磁共振胰胆管成像与超声内镜诊断胆道梗阻疾病的系统评价

MRCP versus EUS for diagnosis of bile duct obstruction： a systematic review

目的评价磁共振胰胆管成像（MRCP）与超声内镜（EUS）对胆道梗阻疾病的诊断准确率。方法检索多家数据库，并手工检索相关会议论文集和中文杂志。2位研究者各自对符合纳入标准的资料进行提取和质量评价，交叉核对，如有分歧，通过讨论或由第3位研究人员协助解决。根据QUADAS质量评估系统评价研究的质量。计算合并统计值敏感度、特异度、阳性似然比、阴性似然比及其95％置信区间，绘制概括性受试者特征性工作曲线（SROC）并计算二者的曲线下面积，通过t检验比较二者的敏感度、特异度差异有无统计学意义。结果共纳入13篇文献，总计1200例患者。MRCP与EUS对胆总管结石所致梗阻诊断的合并敏感度分别为0．870与0．935，合并特异度为0．952与0．947，合并阳性似然比为14．055与16．653，合并阴性似然比为0．177与0．076，SROC曲线下面积为0．9693与0．9771；MRCP与EUS对胆道恶性梗阻诊断的合并敏感度分别为0．805与0．959，合并特异度为0．927与0．975，合并阳性似然比为13．448与23．398，合并阴性似然比为0．134与0．059，SROC曲线下面积为0．9686与0．9870。MRCP与EUS对胆总管结石所致梗阻及胆道恶性梗阻的诊断敏感度、特异度差异均无统计学意义（P〉0．05）。结论MRCP与EUS对胆总管结石及胆道恶性梗阻所致的胆道梗阻疾病均有较高的诊断价值。

Objective To compare the overall diagnostic accuracy of MRCP with EUS for the detection of choledocholithiasis and malignant obstruction in patients with suspected biliary obstruction. Methods A fully recursive literature search was conducted in The Cochrane Central Register of Controlled Trials CENTRAL, MEDLINE or PUBME （1980 -2012）, EMBASE （1980 -2012）, OVID Database （ 1980 - 2012）, CBM （ 1980 - 2012）, VIP database （ 1989 - 2012）, Chinese journal of full-text database （CNKI） （ 1980 - 2012 ）, and WANFANG database （ 1980 - 2012 ）. The prospective diagnostic studies which evaluated or compared the diagnostic accuracy of MRCP and EUS were included combined with manual searches. We also searched the references of all included articles of important meetings and journals. QUA DAS items were used to evaluate the quality of the included studies. SEN, SPE, ± LR, - LR and the areas under SROC were detected, then t-test was used to evaluate whether statistically significant difference existed between EUS and MRCP. Results Thirteen studies including 1200 cases were recruited. The overall pooled sensitivities of MRCP and EUS for the detection of choledocholithiasis were 0. 870 （95% CI：0. 826-0. 906） and 0. 935 （ 95% CI： 0. 90-0. 96 ） respectively, whereas their specificities were 0. 952 （ 95% CI： 0. 926- 0. 971 ） and 0. 947 （95% CI： 0. 920-0. 967 ）, respectively. The overall pooled positive likelihood ratio for MRCP and EUS were 14. 055 （95% CI：6. 259-31. 561） and 16. 653 （95% CI：6. 896-40. 212）, respective- ly, with the corresponding negative likelihood ratio of 0. 177 （ 95 % C1：0. 108-0. 290） and 0. 076 （ 95 % CI：0. 049-0. 118） , respectively. Areas under the ROC curve were 0. 9693 and 0. 9771, respectively. There were no statistically significant difference for sensitivity （0.84 ± 0. 16 vs 0. 93 ± 0. 07, P = 0. 108 ） and specificity （0. 93 s-0. 10 vs 0. 90 ±0. 15, P =0. 555） between MRCP and EUS. The overall pooled sensi- tivities of EUS and MRCP for the detection of malignancy were 0. 959 （ 95% CI： 0. 908-0. 987 ）, and 0. 805 （95% CI： 0. 724-0. 871 ） , respectively, whereas their specificities were 0. 975 （ 95% CI： 0. 954- 0. 988） and 0. 927 （95% CI：0. 897-0. 951 ） ,respectively. The overall pooled positive likelihood ratio for EUS and MRCP were 23. 398 （95% CI： 12. 987-42. 155 ） and 13. 448 （95% CI：4. 961-36. 456） , respectively, with the corresponding negative likelihood ratio of 0. 059 （ 95% CI： 0. 028-0. 122 ） and 0. 134 （ 95% CI：0. 046-0. 391 ） , respectively. Areas under the ROC curve were 0. 9870 and 0. 9686, respective- ly. There were no statistically significant differences for sensitivity （ 0. 95± 0. 55 vs 0. 88 ± 0. 14, P = 0. 22） and specificity （0. 96 ±0. 03 vs 0. 91 ±0. 11, P =0. 31 ） between MRCP and EUS. Conclusion MRCP and EUS are both valuable in differential diagnosis of biliary obstruction caused by choledocholithiasis or malignant lesions.