Systematic Review and Meta-Analysis Medicine ®

OPEN

Efficacy of Xpert MTB/RIF Ultra in diagnosing

tuberculosis meningitis
A systematic review and meta-analysis
Yanqin Shen, Bachelor’s Degree, Guocan Yu, Bachelor’s Degree, Wuchen Zhao, Master’s Degree,
∗
Yazhen Lang, Bachelor’s Degree

Abstract
Background: This study aimed to assess whether Xpert MTB/RIF Ultra (Xpert Ultra) can effectively diagnose tuberculosis
meningitis (TBM) and to simultaneously compare its effectiveness with Xpert in diagnosing TBM in the same population.
Methods: On August 12, 2020, Wanfang Database, China National Knowledge Infrastructure, Embase, Cochrane Library, and
PubMed were searched for studies evaluating the diagnostic accuracy of Xpert Ultra for TBM. Then, we assessed the efficacy of
Xpert Ultra against a composite reference standard and culture. If applicable, we also examined the diagnostic efficacy of Xpert in the
same population. Heterogeneity was then explored by meta-regression, subgroup, and sensitivity analyses.
Results: Six studies containing 601 specimens reported the diagnostic efficacy of Xpert Ultra for TBM, with a composite reference
standard. No study had compared the efficacy between Xpert Ultra and culture. The pooled sensitivity of Xpert Ultra was 64% (95%
confidence interval [CI]: 45–80), and the I2 value was 86% (95% CI: 76–96); its specificity for TBM was consistently 100%. In the same
population, 5 studies compared the diagnostic efficacy between Xpert Ultra and Xpert for TBM. The pooled sensitivity of Xpert Ultra
and Xpert was 68% (95% CI: 46–84; I2 = 87%) and 37% (95% CI: 25–50; I2 = 72%), respectively. The studies were significantly
heterogeneous in terms of sensitivity but not heterogeneous in specificity.
Conclusions: Xpert Ultra was more sensitive than Xpert, but both were specific (100%). Therefore, Xpert Ultra had an excellent
diagnostic efficacy for TBM, and it could be the preferred initial test for TBM.
Abbreviations: AFB = acid-fast bacilli, CI = confidence interval, CRS = composite reference standard, CSF = cerebrospinal fluid,
EPTB = extrapulmonary tuberculosis, FN = false negative, FP = false positive, MTB = Mycobacterium tuberculosis, NAAT = nucleic
acid amplification tests, TBM = tuberculous meningitis, TN = true negative, TP = true positive.
Keywords: efficacy, meta-analysis, tuberculosis meningitis, Xpert, Xpert Ultra

1. Introduction individuals, resulting in 1.5 million deaths[1]; thus, it becomes the

leading cause of death among infectious diseases.[2] According to
Tuberculosis (TB) remains a serious global challenge to public
the presence or absence of lung involvement after Mycobacterium
health.[1] In 2018, TB was found in more than 10 million
tuberculosis (MTB) infection, TB has 2 main categories:
pulmonary tuberculosis and extrapulmonary tuberculosis
Editor: Haider Abdul-Lateef Mousa.
(EPTB).[3] The most serious type of EPTB is tuberculosis
YS and GY contributed equally to this work. meningitis (TBM); although its incidence is low, accounting
The authors have no funding and conflicts of interest to disclose. for only 1% to 5% of new tuberculosis infections, it causes severe
All data generated or analyzed during this study are included in this published disability or death in nearly half of the infected individuals.[4] One
article [and its supplementary information files]. of the main causes of these serious complications is the lack of
Zhejiang Tuberculosis Diagnosis and Treatment Center, Affiliated Hangzhou early and effective diagnostic tools, leading to delayed diagnosis
Chest Hospital, Zhejiang University School of Medicine, Zhejiang Chinese
and consequently, missed treatment.[5] Therefore, early diagnosis
Medicine and Western Medicine Integrated Hospital, Hangzhou, Zhejiang, China.
∗ is paramount in TBM management. Classical microbiological
Correspondence: Yazhen Lang, Zhejiang Tuberculosis Diagnosis and Treatment
Center, Zhejiang Chinese Medicine and Western Medicine Integrated Hospital,
assays cannot achieve an early and rapid diagnosis.[6] The most
Hangzhou, Zhejiang, China (e-mail: [email protected]). widely used tool for TBM diagnosis is cerebrospinal fluid (CSF)
Copyright © 2021 the Author(s). Published by Wolters Kluwer Health, Inc. acid-fast bacilli (AFB) smear with Ziehl–Neelsen staining, owing
This is an open access article distributed under the terms of the Creative to its simplicity and user-friendliness. However, AFB is
Commons Attribution-Non Commercial License 4.0 (CCBY-NC), where it is insensitive, especially when professional testers are unavail-
permissible to download, share, remix, transform, and buildup the work provided able.[7] Although MTB culture is more sensitive than AFB, the
it is properly cited. The work cannot be used commercially without permission
from the journal.
culture requires at least 2 weeks to produce the results; thus, early
clinical applications are not possible.[8] Therefore, MTB culture is
How to cite this article: Shen Y, Yu G, Zhao W, Lang Y. Efficacy of Xpert MTB/
RIF Ultra in diagnosing tuberculosis meningitis: a systematic review and meta- not effective for early diagnosis.[9] Hence, a rapid and effective
analysis. Medicine 2021;100:29(e26778). test for TBM is urgently needed.
Received: 20 January 2021 / Received in final form: 14 June 2021 / Accepted: Meanwhile, nucleic acid amplification tests (NAATs) are
30 June 2021 gaining prominence in rapid TB diagnosis.[10] Xpert MTB/RIF
http://dx.doi.org/10.1097/MD.0000000000026778 (Xpert, Cepheid, Sunnyvale, CA, USA) is the most classical and

