Abstract
Observational studies reached conflicting results regarding the association between gestational diabetes mellitus (GDM) and breast cancer (BC), leaving the causal nature of this relationship unresolved. To clarify this, a Mendelian randomization (MR) study was performed. In brief, genetic instruments for GDM were obtained from the FinnGen project, and data for BC were obtained from previously published consortia. The primary MR analysis methods included the inverse-variance weighted method, with MR-Egger, weighted median, and weighted mode methods serving as validation. Heterogeneity, pleiotropy, and the influence of outliers were evaluated using mandatory sensitivity analyses. Additional sensitivity analyses included Steiger directionality test, radial MR analysis, and multivariable MR analysis. Genetically predicted GDM demonstrated a significant correlation with an increased risk of overall BC (OR = 1.17, 95% CI: 1.03-1.32, p = 0.01) and specifically with ER-negative BC (OR = 1.18, 95% CI: 1.04-1.35, p = 0.01). No such association was observed for ER-positive BC (p = 0.49). However, the causal effect of BC on GDM risk did not emerge in the reverse MR analysis. Comprehensive sensitivity analyses supported these findings. This study provides further evidence that GDM may be a risk factor for BC, particularly the ER-negative subtype, suggesting that GDM may be incorporated into personalized risk assessment models.