Advertisement

Clinical validation of urine-based Xpert® MTB/RIF assay for the diagnosis of urogenital tuberculosis: A systematic review and meta-analysis

Open AccessPublished:March 17, 2020DOI:https://doi.org/10.1016/j.ijid.2020.03.023

      Highlights

      • Low accuracy of conventional testing complicates clinical management of urogenital tuberculosis.
      • Xpert demonstrates promising diagnostic efficiency for urogenital tuberculosis.
      • Xpert performance may not be affected by differing characteristics of included studies.
      • Integrating traditional tests with Xpert assay may increase likelihood of urogenital tuberculosis detection.

      Abstract

      Objectives

      Effective methods for diagnosing urogenital tuberculosis (UGTB) are important for its clinical management. Therefore, we undertook a systematic review to assess the performance of the urine-based Xpert MTB/RIF assay for UGTB.

      Methods

      PubMed, Embase, Web of Science, the Cochrane library, and Scopus were systematically searched up to July 30, 2019. A hierarchical summary receiver operating characteristic (HSROC) was applied to calculate the pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and odds ratio (OR) for the diagnostic accuracy of the Xpert test.

      Results

      Our search identified 858 unique articles from which 69 studies were selected for full-text revision, with 12 studies meeting the inclusion criteria. Eleven studies comprising 1202 samples compared Xpert with mycobacterial culture, while 924 samples from eight studies compared it with a composite reference standard (CRS). The values for pooled sensitivity, specificity, PLR, NLR, and OR were 0.89, 0.95, 20.1, 0.18, and 159.53, respectively, when compared with the mycobacterial culture. Likewise, when compared with a CRS, the respective pooled sensitivity, specificity, PLR, NLR, and OR values were 0.55, 0.99, 40.67, 0.43, and 166.17, thereby suggesting a high level of accuracy for diagnosing UGTB. A meta-regression and sub-group analysis of TB-burden countries, study design, decontamination, concentration, and reference standard could not explain the heterogeneity (p > 0.05) in the diagnostic efficiency.

      Conclusions

      Our results suggested that Xpert is a promising diagnostic tool for the diagnosis of UGTB via urine specimen.

      Keywords

      Introduction

      Tuberculosis (TB) — both pulmonary and extra-pulmonary types of infection, as caused by Mycobacterium tuberculosis (MTB) — is responsible for an estimated global annual incidence of 10.4 million illnesses and 1.7 million deaths, ranking it above HIV/AIDS (
      • World Health Organization
      Global tuberculosis report 2017.
      ). Urogenital TB (UGTB), which refers to the infection of the urogenital organs, including kidney, urinary tract, and genitals, by MTB is the second-most common type of extrapulmonary TB (EPTB) (
      • Kulchavenya E.
      Urogenital tuberculosis – definition and classification.
      ,
      • Figueiredo A.A.
      • Lucon A.M.
      • Srougi M.
      Urogenital tuberculosis.
      ). Nevertheless, the number of people affected by UGTB is likely to be higher, given that UGTB develops in 2–20% of patients with pulmonary TB (PTB) (
      • Figueiredo A.A.
      • Lucon A.M.
      Urogenital tuberculosis: update and review of 8961 cases from the world literature.
      ;
      • Figueiredo A.A.
      • Lucon A.M.
      • Junior R.F.
      • Srougi M.
      Epidemiology of urogenital tuberculosis worldwide.
      ). Also, the diagnosis of UGTB is challenging due to its potentially atypical clinical presentation, and the tendency of UGTB to be masked mainly by urinary tract infections (
      • Kapoor R.
      • Ansari M.
      • Mandhani A.
      • Gulia A.
      Clinical presentation and diagnostic approach in cases of genitourinary tuberculosis.
      ), which often results in a delay or deprivation of treatment. Therefore, a high level of suspicion regarding claims of overlapping infection is often required, along with an urgent and rapid laboratory test for confirmation of diagnosis.
      Traditional diagnostic methods, such as urine microscopy and mycobacterial culture, are utilized as screening laboratory tests. Since UGTB is normally paucibacillary (
      • Tostmann A.
      • Kik S.V.
      • Kalisvaart N.A.
      • Sebek M.M.
      • Verver S.
      • Boeree M.J.
      • et al.
      Tuberculosis transmission by patients with smear-negative pulmonary tuberculosis in a large cohort in the Netherlands.
      ), most urinary specimens examined by microscopy display poor sensitivity and specificity. At present, mycobacterial culture is widely regarded as the method of choice for infection diagnosis (
      • Dunn J.J.
      • Starke J.R.
      • Revell P.A.
      Laboratory diagnosis of Mycobacterium tuberculosis infection and disease in children.
      ), but the identification of UGTB from urine samples is relatively inefficient, time consuming, and labor intensive, and diagnosis is often limited by the presence of low numbers of bacilli. In spite of various limitations, conventional testing remains the primary diagnostic tool in peripheral healthcare facilities. Reports on screening methods, including enzyme-linked immunosorbent assays, slide agglutination tests, and polymerase chain reaction, are available for EPTB; however, the sensitivities and specificities of these diagnostic tests are variable (
      • Barnes P.F.
      Rapid diagnostic tests for tuberculosis: progress but no gold standard.
      ). Additionally, these screening tests necessitate a number of manual steps, and some may require a comparably extended turnaround time.
      The Xpert MTB/RIF (Xpert), which is based on nucleic acid amplification technology, has been proven to be rapid in the detection of PTB and EPTB, with high accuracy (
      • Celik C.
      • Gozel M.G.
      • Bakici M.Z.
      • Berk S.
      • Ozşahin S.L.
      • Gültürk E.
      Applicability of Xpert MTB/RIF assay for routine diagnosis of tuberculosis: a four-year single-center experience.
      ). The assay concurrently detects MTB complex and the mutations conferring rifampicin resistance within 2 h (
      • Steingart K.R.
      • Sohn H.
      • Schiller I.
      • Kloda L.A.
      • Boehme C.C.
      • Pai M.
      • et al.
      Xpert® MTB/RIF assay for pulmonary tuberculosis and rifampicin resistance in adults.
      ,
      • Vadwai V.
      • Boehme C.
      • Nabeta P.
      • Shetty A.
      • Alland D.
      • Rodrigues C.
      Xpert MTB/RIF: a new pillar in diagnosis of extrapulmonary tuberculosis?.
      ). Furthermore, this assay requires minimal training and biosafety facilities, avoids being vulnerable to cross-contamination, and has a high responsiveness in smear-negative MTB (
      • Moure R.
      • Muñoz L.
      • Torres M.
      • Santin M.
      • Martín R.
      • Alcaide F.
      Rapid detection of Mycobacterium tuberculosis complex and rifampin resistance in smear-negative clinical samples by use of an integrated real-time PCR method.
      ). In spite of these advantageous features of Xpert as a tool for the detection of MTB, independent systematic research evaluating the accuracy of the Xpert assay for detection of UGTB in urine specimens remains ambiguous. In light of the need to make clinical decisions regarding the detection of UGTB, we systematically reviewed and analyzed the available data for all patients, including HIV and non-HIV individuals, so as to determine the pooled sensitivity and specificity of the Xpert assay with respect to mycobacterial culture (solid or liquid) and a composite reference standard (CRS).