1
Shen et al. Medicine (2021) 100:29 Medicine

widely used NAAT.[11] Xpert can rapidly diagnose TB; hence, it “Tubercular Meningitis” OR “Meningitides, Tubercular” OR
was recommended for the early diagnosis of pulmonary “Meningitis, Tubercular” OR “Tubercular Meningitides” OR
tuberculosis in 2010 and for some types of EPTB in 2013 by “Meningitis, Tuberculous” OR “Meningitides, Tuberculous”
the World Health Organization.[12] However, the test is still OR “Tuberculous Meningitides” OR “Tuberculous Meningitis”
flawed, and its sensitivity remains low in EPTB (eg, TBM and OR “Tuberculosis Meningitis” OR “Meningitides, Tuberculo-
tuberculous pleurisy) with low bacterial content.[13] For TBM, sis” OR “Meningitis, Tuberculosis” OR “Tuberculosis Menin-
although the World Health Organization recommends Xpert as gitides” OR “Tuberculous Hypertrophic Pachymeningitis” OR
the initial test, its sensitivity remains unsatisfactory.[14] A “Hypertrophic Pachymeningitides, Tuberculous” OR “Hyper-
negative Xpert result does not provide enough confidence to trophic Pachymeningitis, Tuberculous” OR “Pachymeningitides,
rule out TBM. Tuberculous Hypertrophic” OR “Pachymeningitis, Tuberculous
To improve the diagnostic performance for paucibacillary TB, Hypertrophic” OR “Tuberculous Hypertrophic Pachymeningi-
Cepheid developed Xpert MTB/RIF Ultra (Xpert Ultra), which is tides”
the next-generation Xpert; it shares the same equipment platform #2 “Extrapulmonary tuberculosis” OR “Extra pulmonary
with Xpert without updating.[15] Xpert Ultra adds 2 new MTB tuberculosis”
target genes (IS1081 and IS6110) and provides a larger capacity #3 “Meningitis”[Mesh] OR Meningitides OR Pachymeningitis
for DNA amplification reaction chamber.[16] These improve- OR Pachymeningitides
ments greatly enhance the effectiveness of Xpert Ultra in #4 “Cerebrospinal Fluid”[Mesh] OR “Cerebrospinal Fluids” OR
diagnosing TB.[16,17] The diagnostic efficacy of Xpert Ultra is “Fluid, Cerebrospinal” OR “Fluids, Cerebrospinal” OR “Cer-
also beneficial for paucibacillary TB.[18–20] However, the ebro Spinal Fluid” OR “Cerebro Spinal Fluids” OR “Fluid,
diagnostic efficacy of Xpert Ultra for TBM in comparison with Cerebro Spinal” OR “Fluids, Cerebro Spinal” OR “Spinal Fluid,
culture or a composite reference standard (CRS) remains Cerebro” OR “Spinal Fluids, Cerebro”
controversial; 1 study showed that Xpert Ultra was more #5 #1 OR #2 OR #3 OR #4
effective than Xpert[21] but another study demonstrated that the #6 “Xpert Ultra” OR “GeneXpert Ultra”
efficacy of Xpert Ultra revealed no statistically significant # 7 “Xpert” OR “GeneXpert”
improvement than that of Xpert.[22] Hence, this study aimed #8 #5 AND #6 AND #7
to conduct a systematic review and meta-analysis to better assess Similar search formulae were used for Wanfang databases,
the role of Xpert Ultra in TBM diagnosis in comparison with the China National Knowledge Infrastructure, Embase, and the
CRS and to simultaneously compare its effectiveness with Xpert Cochrane Library.
