Tuberculosis is one of the most serious challenges to global public health [
On Jul 6, 2018, we searched PubMed, Embase, the Cochrane Library, China National Knowledge Infrastructure (CNKI), and the Wanfang database for studies evaluating the accuracy of Xpert in LNTB detection. The search formula ((Xpert OR Gene Xpert) AND (Tuberculosis, Lymph Node''[Mesh] OR ''Extra pulmonary tuberculosis'')) was used for PubMed without any limitation. Similar search formulae were used for Embase, the Cochrane Library, CNKI, and Wanfang databases. References cited in the included articles and reviews were further explored for possible candidate studies.
We included full-text original studies that assessed the diagnostic accuracy of Xpert assay for LNTB using FNA or biopsy tissue specimens. Reference standards were well-defined and appropriate in the studies. The articles directly provided true positive (TP), false positive (FP), false negative (FN), and true negative (TN) values for the assay or included the data necessary to calculate these measures. Case reports, studies with < 10 samples, conference reports, and abstracts without full articles were excluded.
A composite reference standard (CRS) or mycobacterial culture was defined as the reference standard in our study. Clinical manifestation, biochemical test results, histopathology, smears, other nucleic acid amplification tests (NAATs), culture, and response to antituberculosis treatment constituted the reference standards in CRS.
Two investigators independently assessed the candidate articles by reviewing titles and abstracts, followed by the full text, for inclusion. Discrepancies between the two investigators were resolved by discussion with a third investigator.
We extracted data including author, year, country, TP, FP, FN, and TN values for the assay, reference standard, and specimen type, along with other parameters. The same two investigators independently extracted the necessary information from each of the included articles; we cross-checked the information obtained by them. Discrepancies between the two datasets were settled by discussion with a third investigator, similar to that during the literature selection phase. Data from studies against two different reference standards were treated separately.
According to the two reference standards (CRS and culture), the two investigators independently divided the studies into two groups and used a revised tool for Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) to assess study quality separately [
We first obtained the values corresponding to TP, FP, FN, and TN in each included study and calculated the estimated pooled sensitivity and specificity of Xpert MTB/RIF associated with 95% CI, against CRS or culture, using bivariate random-effects models. Forest plots for sensitivity and specificity were generated for each study. The area under summary receiver operating characteristic (SROC) curves (AUC) was subsequently calculated. I2 statistics was used to assess heterogeneity between the studies and a reference standard. While 0% indicated no observed heterogeneity, values greater than 50% were considered to signify substantial heterogeneity [
Imperfect reference standards may lead to misclassification of samples in diagnostic validity studies [
Three hundred and four candidate articles, identified from relevant databases using our search strategy, three articles identified from other sources, and twenty-seven qualified articles were included according to the inclusion criteria (Figure
Literature retrieval flow chart. 80, 23, 159, 30, and 12 articles were found from PubMed, the Cochrane Library, Embase, Wanfang database, and CNKI respectively.
When an article reported the use of two different standards in the same study, we considered the article to include two independent studies. In accordance with this principle, 36 independent studies were included: 15 compared Xpert MTB/RIF with CRS and 21 compared Xpert MTB/RIF with culture (Table
Characteristics of the included studies. FNA: fine needle aspiration. CRS: composite reference standard. TP: true positive. FP: false positive. FN: false negative. TN: true negative.