      Methods

      Data source and search strategy

      A computer-aided literature search of reports published up to July 2019 was carried out, using the reference databases PubMed, Embase, Web of Science, the Cochrane Library, and Scopus, for studies evaluating the accuracy of Xpert in UGTB detection. The following search terms were used: “Xpert”, “GeneXpert”, “Roche”, Abbott”, “Cepheid”, “Urogenital tuberculosis”, “Urogenital TB”, “UGTB”, “Renal tuberculosis”, “Genitourinary tuberculosis”, “Extra-pulmonary tuberculosis”, without any limitations. The bibliographies of articles identified and deemed relevant were searched manually for possible additional candidate studies.

      Inclusion criteria

      The inclusion criteria were as follows: (1) search terms included in the title and abstract of published reports; (2) research conducted using human urinary samples; (3) research involving comparative evaluation of the Xpert test against a reference standard; (4) research assessing the diagnostic accuracy of Xpert in UGTB detection, with reliable data on sensitivity and specificity, or predictive values for one or more of the three tests evaluated to calculate these measures; (5) reports presenting sufficient data to calculate odds ratios.

      Exclusion criteria

      The exclusion criteria were as follows: (1) studies describing overviews of UGTB, technical aspects, reviews, mechanism studies, and therapeutic characterization; (2) conference reports, case reports, and abstracts without full texts; (3) non-English publications; (4) non-human samples and animal experiments; (5) specimens testing negative by all reference tests, studies that did not include a reference standard, and studies that did not include enough samples (≤5 urine samples).

      Data extraction

      Two investigators (K. Chen and A.A. Malik) independently screened the potentially relevant articles using a standard protocol to ensure the reproducibility of study selection. In cases of disagreement, a third investigator (S.C. Ojha) independently inspected such articles, and discrepancies were resolved by consensus. The following data were extracted from relevant studies: author, publication year, country, TB incidence per 100 000 population, study type, number of specimens, specimen type, reference standard (CRS/mycobacterial culture), sensitivity, and specificity, along with other parameters. Similar to the literature selection phase, inconsistencies between the two datasets were resolved by discussion with a third investigator. The studies comprising different reference standards were treated as separate studies.

      Quality assessment

      The methodological quality of the studies was assessed using a revised tool for the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) (
      • Whiting P.F.
      • Rutjes A.W.
      • Westwood M.E.
      • Mallett S.
      • Deeks J.J.
      • Reitsma J.B.
      • et al.
      QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies.
      ). QUADAS-2 comprised four domains: patient selection, index test, reference standard, and flow and timing. Signaling questions were included to help assessors judge potential bias. All domains for the potential risk of bias and the first three domains for applicability concerns were assessed by two investigators (K. Chen and A.A. Malik), with critical input from a third (S.C. Ojha). In cases of disagreement, discrepancies between the authors were resolved through discussion.

