.63-0.87; p = 0.001), with consistent findings in sex-specific analysis. This effect was unlikely mediated by HbA1c reduction, as indicated by conventional MR analyses (OR: 1.01 per mmol/mol, 95% CI: 0.99-1.02). No significant association was observed for GPD1 (OR: 1.00, 95% CI: 0.74-1.36; p = 0.98).
Conclusions: Metformin may prevent colorectal cancer via the AMPKγ1 (PRKAG1) target based on genetic evidence, supporting the evaluation of metformin use in colorectal cancer prevention using randomised controlled trials.
Keywords: Mendelian randomization; cancer; genome‐wide association study (GWAS); metformin; reposition.
© 2026 The Author(s). Diabetes, Obesity and Metabolism published by John Wiley & Sons Ltd.