作者：曹溢，丰瑞谊，唐燕，罗强，张艺文，汪汉

单位：

610031四川省成都市，西南交通大学附属医院 成都市第三人民医院心内科

单位（英文）：

Department of Cardiology, Affiliated Hospital of Southwest Jiaotong University/the Third People's Hospital of Chengdu,Chengdu 610031, China

关键词：

肺动脉高压；三甲胺N-氧化物；因果关系；孟德尔随机化分析

关键词（英文）：

Pulmonary arterial hypertension; Trimethylamine N-oxide; Causation; Mendelian randomization analysis

中图分类号：

R 541.5

DOI：

10.12114/j.issn.1008-5971.2023.00.215

基金项目：

国家自然科学基金青年科学基金项目（81300243）；四川省科技厅面上项目（19YYJC0580）

摘要：

　目的　采用两样本孟德尔随机化分析探讨三甲胺N-氧化物（TMAO）与肺动脉高压（PAH）的因果关系。方法　TMAO的数据集来自全基因组关联分析（GWAS），包括弗雷明翰心脏研究的2 076名欧洲参与者的人类代谢组的汇总数据；PAH的数据集来自芬兰数据库，包括162 962名欧洲受试者的16 380 163个单核苷酸多态性（SNP）位点信息。筛选工具变量，采用逆方差加权法（IVW）评估TMAO与PAH之间的关系；其次，采用MR-Egger回归、最大似然比（ML）、加权中位数（WM）、MR多效性残差和与异常值（MR-PRESSO）及MR稳健性评分（MR-RAPS）对IVW结果进行补充。使用MR-PRESSO检测异常值，采用留一法敏感性分析探讨单个SNP对IVW分析结果的影响，采用MR-Egger法的截距表示是否存在水平多效性，采用Cochran's Q检验分析工具变量的异质性，采用漏斗图分析研究结果的潜在偏倚情况。结果　本研究共筛选出9个与TMAO密切相关的工具变量。IVW分析结果显示，TMAO可能会增加PAH发生风险〔OR=1.777，95%CI（1.002，3.149），P=0.048〕；上述关系在ML〔OR=1.806，95%CI（1.004，3.251），P=0.049〕、MR-PRESSO〔OR=2.063，95%CI（1.165，3.652），P=0.037〕及MR-RAPS〔OR=2.063，95%CI（2.054，2.259），P=0.015〕中被证实，但在MR-Egger回归〔OR=1.394，95%CI（0.302，6.422），P=0.685〕、WM〔OR=1.919，95%CI（0.859，4.288），P=0.112〕中未被证实。留一法敏感性分析结果显示，逐一删除单个SNP后IVW分析结果无明显变化。MR-Egger法的截距分析结果显示，本研究不存在任何水平多效性（P=0.369）。Cochran'sQ检验结果显示，工具变量不存在异质性（P=0.586）。漏斗图分析结果显示，所有工具变量呈非对称性分布，表明该研究结果可能存在潜在偏倚。结论　TMAO增加可能导致PAH，TMAO与PAH可能存在因果关系。

英文摘要：

Objective To explore the causal relationship between trimethylamine N-oxide (TMAO) and pulmonaryarterial hypertension (PAH) by two-sample Mendelian randomization analysis. Methods The TMAO dataset was obtained from agenome-wide association study (GWAS) , which included summary data of human metabolome from 2, 076 European participantsin the Framingham Heart Study. The PAH dataset was obtained from the Finnish database, which included 16, 380, 163single nucleotide polymorphism (SNP) loci information from 162, 962 European participants. The instrumental variables werescreened, and the relationship between TMAO and PAH was evaluated by inverse variance weighting (IVW) . Secondly, the resultsof IVW analysis were supplemented by MR-Egger regression, maximum likelihood ratio (ML) , weighted median (WM) , MRpleiotropy residual sum and outlier (MR-PRESSO) and MR-robust adjusted profile score (MR-RAPS) . MR-PRESSO was used todetect abnormal values, and the effect of single SNP on IVW analysis results was analyzed by leave-one-out sensitivity analysis.The intercept of MR-Egger method was used to indicate whether there was horizontal pleiotropy. Cochran's Q test was used toanalyze the heterogeneity of instrumental variables. Funnel plot was used to analyze the potential bias of the research results.Results A total of 9 instrumental variables closely related to TMAO were screened out in this study. The IVW analysis resultsshowed that TMAO may increase the risk of PAH [OR=1.777, 95%CI (1.002, 3.149) , P=0.048] . This relationship was confirmedin ML [OR=1.806, 95%CI (1.004, 3.251) , P=0.049] , MR-PRESSO [OR=2.063, 95%CI (1.165, 3.652) , P=0.037] and MR-RAPS [OR=2.063, 95%CI (2.054, 2.259) , P=0.015] , but not in MR-Egger regression [OR=1.394, 95%CI (0.302, 6.422) , P=0.685] andWM [OR=1.919, 95%CI (0.859, 4.288) , P=0.112] . The results of leave-one-out sensitivity analysis showed that there was nosignificant change in IVW analysis results after deleting a single SNP one by one. The results of the intercept analysis of the MREgger method showed that there was no level of pleiotropy in this study (P=0.369) . Cochran's Q test results showed that there wasno heterogeneity in the instrumental variables (P=0.586) . The results of funnel plot analysis showed that all instrumental variableswere asymmetrically distributed, indicating that there may be potential bias in the results of this study. Conclusion TMAO mayincrease the risk of PAH and there may be a causal relationship between TMAO and PAH.

参考文献：