Author | Year | County | TP | FP | FN | TN | Specimen type | Reference standard | Decontaminate method | Sample condition | Location | Homogenisation | Sample ratio | Patient selection method |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Causse, M. | 2011 | Spain | 16 | 0 | 1 | 70 | Tissue | Cuture | NALC-NaOH | Fresh | Peripheral | No | 2:1 | consecutive |
| ||||||||||||||
Vadwai, V.a | 2011 | India | 32 | 17 | 12 | 127 | Tissue | Cuture | No | Fresh | Peripheral | Mechanical | 2:1 | consecutive |
| ||||||||||||||
Vadwai, V.b | 2011 | India | 49 | 1 | 17 | 122 | Tissue | CRS | No | Fresh | Peripheral | Mechanical | 2:1 | consecutive |
| ||||||||||||||
Armand, S | 2011 | France | 8 | 0 | 8 | 2 | Tissue | Cuture | NALC-NaOH | Frozen | Peripheral | Mechanical | 3:1 | Convenience |
| ||||||||||||||
Hillemann, D. | 2011 | Germany | 6 | 3 | 4 | 52 | Tissue | Cuture | NALC-NaOH | Fresh | Peripheral | Mechanical | 3:1 | Consecutive |
| ||||||||||||||
Zeka, A. N.a | 2011 | Turkey | 11 | 2 | 3 | 10 | Tissue | Cuture | No | Frozen | Peripheral | No | 3:1 | Consecutive |
| ||||||||||||||
Zeka, A. N.b | 2011 | Turkey | 13 | 0 | 4 | 9 | Tissue | CRS | No | Frozen | Peripheral | No | 3:1 | Consecutive |
| ||||||||||||||
Ligthelm, L. J.a | 2011 | South Africa | 29 | 2 | 1 | 16 | FNA | CRS | No | Fresh | Peripheral | No | 2:1 | Consecutive |
| ||||||||||||||
Ligthelm, L. J.b | 2011 | South Africa | 28 | 3 | 1 | 16 | FNA | Cuture | No | Fresh | Peripheral | No | 2:1 | Consecutive |
| ||||||||||||||
Malbruny, B. | 2011 | France | 6 | 0 | 0 | 17 | FNA | Cuture | NALC-NaOH | Fresh | Peripheral | Mechanical | 3:1 | Convenience |
| ||||||||||||||
Moure, R. | 2012 | Spain | 24 | 0 | 10 | 4 | Tissue | Cuture | NALC-NaOH | Frozen | Peripheral | Mechanical | 2:1 | Convenience |
| ||||||||||||||
Tortoli, E.a | 2012 | Italy | 24 | 4 | 5 | 85 | Tissue | Cuture | NALC-NaOH | Frozen | Peripheral | Mechanical | 2:1 | Convenience |
| ||||||||||||||
Tortoli, E.b | 2012 | Italy | 28 | 0 | 5 | 85 | Tissue | CRS | NALC-NaOH | Frozen | Peripheral | Mechanical | 2:1 | Convenience |
| ||||||||||||||
Van Rie, A.a | 2013 | South Africa | 139 | 23 | 10 | 172 | FNA | Cuture | No | Fresh | Peripheral | No | 2:1 | Consecutive |
| ||||||||||||||
Van Rie, A.b | 2013 | South Africa | 160 | 2 | 42 | 144 | FNA | CRS | No | Fresh | Peripheral | No | 2:1 | Consecutive |
| ||||||||||||||
Ablanedo-Terrazas, Y. | 2014 | Mexico | 15 | 0 | 0 | 53 | FNA | Cuture | NALC-NaOH | Frozen | Peripheral | No | 2:1 | Convenience |
| ||||||||||||||
Biadglegne, F. | 2014 | Ethiopia | 29 | 56 | 3 | 132 | FNA | Cuture | NALC-NaOH | Frozen | Peripheral | No | 3:1 | Convenience |
| ||||||||||||||
Coetzee, L. | 2014 | South Africa | 21 | 13 | 4 | 34 | FNA | CRS | No | Fresh | Peripheral | No | 2:1 | Convenience |
| ||||||||||||||
Dhasmana, D. J. | 2014 | United Kingdom | 24 | 3 | 12 | 77 | FNA | CRS | No | Fresh | Mediastinal | No | Unknow | Convenience |
| ||||||||||||||
Scott, L. E. | 2014 | South Africa | 16 | 14 | 4 | 50 | FNA | Cuture | NALC-NaOH | Fresh | Peripheral | No | 2:1 | Convenience |
| ||||||||||||||
Salvador, F.a | 2015 | Spain | 5 | 6 | 2 | 2 | FNA | Cuture | No | Fresh | Peripheral | No | Unknow | Convenience |
| ||||||||||||||
Salvador, F.b | 2015 | Spain | 6 | 0 | 3 | 4 | Tissue | Cuture | No | Fresh | Peripheral | No | Unknow | Convenience |
| ||||||||||||||
Tadesse, M. | 2015 | Ethiopia | 76 | 7 | 11 | 42 | FNA | CRS | NALC-NaOH | Frozen | Peripheral | Mechanical | 3:1 | Consecutive |
| ||||||||||||||
Bholla, M. | 2016 | Tanzania | 11 | 3 | 8 | 40 | FNA | CRS | No | Fresh | Peripheral | No | 3:1 | Convenience |
| ||||||||||||||
Dhooria, S. | 2016 | India | 26 | 2 | 27 | 92 | FNA | CRS | No | Fresh | Mediastinal | No | 2:1 | Convenience |
| ||||||||||||||
Suzana, S.a | 2016 | India | 19 | 19 | 1 | 27 | Tissue | Cuture | No | Fresh | Peripheral | No | 2:1 | Consecutive |
| ||||||||||||||