when diagnosing TBM in the same population.
2.4. Eligibility criteria
2. Material and methods 2.4.1. Type of study. Any type of study that had evaluated the
accuracy of Xpert Ultra for TBM.
2.1. Design and registration
2.4.2. Participants. TBM participants diagnosed using Xpert
We designed a systematic review and meta-analysis to assess the Ultra without any limitations on age, gender, or nationality.
efficacy of Xpert Ultra in diagnosing TBM. We registered the
protocol on the International Platform of Registered Systematic 2.4.3. Index test. Xpert Ultra was considered as the index test.
Review and Meta-Analysis Protocols (INPLASY, registration
2.4.4. Comparator test. A comparator test was optional. A
number: INPLASY202080045).[23] The study results were
study with satisfied participants, intervention, and outcomes can
reported in reference to the Preferred Reporting Items for
be enrolled even if it was a single-arm study.
Systematic Reviews and Meta-Analyses statements.[24] Further-
more, ethical approval was waived for systematic review and 2.4.5. Target conditions. Full-text original studies evaluating
meta-analysis. the Xpert Ultra assay in TBM diagnosis, with clear reference
standards and comprehensive data to extract or calculate true
2.2. Information sources positive (TP), false positive (FP), false negative (FN), and true
negative (TN) values, were considered eligible, thereby included
On August 12, 2020, Wanfang Database, China National in the study. However, studies published in languages other than
Knowledge Infrastructure, Embase, Cochrane Library, and English and Chinese, abstracts and conference reports without
PubMed were searched for studies assessing the diagnostic full text, case reports, and studies with <10 specimens did not
accuracy of Xpert Ultra in diagnosing TBM. The publication time meet the criteria; thus, they were excluded.
period of articles was from the database creation to August 12,
2020. Relevant references cited in the review were also screened 2.4.6. Outcomes. The main outcomes were the sensitivity and
to find studies that potentially met the criteria. specificity of the Xpert Ultra system for TBM. Sensitivity refers to
the probability that the index test result will be positive in an
infected case, the calculation formula is TP/(TP + FN)  100%.
2.3. Search strategy Specificity refers to the probability that the index test result will
YS and GY designed the search strategies for this study. Our be negative in a non-infected case, the calculation formula is TN/
search had no language or time limits. The search formula of (FP + TN)  100%.
PubMed is listed as follows:
2.4.7. Reference standards. Culture or a CRS was used as the
#1 “Tuberculosis, Meningeal”[Mesh] OR “Meningeal Tubercu- reference standard. A CRS comprised clinical symptoms, imaging
loses” OR “Meningeal Tuberculosis” OR “Tuberculoses, features, CSF biochemical analysis, MTB smears, culture, and
Meningeal” OR “TB Meningitis” OR “TB Meningitides” OR effectiveness to anti-TB therapy.