      Statistical analysis

      The meta-analysis was carried out using the following software: RevMan 5.3 for quality assessment and forest-plot generation; Meta-Disc for pooled summary estimates, likelihood ratios, OR, and heterogeneity determination; and R programming environment for hierarchical summary receiver operating characteristic (HSROC) curves. The values for individual parameters comprising true-positive, true-negative, false-positive, and false-negative were obtained from the compiled studies. The pooled sensitivity and specificity of the Xpert assay, along with 95% confidence intervals (CI), were calculated against mycobacterial culture and CRS using a random-effects model. Forest plots for sensitivity and specificity with 95% CI were generated separately to visually assess the heterogeneity among the data. The HSROC curve was subsequently constructed to present an overall summary of the diagnostic assay performance. Furthermore, an I-square (I2) statistical test was applied to assess the heterogeneity between the included studies and a reference standard. While 0% specified no observed heterogeneity, values >50% were considered as signifying substantial heterogeneity (
      • Higgins J.P.
      • Thompson S.G.
      • Deeks J.J.
      • Altman D.G.
      Measuring inconsistency in meta-analyses.
      ,
      • Zhou Y.
      • Dendukuri N.
      Statistics for quantifying heterogeneity in univariate and bivariate meta‐analyses of binary data: the case of meta‐analyses of diagnostic accuracy.
      ). We analyzed varying sample types, sample conditions, national TB burden, concentration method, reference standard, and decontamination as possible sources of heterogeneity, using subgroup and meta-regression analysis. It is speculated that sensitivity and specificity differ across studies due to dissimilarities in study populations, sampling errors, and variances in implicit thresholds as applied to the included data. Therefore, a random-effects model was used to evaluate data heterogeneity.

      Results

      Literature selection

      858 studies were identified from the computer-aided literature searches (PubMed, 174; Embase, 314; Web of Science, 274; the Cochrane Library, 26; Scopus, 70) (Figure 1). Initially, 388 articles were eliminated because of duplication in the databases. Of the remaining 470 identified studies, 69 were deemed potentially relevant based on titles and abstracts. Screening approaches are presented in Supplementary Table S1. Finally, 12 studies that met all of the inclusion criteria were used in the meta-analysis.
      Figure 1
      Figure 1Flow chart representing the selection criteria of eligible studies for meta-analysis.

