肿瘤坏死因子-α对活动性结核病诊断价值的Meta分析

Diagnostic accuracy of tumor necrosis factor-αin active tuberculosis:a Meta-analysis

目的通过Meta分析方法系统评价肿瘤坏死因子-α(TNF-α)对活动性结核病的临床诊断价值。方法选取时间截止至2021年12月,以TNF-α作为标志物诊断活动性结核病的临床研究文献,计算机检索Pubmed、Embase、Cochrane Library、万方数据知识服务平台。分析纳入的独立研究文献的基本特征、质量评价、合并效应量,并对TNF-α诊断活动性结核病敏感度进行亚组结果、发表偏倚分析。结果通过检索筛选共纳入11项独立研究。5项独立研究来自结核高发地区,7项研究纳入未成年人,2项研究包含人类免疫缺陷病毒(HIV)阳性患者。选择的文献分组及诊断标准明确。患者经细菌学培养、涂片镜检或基因检测确诊或经临床资料综合诊断。经诊断准确性研究质量评价(QUADAS)工具评分大多数研究的质量良好。对纳入的11项独立研究进行分析,研究对象共675例。异质性检验结果较大(I2=72.38%),故采用随机效应模型。TNF-α诊断活动性结核病的敏感度为0.86[95%CI=(0.75,0.92)],特异度为0.87[95%CI=(0.83,0.91)]。合并阳性似然比为6.71[95%CI=(4.70,9.60)],合并阴性似然比为0.17[95%CI=(0.09,0.31)],合并诊断优势比为40.47[95%CI=(16.47,99.45)]。根据提取的研究数据绘制TNF-α对活动性结核病诊断价值的合并受试者工作特征曲线(SROC),曲线下面积最大时,敏感度为0.86,特异度为0.87,最大曲线下面积为0.91[95%CI=(0.88,0.93)]。当以TNF-α检测方法分组时,酶联免疫斑点试验(Elispot)检测TNF-α用于诊断活动性结核病的敏感度最高(0.89),其次为多因子检测试验(Luminex)(0.85)和荧光免疫斑点试验(Fluorospot)(0.85),酶联免疫吸附试验(ELISA)敏感度最低(0.75)。结核病高发地区的研究TNF-α检测敏感度(0.85)明显高于低发地区(0.76)。<18岁敏感度(0.86)稍大于≥18岁(0.84),而在1项<18岁与≥18岁均纳入的研究中敏感度仅为0.58。Deeks检验显示,TNF-α对活动性结核病诊断价值研究的Meta分析均无偏倚证据(P=0.538),发表偏倚的风险较低。结论TNF-α对活动性结核病具有一定的临床诊断价值,并且在高发病率地区及<18岁的人群中、采用TNF-αElispot法检测时的敏感度更高。

Objective To systematically evaluate the clinical value of tumor necrosis factor-α(TNF-α)in the diagnosis of active tuberculosis.Methods The literature of clinical studies using TNF-αas a marker for the diagnosis of active tuberculosis was selected up to December 2021,and literature search was performed in Pubmed,Embase,Cochrane Library and Wanfang(in Chinese)databases.The literature of clinical studies using TNF-αas a marker for the diagnosis of active tuberculosis was selected up to December 2021,and computer searches of Pubmed,Embase,Cochrane Library,and Wanfang Data Knowledge Service platform were performed.Results A total of 11 independent studies were included through search and screening,of which 5 studies were from high incidence areas,7 studied included minors,and 2 studies included human immunodeficiency virus(HIV)positive patients.The selection of literature grouping and diagnostic criteria were clear.The patient was confirmed by bacteriological culture,smear microscopy or genetic test,or combined with clinical data.Most studies were rated good quality by the Quality Assessment of Diagnostic Accuracy Studies(QUADAS).11 independent studies were included and a total of 675 subjects were analyzed.Heterogeneity test results were large(I2=72.38%),so the random effects model was used.The sensitivity and specificity of TNF-αin the diagnosis of active tuberculosis were 0.86[95%CI=(0.75,0.92)]and 0.87[95%CI=(0.83,0.91)].The combined positive likelihood ratio was 6.71[95%CI=(4.70,9.60)],the combined negative likelihood ratio was 0.17[95%CI=(0.09,0.31)],and the combined diagnostic advantage ratio was 40.47[95%CI=(16.47,99.45)].The combined subject receiver operating characteristic(SROC)curve for the diagnostic value of TNF-αfor active tuberculosis was plotted based on the extracted study data,with a sensitivity of 0.86,specificity of 0.87,and maximum area under the curve of 0.91[95%CI=(0.88,0.93)]at the maximum area under the curve.When grouped by TNF-αassay,Elispot assay for TNF-αwas the most sensitive(0.89)for the diagnosis of active tuberculosis,followed by Luminex(0.85)and Fluorospot(0.85),and the enzyme-linked immunosorbent assay(ELISA)method was the least sensitive(0.75).The sensitivity of TNF-αassay was significantly higher in studies in areas with high tuberculosis prevalence(0.85)than in areas with low prevalence(0.76).Sensitivity was slightly greater at<18 years(0.86)than at≥18 years(0.84),whereas sensitivity was only 0.58 in one study that included both<18 years and≥18 years.Deeks test showed that Meta-analysis of studies on the diagnostic value of TNF-αin active TB were all unbiased(P=0.538),indicating a low risk of publication bias.Conclusion TNF-αplays a certain value in the diagnosis of active tuberculosis,and is more sensitive in high incidence areas and in people<18 years old when measured by TNF-αElispot.