MR-Egger回归在孟德尔随机化分析中的应用

 目的 探讨MR-Egger回归在孟德尔随机化分析中的应用。方法 利用全基因组关联研究结果确定工具变量，通过MR-Egger回归进行两样本孟德尔随机化分析来检验高密度脂蛋白胆固醇（high density liptein cholesterol，HDL-C）和冠状动脉疾病之间的因果效应，评估潜在工具变量的多效性，并将MR-Egger回归结果与逆方差加权法（inverse-variance weighted，IVW）结果进行比较。结果 共纳入120个单核苷酸多态性构建工具变量。IVW结果表明HDL-C与冠状动脉疾病之间存在显著的因果关联（OR=0.82，95% CI：0.75~0.89）。MR-Egger回归结果显示HDL-C与冠状动脉疾病之间不存在因果关联（OR=0.96，95% CI：0.83~1.11），并提示基因工具变量存在显著的多效性（截距：-0.01，P=0.008）。结论 MR-Egger回归在孟德尔随机化分析中有一定的应用价值，尤其是在多效性偏倚存在的情况下能给出准确的因果效应估计。

Application of MR-Egger regression in Mendelian randomization analysis

Objective To explore the application of MR-Egger regression in Mendelian randomization analysis. Methods Instrumental variables were determined according to the results of genome-wide association study. A two-sample Mendelian randomization analysis was conducted by MR-Egger regression to examine the causal effect between high-density lipoprotein cholesterol (HDL-C) and coronary artery disease. The pleiotropy of potential instrumental variables was evaluated, and the results of MR-Egger regression and inverse-variance weighted (IVW) were further compared. Results A total of 120 single nucleotide polymorphisms were included to construct instrumental variables. The IVW results suggested a significant causal relationship between HDL-C and CAD (OR=0.82, 95%CI: 0.75-0.89), while the MR-Egger regression showed no causal association (OR=0.96, 95%CI: 0.83-1.11), with significant pleiotropy in instrumental variables (the intercept: -0.01, P=0.008). Conclusion MR-Egger regression is valuable in Mendelian randomization analysis, and can give an accurate estimate of the causal effect, especially in the case of pleiotropy bias existing.