      Characteristics of included studies

      The baseline characteristics of included studies are shown in Table 1. 1302 samples from 12 studies were included in this meta-analysis. The number of samples evaluated in each study ranged from 21 to 302, with a median value of 96. Overall, eight (67%) studies were carried out in high-burden TB countries, where the TB prevalence was 63–267 per 100 000 population (
      • Chen Y.
      • Wu P.
      • Fu L.
      • Liu Y.H.
      • Zhang Y.
      • Zhao Y.
      Multicentre evaluation of Xpert MTB/RIF assay in detecting urinary tract tuberculosis with urine samples.
      ,
      • Jing H.
      • Lu Z.
      • Deng Y.
      • Gao D.
      • Li L.
      • Graviss E.A.
      • et al.
      Evaluation of Xpert MTB/RIF in detection of pulmonary and extrapulmonary tuberculosis cases in China.
      ,
      • Khan A.S.
      • Ali S.
      • Khan M.T.
      • Ahmed S.
      • Khattak Y.
      • Abduljabbar
      • et al.
      Comparison of GeneXpert MTB/RIF assay and LED-FM microscopy for the diagnosis of extra pulmonary tuberculosis in Khyber Pakhtunkhwa, Pakistan.
      ,
      • Kim Y.W.
      • Kwak N.
      • Seong M.W.
      • Kim E.C.
      • Yoo C.G.
      • Kim Y.W.
      • et al.
      Accuracy of the Xpert(R) MTB/RIF assay for the diagnosis of extra-pulmonary tuberculosis in South Korea.
      ,
      • Li Y.
      • Pang Y.
      • Zhang T.
      • Xian X.
      • Wang X.
      • Yang J.
      • et al.
      Rapid diagnosis of extrapulmonary tuberculosis with Xpert Mycobacterium tuberculosis/rifampicin assay.
      ,
      • Pang Y.
      • Shang Y.
      • Lu J.
      • Liang Q.
      • Dong L.
      • Li Y.
      • et al.
      GeneXpert MTB/RIF assay in the diagnosis of urinary tuberculosis from urine specimens.
      ,
      • Samuel B.
      • Michael J.
      • Chandrasingh J.
      • Kumar S.
      • Devasia A.
      • Kekre N.
      Efficacy and role of Xpert® Mycobacterium tuberculosis/rifampicin assay in urinary tuberculosis.
      ,
      • Sharma S.K.
      • Kohli M.
      • Chaubey J.
      • Yadav R.N.
      • Sharma R.
      • Singh B.K.
      • et al.
      Evaluation of Xpert MTB/RIF assay performance in diagnosing extrapulmonary tuberculosis among adults in a tertiary care centre in India.
      ). Four (33%) studies were conducted in low-burden TB countries, where the TB incidence was 0.81–7.5 per 100 000 population (
      • Habous M.
      • Elimam M.A.E.
      • Kumar R.
      • Deesi Z.A.L.
      Evaluation of GeneXpert Mycobacterium tuberculosis/rifampin for the detection of Mycobacterium tuberculosis complex and rifampicin resistance in nonrespiratory clinical specimens.
      ,
      • Hillemann D.
      • Rusch-Gerdes S.
      • Boehme C.
      • Richter E.
      Rapid molecular detection of extrapulmonary tuberculosis by the automated GeneXpert MTB/RIF system.
      ,
      • Mazzola E.
      • Arosio M.
      • Nava A.
      • Fanti D.
      • Gesu G.
      • Farina C.
      Performance of real-time PCR Xpert®MTB/RIF in diagnosing extrapulmonary tuberculosis.
      ,
      • Tortoli E.
      • Russo C.
      • Piersimoni C.
      • Mazzola E.
      • Dal Monte P.
      • Pascarella M.
      • et al.
      Clinical validation of Xpert MTB/RIF for the diagnosis of extrapulmonary tuberculosis.
      ). Of the 12 selected studies, seven included both mycobacterial culture and CRS as reference standards, eight used CRS, while eleven studies used mycobacterial culture. The CRS included a study population in which subjects presented with clinical signs of urogenital TB and at least one positive test for the condition according to a positive mycobacterial culture, smear, histopathology, positive cystoscopy biopsy, or radiological signs, and were responsive to anti-TB treatment. All studies were published in English up to July 30, 2018.
      Table 1Characteristics of eligible studies included in the meta-analysis. Abbreviations: UAE, United Arab Emirates; ROK, Republic of Korea; ND, not defined; popn, population; Decontn, decontamination; Concn, concentration; CRS, composite reference standard.
      StudyYearCountryTB incidence/ 100 000 popnTB burdenStudy typeTotal samplesSpecimens/ patientSpecimen conditionDecontnConcnIndex testReference standard
      Chen et al.2019China63 (54–73)HighConvenience3021FreshYesYesXpertCulture, CRS
      Habous et al.2019UAE0.81 (0.69–0.93)LowRetrospective221FreshYesYesXpertCulture
      Hillemann et al.2011Germany7.5 (6.4–8.7)LowConvenience911FreshYesYesXpertCulture
      Jing et al.2017China63 (54–73)HighProspective211FreshNDNDXpertCulture, CRS
      Khan et al.2018Pakistan267 (189–357)HighConvenience47NDFreshYesYesXpertCulture, CRS
      Kim et al.2015ROK70 (65–75)HighRetrospective106> 1Fresh/frozenYesYesXpertCulture, CRS
      Li et al.2017China63 (54–73)HighConvenience301FreshYesYesXpertCulture
      Mazzola et al.2016Italy6.9 (5.9–8)LowRetrospective235> 1FreshYesYesXpertCulture
      Pang et al.2017China63 (54–73)HighProspective1631FreshYesYesXpertCulture, CRS
      Samuel et al.2018India204 (140–281)HighProspective100> 1FreshYesYesXpertCRS
      Sharma et al.2014India204 (140–281)HighConvenience55> 1FreshNDNDXpertCulture, CRS
      Tortoli et al.2012Italy6.9 (5.9–8)LowProspective130> 1FreshYesYesXpertCulture, CRS

      Quality appraisal

      The overall risk of bias and applicability concerns for eligible studies, using the QUADAS-2 checklist (
      • Whiting P.F.
      • Rutjes A.W.
      • Westwood M.E.
      • Mallett S.
      • Deeks J.J.
      • Reitsma J.B.
      • et al.
      QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies.
      ), is summarized in Figure 2. In this study, the patient selection domain was a potential source of bias because a few studies failed to avoid inappropriate exclusion of samples. With regards to applicability, most of the studies included samples from patients with a suggestion of UGTB, indicating a low concern of bias. The index test and reference standard domain were normally at low risk of bias, as Xpert use predefined detection criteria for binary responses. Nevertheless, applicability concerns regarding the index response are unclear because there is no standardized operating protocol for worldwide use. A risk of bias at flow and timing domain was not suspected because both index and reference standards were applied to the same urine specimen. The average quality of studies included in this meta-analysis was satisfactory.
      Figure 2
      Figure 2Forest plot of Xpert MTB/RIF sensitivity and specificity for UGTB detection in urine samples against (A) mycobacterial culture reference standard and (B) composite reference standard. The square represents the sensitivity and specificity of particular study, the black line indicates its confidence interval. Abbreviations: TP, true positive; FP, false positive; FN, false negative; TN, true negative; CI, confidence interval.

