目的： 评价超声内镜（endoscopic ultrasonography,EUS）对胃肠道间质瘤（gastrointestinal stromal tumor,GIST）的诊断价值。方法： 搜索了Cochrane对照试验中央注册库（Cochrane Central Register of Controlled Trials, CENTRAL）、EMBASE（Excerpt Medica Database）、PubMed等8个数据库,并根据参考诊断试验文献质量评价标准-2（quality assessment diagnostic accuracy studies-2,QUADAS-2）分别评估每项研究的质量。使用Stata 16.0软件构建综合受试者工作特征曲线（summary receiver operating characteristic curve, SROC曲线）并计算曲线下面积（area under curve, AUC）。通过多元回归分析可能产生的异质性影响,并进行亚组分析。同时评估出版偏倚。结果： 共筛选出12篇文章,包括529例病变。一项研究在流量和时间解释偏倚方面风险较高,但所有纳入的研究在参考标准方面均存在不明确的偏倚。荟萃分析显示,合并灵敏度和特异度分别为0.88[95%置信区间（confidence interval,CI）：0.78~0.93]和0.84（95%CI：0.70~0.93）。阳性似然比（positive likelihood ratio,PLR）、阴性似然比（negative likelihood ratio, NLR）、诊断率为5.6（95%CI：2.7~11.4）、0.15（95%CI：0.08~0.27）、38（95%CI：14~103）。SROC曲线的AUC为0.93。在Deeks的“漏斗图”中未发现明显的出版偏倚。结论： EUS对临床医师准确诊断GIST具有一定的指导意义。

Objectives To evaluate the diagnostic value of endoscopic ultrasonography(EUS) in the assessment of gastrointestinal stromal tumors (GIST). Methods Eight databases were searched. The authors separately assessed the quality of each study using the quality assessment diagnostic accuracy studies-2(QUADAS-2) criteria. Stata 16.0 software was used to construct a receiver operating characteristic curve(ROC curve) and the area under curve(AUC) was calculated. The possible heterogeneous effects were explored through meta regression, and a subgroup analysis was performed. The publication bias was assessed. Results A total of 12 articles were screened, including 529 cases of lesions. The quality assessment showed that one study had a high risk in terms of flow and time interpretation bias, but all included studies had unclear bias in reference criteria. The meta-analysis showed the summary sensitivity and specificity were 0.88 [95% confidence interval (CI): 0.78-0.93] and 0.84 (95% CI: 0.70-0.93) respectively. The positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio were 5.6 (95% CI :2.7-11.4),0.15 (95% CI: 0.08-0.27), and 38 (95% CI: 14-103), respectively. The AUC of ROC curve was 0.93. There was no significant publication bias in Deeks’ Funnel plot. Conclusions The EUS shows a certain significance in guiding clinicians to diagnose GIST accurately.