2
Shen et al. Medicine (2021) 100:29 www.md-journal.com

2.5. Literature screening and selection 3. Results

The collected works of literature were managed by the 3.1. Characteristics of the studies
ENDNOTE X9.2 literature management software (Clarivate
Corporation, Stanford, USA), where all search records were By searching the relevant databases using our designed search
imported. In selecting eligible articles, YS and GY independently strategies, we acquired 153 candidate articles. Ultimately, we
assessed all imported articles by reviewing their titles and found 6 articles that met the inclusion criteria for the final
abstracts and then the full text according to the inclusion criteria. quantitative analysis[21,22,28–31] (Fig. 1). The study type of all 6
Any dispute that occurred between the 2 researchers was included articles was a prospective study. The kappa value of
discussed with another researcher (YL). First, we gave the agreement between the 2 researchers at the literature screening
controversial literature to a third researcher for independent and data extraction stages was 0.736 (95% CI: 0.489–0.983). All
evaluation, then the 3 researchers discussed and reported the studies were conducted in developing countries with TB
reasons for inclusion or exclusion, respectively, and then the epidemic, with English as the language of publication. In
literature was included or excluded according to the inclusion addition, CSF is the selected specimen type for all articles. All
and exclusion criteria after obtaining agreement. studies used CRS as the reference standard, and no studies used
culture as the reference standard. Five articles simultaneously
compared the diagnostic efficacy of Xpert Ultra and Xpert in
2.6. Data extraction TBM against CRS in the same population.[21,22,28,30,31] The
For each included article, the following were extracted: first range of study specimen volume for Xpert Ultra was 21 to 204,
author’s name; publication year; country; TP, FP, FN, and TN with a median specimen volume of 93.5 and a total specimen
values for the assay; research type; patient selection method; volume of 601. For Xpert, the range of study specimen volume
sample type; sample condition; decontamination method; and was 21 to 166, with a median specimen volume of 102 and a total
homogenization along with other parameters. If an article specimen volume of 478. Table 1 lists the characteristics of the
compared the diagnostic efficacy of the Xpert Ultra and Xpert in included studies. However, 2 articles were excluded because 1
the same population, the relevant data of the Xpert were similarly article only reported sensitivity[32] and the other 1 analyzed the
extracted. The same 2 researchers individually extracted the same data as 1 of the included articles.[33] We also excluded 2
aforementioned relevant data from each included article and then articles that did not separately report data related to Xpert Ultra
cross-checked them, with the disputed data being resolved by for TBM diagnosis.[34,35]
discussing with the third researcher.
3.2. Study quality
2.7. Quality evaluation The results of the overall methodological quality assessment of
The quality of the included literature was assessed by the same 2 the included studies are displayed in Figure 2, with the CRS as the
researchers using a revised tool for the Quality Assessment of reference standard. The risk of bias emerged primarily from
Diagnostic Accuracy Studies independently.[25] Any disagree- patient selection and the reference standard; nonetheless, its flow
ment between researchers was resolved via a discussion with the and timing from the index test were relatively low.
third researcher (YL). According to the Preferred Reporting Items
for Systematic Reviews and Meta-Analysis for Diagnostic Test
3.3. Diagnostic efficacy of Xpert Ultra and Xpert for TBM
Accuracy guideline, publication bias did not need to be
assessed.[26] Six studies containing 601 specimens reported the efficacy of
Xpert Ultra, with a CRS, in diagnosing TBM. The sensitivity of
Xpert Ultra ranged from 44% (95% CI: 29–60) to 95% (95% CI:
2.8. Data synthesis and statistical analysis
77–100), whereas its pooled sensitivity in diagnosing TBM was
Initially, we determined the TP, FP, FN, and TN values for each 64% (95% CI: 45–80), with the I2 value of 86% (95% CI: 76–
study. Then, the pooled sensitivity and specificity with 95% 96); its specificity for TBM was consistent (all 100%) (Fig. 3). The
confidence interval (CI) of the Xpert Ultra or Xpert for TBM studies were significantly heterogeneous in terms of sensitivity but
diagnosis in comparison with the CRS were calculated using the not heterogeneous in specificity.
bivariate random-effects models. The forest plots of the Meanwhile, 5 studies containing 478 specimens reported the
sensitivity and specificity for each study were generated using diagnostic efficacy of Xpert, with a CRS, for TBM. The sensitivity
RevMan version 5.3 (Cochrane Collaboration, Oxford, United of Xpert ranged from 19% (95% CI: 8–33) to 56% (95% CI: 40–
Kingdom). Meanwhile, heterogeneity between the studies was 70), whereas its pooled sensitivity in diagnosing TBM was 37%
assessed using the I2 statistics. A value of 0% indicated no (95% CI: 25–50), with the I2 value of 72% (95% CI: 46–98). The
heterogeneity, whereas a value >50% indicated substantial specificity of the Xpert for TBM was also consistent (all 100%)
heterogeneity.[27] Heterogeneity sources were explored by meta- (Fig. 3). The studies were significantly heterogeneous in terms of
regression, subgroup, and sensitivity analyses. Potential sources sensitivity.
of heterogeneity included the patient selection method, sample For all studies, Xpert Ultra had a higher pooled sensitivity than
condition, decontaminate method, and homogenization. Meta- Xpert, but the specificity of both was consistent. For the same
analyses and meta-regression analyses were conducted using population, 5 studies compared the efficacy of Xpert Ultra and
Stata version 15.0 (Stata Corp., College Station, TX, the USA) Xpert in diagnosing TBM. The pooled sensitivity of Xpert Ultra
and midas. At least 4 studies were required for meta-analysis to and Xpert was 68% (95% CI: 46–84; I2 = 87%) and 37% (95%
obtain the combined effect values for each preset parameter CI: 25–50; I2 = 72%), respectively (Fig. 4). In each study, Xpert
using Stata. Ultra had higher sensitivity and pooled sensitivity than Xpert.

3
Shen et al. Medicine (2021) 100:29 Medicine

Figure 1. Flow chart of literature retrieval. In total, 3, 96, 10, 10, and 34 articles were found in Wanfang Database, China National Knowledge Infrastructure,
Embase, Cochrane Library, and PubMed, respectively.