      Summary estimates

      The 11 studies that met the inclusion criteria for comparison of Xpert with mycobacterial culture accounted for a total of 1202 samples. The sensitivity and specificity of Xpert against mycobacterial culture as a reference standard ranged from 0.33 (95% CI 0.01–0.91) to 1.0 (95% CI 0.4–1.0) and 0.86 (95% CI 0.65–0.97) to 1.0 (95% CI 0.93–1.0), respectively (Figure 3A). The pooled summary estimates were: pooled sensitivity, 0.89 (95% CI 0.82–0.93); pooled specificity, 0.95 (95% CI 0.94–0.96); PLR, 20.1 (95% CI 10.32–39.1); NLR, 0.18 (95% CI 0.09–0.35); OR, 159.53 (95% CI 73.54–346.1). I2 statistical values for sensitivity and specificity were 39.4% and 87.2%, respectively, indicating significant heterogeneity. The area under the curve (AUC) for HSROC was 0.97 (95% CI 95.87–97.89), suggesting very good overall diagnostic validity (Figure 4A).
      Figure 3
      Figure 3Plot of methodological quality as percentages across the eligible studies for (A) mycobacterial culture reference standard and (B) composite reference standard.
      Figure 4
      Figure 4Plot of the HSROC for assessment of Xpert as applied to UGTB detection from confirmed studies of (A) mycobacterial culture and (B) composite reference standard. Gray open circles denote the data point from each of the investigation, red square denotes the summary estimate, and black solid line denotes the HSROC curve.
      Eight studies, including 924 samples, met the criteria for comparison of the Xpert assay with CRS. The Xpert sensitivity and specificity ranged from 0.22 (95% CI 0.03–0.6) to 1.0 (95% CI 0.48–1.0) and 0.93 (95% CI 0.81–0.99) to 1.0 (95% CI 0.98–1.0), respectively (Figure 3B). The pooled summary estimates for Xpert against CRS as reference standard were: pooled sensitivity, 0.55 (95% CI 0.49–0.60); pooled specificity, 0.99 (95% CI 0.98–1.0); PLR, 40.67 (95% CI 14.33–115.43); NLR, 0.43 (95% CI 0.29–0.64); OR, 166.2 (95% CI 60.92–453.27). I2 values for sensitivity and specificity were 83.0% and 49.4%, respectively, indicating significant heterogeneity between studies, in particular for sensitivity. The AUC of HSROC was 0.99 (95% CI 96.96–100), indicating good overall diagnostic validity (Figure 4B).
      Regarding overall UGTB detection in urine specimens, a total of 177 positive samples were detected out of 323 CRS positive samples, while 126 positive samples were detected out of 142 mycobacterial culture confirmed samples. In spite of the fact that overall UGTB detection was higher against CRS, the accuracy of Xpert was relatively higher against mycobacterial culture — 94.36% (95% CI 0.93–0.96), compared with 83.66 (95% CI 81.11–85.99) with CRS as a reference standard.

      Meta-regression and subgroup analysis

      Potential source of heterogeneity among studies were investigated using meta-regression analyses on predefined subgroups used in the assay. Meta-regression analysis indicated that TB burden (high/low), study design (convenience, prospective or retrospective), sample condition (fresh/frozen), decontamination (with or without), and specimen concentration (yes or no) were not significant sources of heterogeneity (meta-regression p > 0.05).

