体质指数与冠心病的因果关系:孟德尔随机化研究

Causal Relationship between Body Mass Index and Coronary Heart Disease:a Mendelian Randomization Study

目的采用孟德尔随机化(MR)方法分析体质指数(BMI)与冠心病的因果关系。方法本研究中BMI为暴露因素,冠心病为结局。通过IEU OpenGWAS prodect获取样本,纳入样本均来源于欧洲人群,其中BMI数据集样本量为461460例,单核苷酸多态性(SNP)数量为9851867个;冠心病数据集样本量为361194例,SNP数量为13295130个。从BMI数据集中选取与BMI高度相关的SNP、从冠心病数据集中选取与BMI有关的SNP,取其交集作为工具变量。采用MR-Egger回归、加权中位数法(WME)、逆方差加权法(IVW)、简单模型、加权模型共5种回归模型分析BMI与冠心病的因果关系。结果本研究筛选出与BMI高度相关的SNP 458个,从冠心病数据集中筛选与BMI高度相关的SNP 460个,然后删除17个具有中间等位基因的回文SNP和2个重复命名的SNP,最终纳入441个SNP。MREgger回归、WME、IVW结果显示,BMI升高是冠心病的危险因素(P<0.01);而单纯模型和加权模型结果未显示BMI升高是冠心病的危险因素(P>0.05),但其β值与MR-Egger回归、WME、IVW的β值方向一致。Cochran Q检验结果显示,SNP间存在统计学异质性(Q=598.844,P<0.001),故主要关注IVW结果;MR-Egger截距法结果显示,MREgger截距项<0.001(P=0.083),提示该研究不存在水平多效性;留一法分析结果显示,逐一剔除SNP后,剩余SNP仍位于无效线右边,且其OR值及其95%CI未发生明显改变,提示本研究MR结果较稳健。漏斗图分析结果显示,所有SNP分布基本对称,提示SNP作为工具变量时,推断出的因果关系受潜在因素的影响相对较小。结论BMI升高是冠心病的危险因素,故建议通过控制BMI而预防冠心病。

Objective To analyze the causal relationship between body mass index(BMI)and coronary heart disease by Mendelian randomization(MR)methods.Methods In this study,BMI was the exposure factor and coronary heart disease was the outcome.The samples were obtained by IEU OpenGWAS prodect,and the included samples were all from the European population.The sample size of BMI data set was 461460 cases,the number of single nucleotide polymorphism(SNP)was 9851867;the sample size of coronary heart disease data set was 361194 cases,the number of SNP was 13295130.SNP highly related to BMI were selected from the BMI data set,and SNP related to BMI were selected from the coronary heart disease data set,and their intersections were taken as instrumental variables.Five regression models including MR-Egger regression,weighted median method(WME),inverse variance weighting(IVW),simple model and weighted model were used to analyze the causal relationship between BMI and coronary heart disease.Results In this study,458 SNP highly related to BMI and 460 SNP highly correlated with BMI were screened from the coronary heart disease dataset,and then 17 palindromic SNP with intermediate alleles and 2 repetitively named SNP were deleted,and 441 SNP were finally included.The results of MR-Egger regression,WME and IVW showed that elevated BMI was a risk factor for coronary heart disease(P<0.01).While the results of the simple model and the weighted model showed that the elevated BMI was not a risk factor for coronary heart disease(P>0.05),but the results of the simple and weighted models showed that the beta values were in the same direction as those of the MR-Egger regression,WME,and IVW.The results of Cochran Q test showed that there was statistical heterogeneity among SNP(Q=598.844,P<0.001),so the IVW results were mainly concerned.The results of MR-Egger intercept method showed that the MR-Egger intercept term was<0.001(P=0.083),suggesting that there was no horizontal pleiotropy in this study.The results of the leaveone-out analysis showed that after the SNP were removed one by one,the remaining SNP were still on the right side of the invalid line,and their OR value and 95%CI did not change significantly,suggesting that the MR results of this study were more robust.The results of funnel plot analysis showed that the distribution of all SNP was basically symmetrical,suggesting that when SNP was used as an instrumental variable,the causal relationship inferred was less affected by potential factors.Conclusion Elevated BMI is a risk factor for coronary heart disease,therefore,it is recommended to prevent coronary heart disease by controlling BMI.