The heterogeneity between studies in terms of sensitivity was tion, decontamination method, and homogenization method did
determined by meta-regression, subgroup, and sensitivity not affect the sensitivity of Xpert Ultra and Xpert for TBM
analyses. The patient selection method, sample condition, diagnosis in comparison with the CRS (meta-regression P > .05).
decontamination method, and homogenization method used in In the subgroup analyses performed on the subgroups of the
the assay of the predefined subgroups were assessed by meta- consecutive patient selection method, decontamination without
regression and subgroup analyses. Meta-regression analysis N-acetyl-L-cysteine-sodium hydroxide (NALC-NaOH), and
demonstrated that the patient selection method, sample condi- mechanical homogenization method of Xpert Ultra, the

Table 1
Characteristics of the included studies.
patient
Sample Research Decontaminate Sample selection
Test Author Year County type Reference type N TP FP FN TN method condition Homogenisation method
Xpert Ultra Bahr et ala 2018 Uganda CSF CRS Prospective 129 21 0 1 107 No Frozen Mechanical Consecutive
Wang et ala 2019 China CSF CRS Prospective 60 19 0 24 17 NALC-NaOH Frozen Mechanical Consecutive
Zhang et ala 2019 China CSF CRS Prospective 21 10 0 5 6 NALC-NaOH Fresh Mechanical Convenience
Cresswell et ala 2020 Uganda CSF CRS Prospective 204 39 0 12 153 No Fresh No Consecutive
Donovan et ala 2020 Vietnam CSF CRS Prospective 103 25 0 28 50 No Fresh Mechanical Convenience
Shao et al 2020 China CSF CRS Prospective 84 28 0 32 24 No Frozen No Consecutive
Xpert Bahr et alb 2018 Uganda CSF CRS Prospective 129 10 0 12 107 No Frozen Mechanical Consecutive
Wang et al.b 2019 China CSF CRS Prospective 60 8 0 35 17 NALC-NaOH Frozen Mechanical Consecutive
Zhang et alb 2019 China CSF CRS Prospective 21 4 0 11 6 NALC-NaOH Fresh Mechanical Convenience
Cresswell et alb 2020 Uganda CSF CRS Prospective 166 25 0 20 121 No Fresh No Consecutive
Donovan et alb 2020 Vietnam CSF CRS Prospective 102 21 0 32 49 No Fresh Mechanical Convenience
CRS = composite reference standard, CSF = cerebrospinal fluid, FN = false negative, FP = false positive, TN = true negative, TP = true positive.

4
Shen et al. Medicine (2021) 100:29 www.md-journal.com

Figure 2. Methodological quality graphs of risk of bias and applicability concerns, presented as percentages across the included studies against the composite
reference standard. (A) Xpert Ultra. (B) Xpert.

heterogeneity between studies in the subgroups remained highly preventing patients to receive optimal treatments and leading to
significant (I2 > 50%). For the Xpert, the heterogeneity also serious consequences, such as severe disability and death. As a
remained highly significant (I2 > 50%) according to the subgroup result, the patient’s prognosis worsens, and the burden on the
analyses that were performed on the subgroup of mechanical patient’s family increases.[22] In patients with suspected TBM, a
homogenization method. However, the studies included in other lumbar puncture to obtain a CSF sample for correlation testing is
subgroups of Xpert were limited; thus, subgroup analysis could the most common and critical step. However, the MTB content in
not be conducted. In sensitivity analyses, a particular study was CSF is extremely low, and the probability of obtaining a positive
either included or excluded to reanalyze whether the conclusions result by AFB and culture is also very low[37]; therefore,
had changed. Unfortunately, sensitivity analysis did not identify diagnosing TBM using these 2 methods alone still does not
studies that resulted in significant heterogeneity. meet the clinical needs.[37] In the studies included in this meta-
analysis, no study had compared the sensitivity and specificity
between Xpert Ultra and culture; thus, the current study used
4. Discussion
CRS as the reference standard for correlation analysis. However,
Similar to other paucibacillary EPTB, the MTB content of the test larger clinical studies are required to confirm the relevant
specimen was low, making the early diagnosis of TBM extremely diagnostic efficacy of Xpert Ultra in comparison with that of
difficult.[36] Delays in early diagnosis allow for treatment deficits, culture.

Figure 3. Forest plot for the sensitivity and specificity of Xpert Ultra and Xpert for the diagnosis of tuberculosis meningitis diagnosis against a composite reference
standard.

5
Shen et al. Medicine (2021) 100:29 Medicine

Figure 4. Forest plot for the sensitivity and specificity of Xpert Ultra and Xpert for the diagnosis of tuberculosis meningitis diagnosis against a composite reference
standard in the same population.