      Discussion

      Diagnosis of UGTB is challenging, and often delayed, because of the paucibacillary nature of the disease, non-specific clinical manifestations, and suboptimal accuracy of conventional tests (
      • Kapoor R.
      • Ansari M.
      • Mandhani A.
      • Gulia A.
      Clinical presentation and diagnostic approach in cases of genitourinary tuberculosis.
      ,
      • Tostmann A.
      • Kik S.V.
      • Kalisvaart N.A.
      • Sebek M.M.
      • Verver S.
      • Boeree M.J.
      • et al.
      Tuberculosis transmission by patients with smear-negative pulmonary tuberculosis in a large cohort in the Netherlands.
      ). An underdiagnosed infection can result in a substantial delay in treatment, which may contribute to the high morbidity and mortality associated with UGTB (
      • Kapoor R.
      • Ansari M.
      • Mandhani A.
      • Gulia A.
      Clinical presentation and diagnostic approach in cases of genitourinary tuberculosis.
      ;
      • Figueiredo A.A.
      • Lucon A.M.
      • Junior R.F.
      • Srougi M.
      Epidemiology of urogenital tuberculosis worldwide.
      ). The current microbiological methods of choice for UGTB diagnosis include smear examination, because of its simplicity and convenience; however, its sensitivity ranges from 42% to 52% for a mycobacterial positive culture (
      • Samuel B.
      • Michael J.
      • Chandrasingh J.
      • Kumar S.
      • Devasia A.
      • Kekre N.
      Efficacy and role of Xpert® Mycobacterium tuberculosis/rifampicin assay in urinary tuberculosis.
      ). In contrast, urine mycobacterial culture is considered fairly sensitive in diagnosing UGTB, but the prolonged turnaround time means that initiation of treatment is often delayed (
      • Samuel B.
      • Michael J.
      • Chandrasingh J.
      • Kumar S.
      • Devasia A.
      • Kekre N.
      Efficacy and role of Xpert® Mycobacterium tuberculosis/rifampicin assay in urinary tuberculosis.
      ,
      • Altez-Fernandez C.
      • Ortiz V.
      • Mirzazadeh M.
      • Zegarra L.
      • Seas C.
      • Ugarte-Gil C.
      Diagnostic accuracy of nucleic acid amplification tests (NAATs) in urine for genitourinary tuberculosis: a systematic review and meta-analysis.
      ). With existing reference standards presenting a number of limitations, a urine-based molecular technique for diagnosing UGTB could provide a more accurate estimation.
      The Xpert assay is currently one of the most used nucleic acid detection methods in TB diagnosis (
      • Celik C.
      • Gozel M.G.
      • Bakici M.Z.
      • Berk S.
      • Ozşahin S.L.
      • Gültürk E.
      Applicability of Xpert MTB/RIF assay for routine diagnosis of tuberculosis: a four-year single-center experience.
      ,
      • Kohli M.
      • Schiller I.
      • Dendukuri N.
      • Dheda K.
      • Denkinger C.M.
      • Schumacher S.G.
      • et al.
      Xpert® MTB/RIF assay for extrapulmonary tuberculosis and rifampicin resistance.
      ). This assay has also been advocated by WHO for the early detection of UGTB. A systematic review on EPTB, which involved detection of UGTB via urine specimens, suggested a good diagnostic performance of Xpert against mycobacterial culture as a reference standard (
      • Kohli M.
      • Schiller I.
      • Dendukuri N.
      • Dheda K.
      • Denkinger C.M.
      • Schumacher S.G.
      • et al.
      Xpert® MTB/RIF assay for extrapulmonary tuberculosis and rifampicin resistance.
      ). But the EPTB systemic review included case-control studies, and those with samples of less than five, which could overestimate the diagnostic accuracy of Xpert. Since there is lack of consensus regarding the use of these tests in UGTB detection, this study aimed to establish summary estimates for sensitivity and specificity of the Xpert assay in the diagnosis of UGTB compared with mycobacterial culture and a composite reference standard, and to simultaneously evaluate the heterogeneity among studies.
      This meta-analysis presents a comprehensive validation of the accuracy of Xpert with urine specimens collected from individuals with clinical suspicion of UGTB. Our study found that the pooled sensitivity and specificity of Xpert for UGTB detection were, respectively, 0.89 (95% CI 0.82–0.93) and 0.95 (95% CI 0.94–0.96) versus mycobacterial culture, and 0.55 (95% CI 0.49–0.60) and 0.99 (95% CI 0.98–1.0) versus CRS. In a meta-analysis by
      • Altez-Fernandez C.
      • Ortiz V.
      • Mirzazadeh M.
      • Zegarra L.
      • Seas C.
      • Ugarte-Gil C.
      Diagnostic accuracy of nucleic acid amplification tests (NAATs) in urine for genitourinary tuberculosis: a systematic review and meta-analysis.
      , the pooled sensitivity and specificity against assorted reference standard were 0.87 (0.66–0.96) and 0.91 (0.84–0.95), respectively. Although the diagnostic accuracy of Xpert for detection of UGTB in urine sample appear consistent with mycobacterial culture reference standard of our present study, the meta-analysis by
      • Altez-Fernandez C.
      • Ortiz V.
      • Mirzazadeh M.
      • Zegarra L.
      • Seas C.
      • Ugarte-Gil C.
      Diagnostic accuracy of nucleic acid amplification tests (NAATs) in urine for genitourinary tuberculosis: a systematic review and meta-analysis.
      only comprised two studies and also suffered from a lack of a precise reference standard. In our study, Xpert displayed high specificity but modest sensitivity against mycobacterial culture or CRS while diagnosing UGTB in urine specimens. Nevertheless, it can be used as an initial screening test in patients suspected of having UGTB in order to reduce diagnostic and treatment delays. Owing to the paucibacillary nature of the disease, a CRS could possibly offer a more precise reference standard.
      The positive attributes of our study include the comprehensive search strategy for data extraction and manipulation by independent analysts. The use of a standardized protocol, a uniform rigorous reference standard, a bivariate random-effects model for data analysis, and pre-specified subgroups to account for heterogeneity among datasets were also positive considerations. In addition, studies involving small sample sizes, which tend to overestimate the diagnostic performance of index tests, were not included.
      Our study also had several limitations. We did not include unpublished data, abstracts, conference papers, and non-English publications. The outcome of this meta-analysis could have introduced bias due to differences between studies, including patient selection, sample sizes of individual studies, study design, and non-standardized sample processing in independent studies. It is possible that we could have missed some important studies in spite of the comprehensive literature database search. In addition, the meta-analysis was limited by the small number of studies, particularly those using a CRS. It should also be noted that the CRS standards used for the included studies were also different. A number of studies included the results from clinical features suggestive of UGTB, AFB smear microscopy, cytology/histology, and responses to anti-tuberculosis treatment, while a few studies even considered the results of mycobacterial culture as part of the CRS. This could be a significant source of heterogeneity.
      Although the study designs, high-burden TB countries, decontamination, specimen concentration, and the reference standard (mycobacterial culture or CRS) were not significant sources of heterogeneity in this meta-regression analysis, these variables could possibly enhance heterogeneity and limit the generalizability of the overall performance of Xpert.
      In conclusion, the findings of this study demonstrated that Xpert is a promising tool for the early diagnosis of UGTB and supports the WHO guidelines for TB. The performance of this assay may not be affected by sampling from a high-burden TB country, or by different study designs, reference standards, and decontamination steps. However, the results of this meta-analysis indicated the necessity for integrating traditional tests to establish a diagnosis given their complementarity in terms of being able to identify the affected patients, degree of invasiveness, and causative agents. Regarding the higher diagnostic accuracy, further well-designed, large, multicenter, prospective studies are still needed to confirm this finding.