Suzana, S.b | 2016 | India | 38 | 0 | 18 | 11 | Tissue | CRS | No | Fresh | Peripheral | No | 2:1 | Consecutive |
| ||||||||||||||
Wang Xiao | 2016 | China | 1 | 24 | 0 | 7 | FNA | Cuture | No | Fresh | Peripheral | No | 2:1 | Convenience |
| ||||||||||||||
Nataraj, G. | 2016 | India | 29 | 1 | 9 | 87 | FNA | Cuture | No | Fresh | Peripheral | No | 2:1 | Convenience |
| ||||||||||||||
Xu Meili | 2016 | China | 73 | 0 | 7 | 40 | FNA | CRS | No | Fresh | Peripheral/Mediastinal | Mechanical | 2:1 | Convenience |
| ||||||||||||||
Huang Shaojun.a | 2016 | China | 48 | 0 | 2 | 13 | FNA | CRS | NALC-NaOH | Fresh | Peripheral | Mechanical | 2:1 | Convenience |
| ||||||||||||||
Huang Shaojun.b | 2016 | China | 26 | 0 | 5 | 13 | Tissue | CRS | NALC-NaOH | Fresh | Peripheral | Mechanical | 2:1 | Convenience |
| ||||||||||||||
Pandey, S.a | 2017 | Australia | 9 | 1 | 1 | 3 | FNA | Cuture | NALC-NaOH | Fresh | Peripheral | No | 3:1 | Convenience |
| ||||||||||||||
Pandey, S.b | 2017 | Australia | 5 | 2 | 3 | 1 | Tissue | Cuture | NALC-NaOH | Fresh | Peripheral | No | 3:1 | Convenience |
| ||||||||||||||
Jorstad, M. D. | 2018 | Zanzibar | 6 | 0 | 16 | 20 | FNA | CRS | No | Fresh | Peripheral | No | Unknow | Convenience |
| ||||||||||||||
Perez-Risco, D. | 2018 | Spain | 16 | 0 | 1 | 8 | FNA | Cuture | NALC-NaOH | Frozen | Peripheral | No | 2:1 | Convenience |
The overall methodological quality of the included studies, using a CRS and culture, is summarized in Figure
Methodological quality graphs (risk of bias and applicability concerns) as percentages across the included studies. (a) Composite reference standard. (b) Culture reference standard.
Fifteen studies included comparison of 1597 FNA or tissue samples with a CRS; Xpert MTB/RIF sensitivity ranged from 49% (95% CI 35–63%) to 97% (95% CI 83–100%). Pooled sensitivity of Xpert MTB/RIF assay for LNTB was 79% (95% CI 69–86%) and I2 statistical values were 86%. Xpert MTB/RIF specificity ranged from 72% (95% CI 57–84%) to 100% (95% CI 97–100%). Pooled specificity of Xpert MTB/RIF assay was 98% (95% CI 94–99%) and I2 statistical values were 89% (Figure
Forest plot of Xpert sensitivity and specificity for tuberculosis detection in LNTB compared with a composite reference standard. The squares represent the sensitivity and specificity of a study, and the black line represents their confidence intervals. The diamonds represent the pooled sensitivity and specificity and their confidence intervals. LNTB: lymph nodes tuberculosis.
Compared to a culture reference standard, the pooled sensitivity of Xpert MTB/RIF was 84% (95% CI 77–90%) with I2 = 67% and specificity was 91% (95% CI 78–96%) with I2 = 92% for 1629 FNA or tissue specimens from 21 studies (Figure
Forest plot of Xpert sensitivity and specificity for tuberculosis detection in LNTB compared with culture reference standard. The squares represent the sensitivity and specificity of a study, and the black line represents their confidence intervals. The diamonds represent the pooled sensitivity and specificity and their confidence intervals. LNTB: lymph nodes tuberculosis.
We explored the heterogeneity among studies, using subgroup and metaregression analyses on predefined subgroups of sample types, decontamination method, sample condition, and homogenization used in the assay.
In case of Xpert MTB/RIF using FNA samples, sensitivity ranged from 27% (95% CI 11–50%) to 97% (95% CI 83–100%) against a CRS. The pooled sensitivity was 80% (95% CI 63–90%) with I2 = 90%, and the pooled specificity was 96% (95% CI 90–98%) with I2 = 89% (Figure
Forest plot of Xpert sensitivity and specificity for tuberculosis detection in LNTB versus composite reference standard. (a) FNA samples. (b) Tissue samples. The squares represent the sensitivity and specificity of a study, and the black line represents their confidence intervals. The diamonds represent the pooled sensitivity and specificity and their confidence intervals. LNTB: lymph nodes tuberculosis. FNA: fine needle aspiration.