Aside from AFB and culture, other valid tests have also been specificity between the 2 tests and culture examination. As a
explored for TBM diagnosis. Owing to molecular biology composite standard with multiple factors, CRS might lead to
advancement, rapid microbiological testing was made possible, sensitivity decrement and specificity increment. Therefore, for
with NAAT as the crucial step. Xpert is the leading NAAT in TB TBM confirmed by a positive CSF culture, the sensitivity of Xpert
diagnostics and is widely used. This test has led to a significant Ultra should be higher than that observed in the current study.
increase in the diagnostic efficacy of TB because it shortens the Moreover, we observed significant heterogeneity in the
diagnosis time and allows a considerably earlier therapeutic sensitivity of Xpert Ultra and Xpert. The patient selection
window, thereby beneficial for TB control.[11] However, Xpert still method, sample condition, decontamination method, and
seems impractical for paucibacillary EPTB such as TBM. Xpert’s homogenization method were different among the studies. The
diagnostic performance in TBM remains unsatisfactory. There- sources of heterogeneity were explored in these terms. However,
fore, more efficient detection methods are still being developed. according to the meta-regression and subgroup analyses, these
Based on the Xpert, the second-generation Xpert Ultra was factors did not affect the sensitivity. Similarly, the sensitivity
developed by Cepheid. According to numerous reported improve- analysis did not reveal particularly heterogeneous articles. In
ments, the Xpert Ultra is considerably more sensitive than the addition, CRS might be different in each independent study (eg,
Xpert. However, the diagnostic performance of the Xpert Ultra for some studies had not included treatment response), which itself
TBM remains unclear. Most of the studies agreed that the Xpert might be a source of heterogeneity.
Ultra is more sensitive than the Xpert statistically,[21,28,29] but This meta-analysis had several limitations. Some studies might
another study reported otherwise.[22] Although a meta-analysis be missed during literature screening, although we had tried to
mentioned the role of Xpert Ultra in TBM,[38] the study included expand the search the best that we could. In addition, some
only 2 papers on TBM, an extremely limited number, and our study studies did not report TBM data separately; hence, the final
included more studies, and the results might be more informative. results might be biased. The CRS might also differ across
To the best of our knowledge, no independent systematic review individual studies. The sensitivity between the studies was clearly
and meta-analysis on the efficacy of Xpert Ultra in TBM diagnosis heterogeneous; therefore, the pooled estimates of sensitivity must
are available; hence, we conducted the present study. be treated prudently.
Six studies using CRS as the gold standard were included in this
study; they demonstrated that Xpert Ultra had a pooled
sensitivity of 64% and a specificity of 100% for TBM diagnosis, 5. Conclusions
indicating that Xpert Ultra has good diagnostic efficacy for TBM. This was the first independent systematic review and meta-analysis
The sensitivity results across independent studies were strikingly on the efficacy of Xpert Ultra in TBM diagnosis. The pooled
different, and a substantial level of heterogeneity was observed. sensitivity and specificity of Xpert Ultra for TBM diagnosis were
Moreover, 5 studies compared the efficacy between Xpert Ultra 64% and 100%, respectively. Xpert Ultra was more sensitive than
and Xpert for TBM diagnosis in the same population. The pooled Xpert, and both were identical in terms of specificity (100%).
sensitivity of Xpert Ultra and Xpert was 68% and 37%, Therefore, Xpert Ultra is highly effective in diagnosing TBM, and it
respectively, and the specificity of both tests was tremendously could be the preferred initial test for TBM.
high (100%). In the same population, the sensitivity of Xpert
Ultra was significantly better than that of Xpert. The sensitivity of
Xpert in this population was low, possibly related to the low Acknowledgments
number of specimens included in the study. A recent meta- We thank the patients and their families who were included in this
analysis on Xpert for TBM diagnosis showed that the Xpert had a study, and our colleagues in the department.
pooled sensitivity of 63% and a pooled specificity of 98.1%.[39]
Although the Xpert sensitivity of that study was higher than ours,
Author contributions
it was still lower than the sensitivity of Xpert Ultra in this study.
Meanwhile, the present study did not compare the sensitivity and Data curation: Yanqin Shen, Guocan Yu, Yazhen Lang.

