Abstract: Objective To investigate the causal associations between antidiabetic medications(exogenous insulin, metformin, rosiglitazone, and gliclazide)and the risk of malignant tumours in women using Mendelian randomisation. Methods Summary-level data from genome-wide association studies(GWAS)in the IEU OpenGWAS project were analyzed. Exposures included four antidiabetic drugs(insulin, metformin, rosiglitazone, and gliclazide), with female malignant tumours as the outcome. Two-sample Mendelian randomisation analyses were conducted using inverse variance weighting(IVW, primary method), MR-Egger regression, and weighted median approaches. Heterogeneity was assessed via IVW and MR-Egger, pleiotropy was evaluated using MR-Egger intercepts, and leave-one-out sensitivity analysis tested result robustness. Results This study showed a positive causal association between insulin treatment and the risk of ovarian cancer(OR=1.08,95%CI: 1.03-1.14; P=0.004), a negative causal association with the risk of cervical cancer(OR=6.14×10^-11,95%CI: 2.00×10^-19-1.88×10^-2; P=0.018), endometrial cancer(OR=1.86×10^-3,95%CI: 6.47×10^-6-0.533; P=0.029), breast cancer(OR=4.83×10^-4;95%CI: 2.02×10^-6-0.116; P=0.006), and HER(+)breast cancer subtype(OR=2.4×10^-4;95%CI: 1.64×10^-7-0.351; P=0.025), and a negative causal association between metformin treatment and human epidermal growth factor receptor(HER)(+)breast cancer subtypes(OR=2.14×10^-2;95%CI: 1.36×10^-3-0.337; P=0.006). No horizontal pleiotropy was found in any of the above studies. Conclusion Exogenous insulin increases ovarian cancer risk but reduces risks of cervical, endometrial, breast, and HER(+)breast cancers. Metformin lowers HER+ breast cancer risk.

Key words: Female malignant tumours, Breast cancer, Insulin, Metformin, Mendelian randomisation