In Xpert MTB/RIF using FNA samples, sensitivity ranged from 71% (95% CI 29–96%) to 100% (95% CI 78–100%) against culture. The pooled sensitivity was 90% (95% CI 83–95%) with I2 = 52%, and pooled specificity was 89% (95% CI 65–97%) with I2 = 95% (Figure
Forest plot of Xpert sensitivity and specificity for tuberculosis detection in LNTB versus culture. (a) FNA samples. (b) Tissue samples. The squares represent the sensitivity and specificity of a study, and the black line represents their confidence intervals. The diamonds represent the pooled sensitivity and specificity and their confidence intervals. LNTB: lymph nodes tuberculosis. FNA: fine needle aspiration.
Metaregression analysis showed that studies with FNA samples and tissue samples, compared to a CRS, showed similar sensitivity (79% and 78%, metaregression P = 0.35). However, studies with FNA samples, compared to a CRS, showed lower specificity (96%) than those using tissue samples (100%); the difference was statistically significant (metaregression P < 0.01). The sensitivity of studies using FNA samples was higher than that of studies using tissue samples, compared to culture (90% versus 76%); however, the difference was not statistically significant (meta-regression P = 0.43). The specificity of studies using FNA samples was lower than that of studies using tissue samples, compared to culture (89% versus 93%); this difference was also not statistically significant (metaregression P = 0.56).
Metaregression analysis showed that decontamination method (with NALC-NaOH or without), sample condition (fresh or frozen), homogenization (mechanical or otherwise), and patient population (high/low income) did not have any effect on Xpert MTB/RIF sensitivity and specificity (meta-regression P > 0.05), compared to that in CRS. These factors were, therefore, obviously not a source of heterogeneity across the studies.
Diagnosis of EPTB, including LNTB, is very challenging, due to the characteristics of a low bacterial load. In most cases, invasive examinations are necessary. For LNTB, the most commonly used invasive procedures are FNA and biopsy. Specimens obtained by different methods have different sensitivity for the diagnosis of LNTB. Sensitivity of pathological diagnosis is the highest with the most invasive biopsy specimens [
Nucleic acid detection has been widely used in the diagnosis of tuberculosis [
Our research found that the pooled sensitivity and specificity of Xpert MTB/RIF for LNTB were 79% (95% CI 69–86%) and 98% (95% CI 94–99%) versus CRS and 84% (95% CI 77–90%) and 91% (95% CI 78–96%) versus culture, respectively. In a meta-analysis by Denkinger et al., Xpert MTB/RIF pooled sensitivity and specificity were 81.2% (95% CI 72.4–87.7%) and 99.1% (95% CI 94.5–99.9%) versus CRS, and 83.1% (95% CI 71.4–90.7%) and 93.6% (95% CI 87.9–96.8%) versus culture for lymph node tissues or aspirates [
Sample processing of lymph node specimens, such as decontamination, sample condition, and homogenization, was variable across studies, but metaregression analysis showed that these factors did not affect the result and hence were not the sources of heterogeneity. In addition, we also found the patient's financial status to not affect the outcome.
According to the findings of this study, a multistep approach could be adopted for the management of suspected LNTB: FNA should be performed as the first step due to its less invasiveness, followed by complete relevant examinations, including Xpert MTB/RIF and pathological tests to improve diagnostic sensitivity. When the inspection result of the first step is negative, a more invasive technique (like biopsy) should be performed; pathological examination may be used for further validation, rather than Xpert MTB/RIF, since the assay did not increase the sensitivity of diagnosis when performed in biopsy-obtained samples.
Our meta-analysis also had several limitations. We realize that we may have missed some studies, despite the comprehensive search, and some studies that failed to distinguish specimen types. In addition, some included studies used multiple sample types, which may have led to some bias in our results. In addition, sample processing of lymph node specimens was highly variable across and within studies, since the assay, designed for respiratory samples, may slightly vary for other specimens. Additionally, the CRS standard for the studies was also different. Heterogeneity among the studies was remarkable, and the pooled estimates need to be interpreted with caution.
In this meta-analysis, we observed that the pooled sensitivity and specificity of Xpert MTB/RIF were 79% and 98%, respectively, when compared with a CRS, and 84% and 91%, respectively, when compared with culture. When performed on FNA samples, the pooled sensitivity and specificity were 80% and 96% versus CRS and 90% and 89% versus culture, respectively. When performed on tissue samples, the pooled sensitivity and specificity were 76% and 100% versus CRS, and 76% and 92% versus culture, respectively. There was no significant difference in the diagnostic efficiency for specimens obtained via different routes. Xpert MTB/RIF showed a good diagnostic efficiency on LNTB and was not related to the type of specimen.
The authors have no funding or conflicts of interest to disclose.
Search strategies for each database.