用Network Meta分析系统评价GLP-1受体激动剂类降糖药的心血管安全性

目的 使用Network Meta分析系统评价胰高血糖素样肽1（GLP-1）受体激动剂类降糖药的心血管安全性.方法 系统检索Medline、Embase、ClinicalTrials.gov和Cochrane Library数据库（截止2011年10月）中比较GLP-1受体激动剂与其他降糖药物或安慰剂的心血管安全性的随机对照研究（RCT）,采用传统Meta分析和Network Meta分析方法对纳入的RCT的研究结果进行合并.结果 共纳入45项研究,15 883例糖尿病患者,包括八种干预措施（六种GLP-1类药：艾塞那肽、利拉鲁肽、他司鲁肽、阿必鲁肽、利西拉来和LY2189265,以及安慰剂和传统降糖药）,研究总臂数为95.传统Meta分析和Network Meta分析结果相近,均未显示GLP-1受体激动剂与其他降糖药物或安慰剂之间心血管疾病安全性有统计学差异（P均＞0.05）.此外,结合直接和间接比较的Network Meta分析显示六种GLP-1类药之间两两比较的心血管安全性也均无统计学差异（P均＞0.05）.基于贝叶斯理论的Network Meta分析可对八种干预措施进行排序,显示安慰剂心血管风险最大.结论 尽管单个研究报道GLP-1类药有潜在的心血管保护效应,但目前Network Meta分析仍无法定论,仍有待专门设计的大型前瞻性研究加以验证.

Cardiovascular safety of antidiabetic agents including GLP-1 receptor agonist reviewed systematically by Network Meta-analysis

Objective To review systematically the cardiovascular safety of antidiabetic agents including glucagon-like peptide-1 (GLP-1) receptor agonist by applying Network Meta-analysis. Methods The randomized controlled trials (RCT) about the comparison of cardiovascular safety of GLP-1 receptor agonist and other antidiabetic agents or placebo were retrieved systematically from Medline, Embase, ClinicalTrials.gov and Cochrane Library (deadline to Oct. 2011). The outcomes of included RCT were combined by using traditional Meta-analysis and Network Meta-analysis. Results There were totally 45 RCT included involving 15 883 diabetic patients, totally 95 arms and 8 kinds of intervention methods (6 kinds of GLP-1-like antidiabetic agents: exenatide, liraglutide, taspoglutide, albiglutide, lixisenatide, LY2189265, and placebo and conventional antidiabetic agents). The results of traditional Meta-analysis were similar to those of Network Meta-analysis, which did not show statistical difference in cardiovascular safety between GLP-1 receptor agonist and other antidiabetic agents or placebo (all P>0.05). Additionally, Network Meta-analysis combing direct and indirect comparison showed that there was no statistical difference in cardiovascular safety among 6 kinds of GLP-1-like antidiabetic agents in a pairwise comparison (all P>0.05). Network Meta-analysis based on Bayesian theory could sort 8 kinds of intervention methods and showed that placebo had the highest cardiovascular risk. Conclusion Although previous single study has reported GLP-1-like antidiabetic agents had potential protective effect on heart and vessels, current Network Meta-analysis cannot give a certain conclusion. It is needed a large perspective study designed specially for confirmation.