We identified a significant protective effect of LDL-c against T2D (OR = 0.84, PGSMR = 1.1 × 10−4) in the case–control data, which might explain the observation from a previous study that lowering LDL-c using statin therapy is associated with a slightly increased risk of T2D39. The estimate was not significant in the community data (likely due to the lack of power) but in a consistent direction (OR = 0.95, PGSMR = 0.08). Given the strong genetic correlation between the two T2D data sets (rg = 0.98, SE = 0.062) as estimated by the bivariate LDSC analysis30, we meta-analyzed the two data sets using the inverse-variance approach, and performed the GSMR analysis to re-estimate the effect of LDL-c on T2D using the T2D meta-analysis data. The effect size was highly significant (OR = 0.88, PGSMR = 3.0 × 10−7).
