不同类型CT和MR血管成像诊断颅内动脉瘤准确率的Meta分析

目的 分析评价不同类型CT血管成像（CTA）和MR血管成像（MRA）诊断颅内动脉瘤（IAN）的准确率。 方法 检索PubMed、EMbase、Cochralle图书馆和中国知网、中国生物医学文献数据库、维普网、万方数据库，收集2018年1月以前有关不同类型CTA和MRA诊断IAN的中、英文文献，按照诊断性研究的纳入标准筛选文献，并依据诊断准确性试验质量评价工具2标准进行质量评价。采用Stata 12.0及Meta-Disc 1.4软件对纳入的文献进行Meta分析，分别比较CTA、MRA及亚组之间的合并灵敏度（SEN合并）、合并特异度（SPE合并）、诊断比值比（DOR）、合并阳性似然比（PLR合并）、合并阴性似然比（NLR合并）、验前概率和验后概率等性能指标，绘制森林图和综合受试者工作特征（SROC）曲线并计算曲线下面积（AUC）。采用Stata 12.0统计分析软件绘制Deek's漏斗图评估纳入文献的发表偏倚。通过Q检验和I2进行异质性分析。 结果 检索获得1754篇文献，最终纳入38篇临床诊断研究，包括5212例患者。Meta分析结果显示：①CTA诊断IAN的SEN合并、SPE合并、DOR、PLR合并、NLR合并、AUC、验前概率和验后概率分别为0.947（95%CI：0.926，0.963）、0.916（95%CI：0.864，0.949）、195.328（95%CI：97.367，391.847）、11.218（95%CI：6.861，18.341）、0.057（95%CI：0.040，0.082）、0.98（95%CI：0.96，0.99）、20%和74%；②MRA诊断IAN的SEN合并、SPE合并、DOR、PLR合并、NLR合并、AUC、验前概率和验后概率分别为0.935（95%CI：0.899，0.958）、0.956（95%CI：0.917，0.977）、311.421（95%CI：126.935，764.038）、21.285（95%CI：11.114，40.766）、0.068（95%CI：0.044，0.107）、0.98（95%CI：0.97，0.99）、20%和84%；③亚组分析结果显示：不同类型的CTA和MRA诊断IAN之间AUC的差异均无统计学意义。 结论 不同类型CTA与MRA均可以诊断IAN，诊断准确性均较高且基本一致，CTA和MRA均可作为诊断IAN的首选检查手段。

Diagnostic accuracy of different types of computed tomographic angiography and magnetic resonance angiography for intracranial aneurysm: a Meta-analysis

Objective To perform a Meta-analysis to systematically review the accuracy different types of computed tomographic angiography (CTA) and magnetic resonance angiography (MRA) for intracranial aneurysm (IAN). Methods PubMed, EMbase, Cochralle library, Cnki, China biomedical literature database, weipu, wanfang and other databases were searched to collect Chinese and English literatures on IAN diagnosis by different types of CTA and MRA before January 2018．The references were evaluated and screened according to the criteria of diagnostic research. In addition, quality assessment was conducted according to the Quality Assessment of Diagnostic Accuracy Studies 2, RevMan 5.3 and Metadisc 1.4 software was used to respectively compare the summary sensitivity (SEN), summary specificity (SPE), diagnostic odds ratio (DOR), summary positive likelihood ratio (PLR), summary negative likelihood ratio (NLR), pre-test probability, post-test probability between CTA, MRA and subgroup. The forest figure and a summary of receiver operating characteristic curve (SROC) was drawn out and the area under curve (AUC) was calculated. Deek's funnel plot was drawn by Stata 12.0 statistical analysis software to evaluate the publication bias of included literature. Heterogeneity was analyzed by Q test and I2. P<0.05 indicated that the difference was statistically significant. Results A total of 1754 records was retrieved and 38 studies involving 5212 patients were included. The results of meta-analysis showed that: ① The summary SEN, summary SPE, DOR, summary PLR, summary NLR, AUC, pre-test probability, post-test probability of CTA were: 0.947 (95%CI: 0.926, 0.963), 0.916 (95%CI: 0.864, 0.949), 195.328 (95%CI: 97.367, 391.847), 11.218 (95%CI: 6.861, 18.341), 0.057 (95%CI: 0.040, 0.082), 0.98 (95%CI: 0.96, 0.99), 20% and 74%; ② The summary SEN, summary SPE, DOR, summary PLR, summary NLR, AUC, pre-test probability, post-test probability of MRA were: 0.935 (95%CI: 0.899, 0.958), 0.956 (95%CI: 0.917, 0.977), 311.421 (95%CI:126.935, 764.038), 21.285 (95%CI: 11.114, 40.766), 0.068 (95%CI: 0.044, 0.107), 0.98 (95%CI: 0.97, 0.99), 20% and 84%; ③ The results of subgroup analysis showed that: there were no statistical difference between the different types of CTA and MRA in SROC. Conclusions The different types of CTA and MRA can be used to diagnose IAN, and the diagnostic accuracy is relatively high and basically consistent. Both CTA and MRA can be used as the first choice for diagnosing IAN.