      Author contributions

      S.C.O. conceived the idea and contributed to the research design. K.C., A.A.M., C.N., and S.C.O. contributed to data curation, formal analysis, writing, review, and editing. S.A., O.C., C.S., S-Y.J., W.C., W.G., and C.L.D. contributed to verificating, reviewing, and editing the manuscript. All authors approved the final version of the manuscript before submission to the journal.

      Conflicts of interest

      None.

      Ethical approval

      Not applicable

      Acknowledgements

      S.C.O. gratefully acknowledges the financial support provided by the Affiliated Hospital of Southwest Medical University, China. C.N. and A.A.M. are grateful for support from the Center of Excellence on Medical Biotechnology (CEMB), the S&T Postgraduate Education and Research Development Office (PERDO), and the Office of Higher Education Commission (OHEC), Thailand.

      Appendix A. Supplementary data

      The following is Supplementary data to this article:

      References

        • Altez-Fernandez C.
        • Ortiz V.
        • Mirzazadeh M.
        • Zegarra L.
        • Seas C.
        • Ugarte-Gil C.
        Diagnostic accuracy of nucleic acid amplification tests (NAATs) in urine for genitourinary tuberculosis: a systematic review and meta-analysis.
        BMC Infect Dis. 2017; 17: 390
        • Barnes P.F.
        Rapid diagnostic tests for tuberculosis: progress but no gold standard.
        Am J Resp Crit Care Med. 1997; 155: 1497-1498
        • Celik C.
        • Gozel M.G.
        • Bakici M.Z.
        • Berk S.
        • Ozşahin S.L.
        • Gültürk E.
        Applicability of Xpert MTB/RIF assay for routine diagnosis of tuberculosis: a four-year single-center experience.
        Turk J Med Sci. 2015; 45: 1329-1334
        • Chen Y.
        • Wu P.
        • Fu L.
        • Liu Y.H.
        • Zhang Y.
        • Zhao Y.
        Multicentre evaluation of Xpert MTB/RIF assay in detecting urinary tract tuberculosis with urine samples.
        Sci Rep. 2019; 9: 11053
        • Dunn J.J.
        • Starke J.R.
        • Revell P.A.
        Laboratory diagnosis of Mycobacterium tuberculosis infection and disease in children.
        J Clin Microbiol. 2016; 54: 1434-1441
        • Figueiredo A.A.
        • Lucon A.M.
        Urogenital tuberculosis: update and review of 8961 cases from the world literature.
        Rev Urol. 2008; 10: 207-217
        • Figueiredo A.A.
        • Lucon A.M.
        • Srougi M.
        Urogenital tuberculosis.
        Microbiol Spectr. 2017; 5: 1-16
        • Figueiredo A.A.
        • Lucon A.M.
        • Junior R.F.
        • Srougi M.
        Epidemiology of urogenital tuberculosis worldwide.
        Int J Urol. 2008; 15: 827-832
        • Habous M.
        • Elimam M.A.E.
        • Kumar R.
        • Deesi Z.A.L.
        Evaluation of GeneXpert Mycobacterium tuberculosis/rifampin for the detection of Mycobacterium tuberculosis complex and rifampicin resistance in nonrespiratory clinical specimens.
        Int J Mycobacteriol. 2019; 8: 132-137
        • Higgins J.P.
        • Thompson S.G.
        • Deeks J.J.
        • Altman D.G.
        Measuring inconsistency in meta-analyses.
        BMJ. 2003; 327: 557-560
        • Hillemann D.
        • Rusch-Gerdes S.
        • Boehme C.
        • Richter E.
        Rapid molecular detection of extrapulmonary tuberculosis by the automated GeneXpert MTB/RIF system.
        J Clin Microbiol. 2011; 49: 1202-1205
        • Jing H.
        • Lu Z.
        • Deng Y.
        • Gao D.
        • Li L.
        • Graviss E.A.
        • et al.
        Evaluation of Xpert MTB/RIF in detection of pulmonary and extrapulmonary tuberculosis cases in China.
        Int J Clin Exp Pathol. 2017; 10: 4847-4851
        • Kapoor R.
        • Ansari M.
        • Mandhani A.
        • Gulia A.
        Clinical presentation and diagnostic approach in cases of genitourinary tuberculosis.
        Indian J Urol. 2008; 24: 401-405
        • Khan A.S.
        • Ali S.
        • Khan M.T.
        • Ahmed S.
        • Khattak Y.
        • Abduljabbar
        • et al.
        Comparison of GeneXpert MTB/RIF assay and LED-FM microscopy for the diagnosis of extra pulmonary tuberculosis in Khyber Pakhtunkhwa, Pakistan.
