Objective: To systematically review the diagnostic accuracy of Xpert® Mycobacterium tuberculosis/rifampicin (Xpert® MTB/RIF) for the detection of active tuberculosis (TB) and rifampicin-resistance TB in Chinese patients.
Methods: Four Chinese databases (SinoMed, CNKI, WanFang database, and VIP) and three English databases (PubMed, Embase, and The Cochrane Library) were searched from January 1, 2000 to September 15, 2017, to identify diagnostic tests about the accuracy of Xpert® MTB/RIF in Chinese patients. Two investigators screened the articles and extracted the information independently, and then the quality of each included study was evaluated by Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2. Bivariate random-effects meta-analysis was conducted to pool the sensitivity and specificity. In addition, subgroup analyses were performed based on patient type (TB patient and TB suspected patient), sample type (sputum, bronchoalveolar lavage fluid and others). All statistical analyses were conducted with Stata version 13.0.
Results: A total of 47 articles were included in this systematic review. Most of them (38 articles) were in Chinese and only 9 articles were in English. All the articles were published during 2014 to 2017, and the sample size ranged from 31 to 3 151. Forty articles including 42 comparisons about TB were finally included with the pooled sensitivity of 0.94 (95%CI: 0.92, 0.95) and the pooled specificity of 0.87 (95%CI: 0.84, 0.91). Subgroup analysis showed that different patient and specimen types had no significant differences on sensitivity, but the specificity of sputum group was higher than that of bronchoalveolar lavage fluid. As for the detection of rifampicin-resistant TB, 33 articles (38 comparisons) were analyzed, the pooled sensitivity and specificity were 0.92 (95%CI: 0.89, 0.94) and 0.98 (95%CI: 0.97, 0.99) respectively. There were no significant differences between the patient and specimen in the subgroup analyses. The Deeks funnel plot showed a possible publication bias for detecting active tuberculosis (P=0.08) and no publication bias for rifampicin-resistant TB (P=0.24). The likelihood ratio scatter gram showed that in clinical applications, Xpert® MTB/RIF had a good diagnostic ability for detecting active tuberculosis, and it had good clinical diagnostic value in detecting rifampicin-resistant TB.
Conclusion: Xpert® MTB/RIF has good sensitivity and specificity in detecting TB and rifampicin-resistant TB in Chinese people. In particular, it has good clinical value in diagnosing rifampicin-resistance TB.
目的: 系统评价结核分枝杆菌/利福平耐药实时荧光定量核酸扩增检测方法(Xpert® Mycobacterium tuberculosis/rifampicin，Xpert® MTB/RIF)对中国人群肺结核(活动性肺结核和利福平耐药肺结核)的诊断准确性。
方法: 计算机检索SinoMed、CNKI、万方、维普、PubMed、Embase、The Cochrane Library数据库，搜集于中国人群中应用Xpert® MTB/RIF方法诊断的研究，检索时间限定为2000年1月1日—2017年9月15日。由2名研究者独立筛选文献、提取资料，使用诊断试验准确性研究的偏倚评估工具(Quality Assessment of Diagnostic Accuracy Studies，QUADAS)-2对纳入的研究进行质量评价。采用Stata 13.0软件进行meta分析，使用双变量随机效应模型合并灵敏度与特异度等诊断准确性指标，根据患者类型和样本类型进行亚组分析。
结果: 最终共纳入47篇文献，其中英文9篇，中文38篇，发表时间为2014—2017年，样本量范围为31~3 151例，活动性肺结核共纳入40篇文献(42个比较组)，合并灵敏度为0.94(95%CI: 0.92，0.95)，合并特异度为0.87(95%CI: 0.84，0.91)。亚组分析显示，对于不同的样本类型，灵敏度无明显差异，但痰液样本的特异度高于支气管灌洗液样本。利福平耐药肺结核共纳入33篇文献(38个比较组)，合并灵敏度为0.92(95%CI: 0.89，0.94)，特异度为0.98(95%CI: 0.97，0.99)。Deeks法漏斗图显示，检测活动性肺结核方面可能存在发表偏倚(P=0.08)，利福平耐药肺结核方面均不存在发表偏倚(P=0.24)。
结论: 使用Xpert® MTB/RIF方法检测活动性肺结核和利福平耐药肺结核具有较好的灵敏度和特异度，在检测利福平耐药方面具有较好的临床诊断价值。
Keywords: Meta-analysis; Multidrug-resistant tuberculosis; Pulmonary tuberculosis; Xpert® MTB/RIF.