6
Shen et al. Medicine (2021) 100:29 www.md-journal.com

Formal analysis: Yanqin Shen, Guocan Yu. [17] Horne DJ, Kohli M, Zifodya JS, et al. Xpert MTB/RIF and Xpert MTB/
RIF Ultra for pulmonary tuberculosis and rifampicin resistance in adults.
Investigation: Yanqin Shen, Wuchen Zhao.
Cochrane Database Syst Rev 2019;6:Cd009593.
Methodology: Yanqin Shen, Guocan Yu, Wuchen Zhao, Yazhen [18] Bisognin F, Lombardi G, Lombardo D, Re MC, Dal Monte P.
Lang. Improvement of Mycobacterium tuberculosis detection by Xpert
Project administration: Yazhen Lang. MTB/RIF Ultra: a head-to-head comparison on Xpert-negative samples.
Resources: Yanqin Shen, Guocan Yu. PLoS One 2018;13:e0201934.
[19] Dorman SE, Schumacher SG, Alland D, et al. Xpert MTB/RIF Ultra for
Software: Yanqin Shen, Guocan Yu, Yazhen Lang. detection of Mycobacterium tuberculosis and rifampicin resistance: a
Supervision: Wuchen Zhao. prospective multicentre diagnostic accuracy study. Lancet Infect Dis
Validation: Wuchen Zhao. 2018;18:76–84.
Visualization: Yazhen Lang. [20] Wang G, Wang S, Yang X, et al. Accuracy of Xpert MTB/RIF Ultra for
the diagnosis of pleural TB in a multicenter cohort study. Chest
Writing – original draft: Yanqin Shen, Guocan Yu.
2019;157:268–75.
Writing – review & editing: Yanqin Shen, Guocan Yu, Yazhen [21] Cresswell FV, Tugume L, Bahr NC, et al. Xpert MTB/RIF Ultra for the
Lang. diagnosis of HIV-associated tuberculous meningitis: a prospective
validation study. Lancet Infect Dis 2020;20:308–17.
[22] Donovan J, Thu DDA, Phu NH, et al. Xpert MTB/RIF Ultra versus Xpert
References MTB/RIF for the diagnosis of tuberculous meningitis: a prospective,
[1] World Health Organization. Global Tuberculosis Report 2019. 2019. randomised, diagnostic accuracy study. Lancet Infect Dis 2020;20:299–307.
[2] Cresswell FV, Ellis J, Kagimu E, et al. Standardized urine-based [23] Shen Y, Yu G, Zhao W, Lang LG. Diagnostic accuracy of Xpert MTB/RIF
tuberculosis (TB) screening with TB-lipoarabinomannan and Xpert Ultra for tuberculous meningitis: a protocol of systematic review and meta-
MTB/RIF Ultra in Ugandan adults with advanced human immunodefi- analysis. Inplasy Protocol 2020;doi: 10.37766/inplasy2020.8.0045.
ciency virus disease and suspected meningitis. Open Forum Infect Dis [24] Moher D, Liberati A, Tetzlaff J, Altman DG. PRISMA GroupPreferred
2020;7:ofaa100. reporting items for systematic reviews and meta-analyses: the PRISMA
[3] Sunnetcioglu A, Sunnetcioglu M, Binici I, Baran AI, Karahocagil MK, statement. PLoS Med 2009;6:e1000097.
Saydan MR. Comparative analysis of pulmonary and extrapulmonary [25] Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised tool
tuberculosis of 411 cases. Ann Clin Microbiol Antimicrob 2015;14:34. for the quality assessment of diagnostic accuracy studies. Ann Intern Med
[4] Thao LTP, Heemskerk AD, Geskus RB, et al. Prognostic models for 9- 2011;155:529–36.
month mortality in tuberculous meningitis. Clin Infect Dis 2018;66: [26] McInnes MDF, Moher D, Thombs BD, et al. Preferred reporting items
523–32. for a systematic review and meta-analysis of diagnostic test accuracy
[5] Sheu JJ, Yuan RY, Yang CC. Predictors for outcome and treatment delay studies: the PRISMA-DTA statement. JAMA 2018;319:388–96.
in patients with tuberculous meningitis. Am J Med Sci 2009;338:134–9. [27] Yu G, Zhong F, Ye B, Xu X, Chen D, Shen Y. Diagnostic accuracy of the
[6] Thwaites GE. Advances in the diagnosis and treatment of tuberculous Xpert MTB/RIF assay for lymph node tuberculosis: a systematic review
meningitis. Curr Opin Neurol 2013;26:295–300. and meta-analysis. Biomed Res Int 2019;2019:4878240.
[7] Heemskerk AD, Donovan J, Thu DDA, et al. Improving the [28] Bahr NC, Nuwagira E, Evans EE, et al. Diagnostic accuracy of Xpert