        Braz J Microbiol. 2018; 49: 909-913
        • Kim Y.W.
        • Kwak N.
        • Seong M.W.
        • Kim E.C.
        • Yoo C.G.
        • Kim Y.W.
        • et al.
        Accuracy of the Xpert(R) MTB/RIF assay for the diagnosis of extra-pulmonary tuberculosis in South Korea.
        Int J Tuberc Lung Dis. 2015; 19: 81-86
        • Kohli M.
        • Schiller I.
        • Dendukuri N.
        • Dheda K.
        • Denkinger C.M.
        • Schumacher S.G.
        • et al.
        Xpert® MTB/RIF assay for extrapulmonary tuberculosis and rifampicin resistance.
        Cochrane Database Syst Rev. 2018; 8CD012768
        • Kulchavenya E.
        Urogenital tuberculosis – definition and classification.
        Current therapy and surgery for urogenital tuberculosis. Springer, 2016: 31-49
        • Li Y.
        • Pang Y.
        • Zhang T.
        • Xian X.
        • Wang X.
        • Yang J.
        • et al.
        Rapid diagnosis of extrapulmonary tuberculosis with Xpert Mycobacterium tuberculosis/rifampicin assay.
        J Med Microbiol. 2017; 66: 910-914
        • Mazzola E.
        • Arosio M.
        • Nava A.
        • Fanti D.
        • Gesu G.
        • Farina C.
        Performance of real-time PCR Xpert®MTB/RIF in diagnosing extrapulmonary tuberculosis.
        Infez Med. 2016; 24: 304-309
        • Moure R.
        • Muñoz L.
        • Torres M.
        • Santin M.
        • Martín R.
        • Alcaide F.
        Rapid detection of Mycobacterium tuberculosis complex and rifampin resistance in smear-negative clinical samples by use of an integrated real-time PCR method.
        J Clin Microbiol. 2011; 49: 1137-1139
        • Pang Y.
        • Shang Y.
        • Lu J.
        • Liang Q.
        • Dong L.
        • Li Y.
        • et al.
        GeneXpert MTB/RIF assay in the diagnosis of urinary tuberculosis from urine specimens.
        Sci Rep. 2017; 7: 6181
        • Samuel B.
        • Michael J.
        • Chandrasingh J.
        • Kumar S.
        • Devasia A.
        • Kekre N.
        Efficacy and role of Xpert® Mycobacterium tuberculosis/rifampicin assay in urinary tuberculosis.
        Indian J Urol. 2018; 34: 268-272
        • Sharma S.K.
        • Kohli M.
        • Chaubey J.
        • Yadav R.N.
        • Sharma R.
        • Singh B.K.
        • et al.
        Evaluation of Xpert MTB/RIF assay performance in diagnosing extrapulmonary tuberculosis among adults in a tertiary care centre in India.
        Eur Respir J. 2014; 44: 1090-1093
        • Steingart K.R.
        • Sohn H.
        • Schiller I.
        • Kloda L.A.
        • Boehme C.C.
        • Pai M.
        • et al.
        Xpert® MTB/RIF assay for pulmonary tuberculosis and rifampicin resistance in adults.
        Cochrane Database Syst Rev. 2013; CD009593
        • Tortoli E.
        • Russo C.
        • Piersimoni C.
        • Mazzola E.
        • Dal Monte P.
        • Pascarella M.
        • et al.
        Clinical validation of Xpert MTB/RIF for the diagnosis of extrapulmonary tuberculosis.
        Eur Respir J. 2012; 40: 442-447
        • Tostmann A.
        • Kik S.V.
        • Kalisvaart N.A.
        • Sebek M.M.
        • Verver S.
        • Boeree M.J.
        • et al.
        Tuberculosis transmission by patients with smear-negative pulmonary tuberculosis in a large cohort in the Netherlands.
        Clin Infect Dis. 2008; 47: 1135-1142
        • Vadwai V.
        • Boehme C.
        • Nabeta P.
        • Shetty A.
        • Alland D.
        • Rodrigues C.
        Xpert MTB/RIF: a new pillar in diagnosis of extrapulmonary tuberculosis?.
        J Clin Microbiol. 2011; 49: 2540-2545
        • Whiting P.F.
        • Rutjes A.W.
        • Westwood M.E.
        • Mallett S.
        • Deeks J.J.
        • Reitsma J.B.
        • et al.
        QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies.
        Ann Intern Med. 2011; 155: 529-536
        • World Health Organization
        Global tuberculosis report 2017.
        WHO, Geneva2017
        • Zhou Y.
        • Dendukuri N.
        Statistics for quantifying heterogeneity in univariate and bivariate meta‐analyses of binary data: the case of meta‐analyses of diagnostic accuracy.
        Stat Med. 2014; 33: 2701-2717