2.3 Statistical analysis
We used the “TwoSampleMR” and “MRPRESSO”, “mr.raps” packages of the R software (version 4.2.2) to perform MR analysis. Multiple MR approaches were utilized to explore causal relationships between exposure and outcome. Causal effect estimates from the IVW was considered the criterion for establishing a causal relationship between exposure and outcome(33). It is suggested that if the IVs meet the three major assumption of MR and the selected SNPs are all valid IVs, result from IVW model provides an efficient and reliable estimate for the causal effect(34). As an extension to IVW, MR-PRESSO approach attempts to remove outliers based on their contributions to heterogeneity, while other MR Methods were regarded as a supplement and reference(35). After MR analysis, we assessed the quality of the harmonization and whether the variants were difficult to harmonize by performing Cochran’s Q test and MR Egger regression analysis, as well as leave-one-out method. The IVW model was applied to the heterogeneity analysis, where Cochran’s Q statistic provided evidence for heterogeneity and invalid instruments, with p value greater than 0.05 indicating the absence of heterogeneity(34). The intercept of MR-Egger regression was used to explore and account for the impact of horizontal pleiotropy, and p value greater than 0.05 is evidence for credible instrumental variables(36). To further assess the reliability of the MR results, we evaluated the impact of one single SNP for estimates by performing the leave-one-out method, in which new MR results was recalculated after leaving out each SNP in turn. Finally, we visualized the main results of the MR analysis using scatter plots, forest plots, and funnel plots. Figure 1B shows a detailed flowchart of this MR study.
Results