γ干扰素释放试验诊断儿童潜伏结核菌感染的Meta分析

目的以结核菌素皮肤试验(TST)为参考试验,评价QuantiFERON-TB Gold test(QFT)和T-SPOT.TB两种IFN-γ释放试验诊断儿童潜伏结核菌感染的准确性。方法计算机检索EMBASE、PubMed、Cochrane图书馆、中文科技期刊全文数据库、中国期刊全文数据库和万方数字化期刊群等数据库,检索时间均为建库至2010年3月。全面检索IFN-γ释放试验诊断儿童潜伏结核菌感染的文献,按照诊断试验的纳入标准筛选文献,提取纳入文献的特征信息(研究背景、设计信息和诊断参数信息)。根据QUADAS质量评价标准评价纳入文献的质量。采用Meta-Disc1.4软件进行Meta分析,检验异质性,并根据异质性结果选择相应的效应模型。对纳入文献予以加权定量合并,计算汇总敏感度、特异度、阳性似然比、阴性似然比和诊断优势比及其95%CI,绘制汇总受试者工作特征(SROC)曲线,并计算曲线下面积(AUC)。结果共检出相关文献285篇,按照文献纳入标准,最终纳入13篇文献(英文文献12篇和中文文献1篇)。10篇文献报道了QFT试验对儿童潜伏结核菌感染的诊断价值,汇总敏感度为0.40(95%CI:0.37~0.44)、汇总特异度为0.87(95%CI:0.85~0.88)、汇总阳性似然比为6.21(95%CI:3.07~12.54)、汇总阴性似然比为0.46(95%CI:0.31~0.68)、汇总诊断优势比为15.58(95%CI:7.47~32.48),SROC AUC为0.8931。4篇文献报道了T-SPOT.TB试验对儿童潜伏结核菌感染的诊断价值,汇总敏感度为0.74(95%CI:0.68~0.79)、汇总特异度为0.84(95%CI:0.81~0.87)、汇总阳性似然比为4.66(95%CI:1.27~17.12)、汇总阴性似然比为0.42(95%CI:0.18~0.99)、汇总诊断优势比为13.71(95%CI:3.71~50.72),SROCAUC为0.8306。结论 IFN-γ释放试验适用于进一步鉴别诊断儿童是否处于潜伏结核菌感染,不适合儿童潜伏结核菌感染的筛查。

Interferon-gamma release assays for diagnosis of latent Mycobacterium tuberculosis infection in children: a meta-analysis

Objective Regarded tuberculin skin test (TST) as the reference standard, to assess the diagnostic value of two interferon-gamma（IFN-γ） release assays ［QuantiFERON-TB Gold test(QFT) and T-SPOT.TB］ for latent Mycobacterium tuberculosis infection in children. Methods We searched EMBASE（1974-2010.3）, PubMed（1966-2010.3）, Cochrane Library（from establishment to 2010.3）,CNKI（1994-2010.3）, CSJD(1989-2010.3）and Wanfang（2000-2010.3）to find all diagnostic tests about IFN-γ release assays for latent Mycobacterium tuberculosis infection in children. After collecting studies according to inclusion criteria of diagnostic tests, data （study background, design information and diagnostic parameters） were extracted.QUADAS items were used to evaluate the qualities of the included studies. Meta-disc software was used to handle data of included studies and to examine heterogeneity. The effect-model was selected according to outcomes of heterogeneity. All included studies were combined with weighted quantity, sensitivity, specificity, positive likelihood ratio(PLR), negative likelihood ratio(NLR), diagnostic odds ratio(DOR) and their 95％CI were calculated, and SROC curve was drawn. Results We got potentially relevant 285 studies.According to eligibility criteria, 13 studies were included. In 10 studies QFT was compared with TST, the outcomes were as follows: pooled sensitivity 0.40（95%CI：0.37-0.44）, pooled specificity 0.87（95%CI：0.85-0.88）, PLR 6.21（95%CI：3.07-12.54）, NLR 0.46（95%CI：0.31-0.68）,DOR 15.58（95%CI：7.47-32.48）and SROC AUC 0.893 1. Comparing T-SPOT.TB with TST, the outcomes of 4 studies were as follows: pooled sensitivity 0.74（95%CI：0.68-0.79）, pooled specificity 0.84（95%CI：0.81-0.87）, PLR 4.66（95%CI：1.27-17.12）, NLR 0.42（95%CI：0.18-0.99）,DOR 13.71（95%CI：3.71-50.72）and SROC AUC 0.830 6. Conclusions There was not sufficient evidence to support the IFN-γ release assays in the diagnosis of latent Mycobacterium tuberculosis infection in children, and IFN-γ release assays could not be regarded as an effective and feasible method for screening clinically. The diagnostic values of IFN-γ release assays needs to be studied further in future because of the limitation of included studies.