microbiological diagnosis of tuberculous meningitis: a prospective, MTB/RIF Ultra for tuberculous meningitis in HIV-infected adults: a
international, multicentre comparison of conventional and modified prospective cohort study. Lancet Infect Dis 2018;18:68–75.
Ziehl-Neelsen stain, GeneXpert, and culture of cerebrospinal fluid. J [29] Shao L, Qiu C, Zheng L, et al. Comparison of diagnostic accuracy of the
Infect 2018;77:509–15. GeneXpert Ultra and cell-free nucleic acid assay for tuberculous meningitis:
[8] Wilkinson RJ, Rohlwink U, Misra UK, et al. Tuberculous meningitis. Nat a multicentre prospective study. Int J Infect Dis 2020;98:441–6.
Rev Neurol 2017;13:581–98. [30] Wang G, Wang S, Jiang G, et al. Xpert MTB/RIF Ultra improved the
[9] Wang YY, Xie BD. Progress on diagnosis of tuberculous meningitis. diagnosis of paucibacillary tuberculosis: a prospective cohort study. J
Methods Mol Biol 2018;1754:375–86. Infect 2019;78:311–6.
[10] Chitnis AS, Davis JL, Schecter GF, Barry PM, Flood JM. Review of [31] Zhang P, Zheng J, Fu L, et al. Value of Xpert MTB /RIF Ultra in the
nucleic acid amplification tests and clinical prediction rules for diagnosis diagnosis of tuberculous meningitis. J Tuberc Lung Health 2019;8:42–7.
of tuberculosis in acute care facilities. Infect Control Hosp Epidemiol Chinese.
2015;36:1215–25. [32] Chin JH, Musubire AK, Morgan N, et al. Xpert MTB/RIF Ultra for
[11] Kohli M, Schiller I, Dendukuri N, et al. Xpert(®) MTB/RIF assay for detection of Mycobacterium tuberculosis in cerebrospinal fluid. J Clin
extrapulmonary tuberculosis and rifampicin resistance. Cochrane Microbiol 2019;57.
Database Syst Rev 2018;8:Cd012768. [33] Wang S. Diagnostic Value of Xpert MTB /RIF Ultra in Smear Negative
[12] WHO Guidelines Approved by the Guidelines Review CommitteeAuto- Pulmonary Tuberculosis, Tuberculous Pleurisy and Tuberculous Men-
mated Real-Time Nucleic Acid Amplification Technology for Rapid and ingitis. 2019;Beijing Tuberculosis Chest Cancer Institute, Chinese.
Simultaneous Detection of Tuberculosis and Rifampicin Resistance: [34] Menichini M, Lari N, Lupetti A, Rindi L. Evaluation of Xpert MTB/RIF
Xpert MTB/RIF Assay for the Diagnosis of Pulmonary and Extrap- Ultra assay for rapid diagnosis of pulmonary and extra-pulmonary
ulmonary TB in Adults and Children: Policy Update. Geneva: World tuberculosis in an Italian center. Eur J Clin Microbiol Infect Dis
Health Organization; 2013. 2020;39:1597–600.
[13] Huo ZY, Peng L. Is Xpert MTB/RIF appropriate for diagnosing [35] Wu X, Tan G, Gao R, et al. Assessment of the Xpert MTB/RIF Ultra
tuberculous pleurisy with pleural fluid samples? A systematic review. assay on rapid diagnosis of extrapulmonary tuberculosis. Int J Infect Dis
BMC Infect Dis 2018;18:284. 2019;81:91–6.
[14] Bahr NC, Marais S, Caws M, et al. GeneXpert MTB/Rif to diagnose [36] Arshad A, Dayal S, Gadhe R, et al. Analysis of tuberculosis meningitis
tuberculous meningitis: perhaps the first test but not the last. Clin Infect pathogenesis, diagnosis, and treatment. J Clin Med 2020;9.
Dis 2016;62:1133–5. [37] Bahr NC, Meintjes G, Boulware DR. Inadequate diagnostics: the case to
[15] Opota O, Mazza-Stalder J, Greub G, Jaton K. The rapid molecular test move beyond the bacilli for detection of meningitis due to Mycobacteri-
Xpert MTB/RIF ultra: towards improved tuberculosis diagnosis and um tuberculosis. J Med Microbiol 2019;68:755–60.
rifampicin resistance detection. Clin Microbiol Infect 2019;25:1370–6. [38] Jiang J, Yang J, Shi Y, et al. Head-to-head comparison of the diagnostic
[16] Chakravorty S, Simmons AM, Rowneki M, et al. The new Xpert MTB/ accuracy of Xpert MTB/RIF and Xpert MTB/RIF Ultra for tuberculosis:
RIF Ultra: improving detection of Mycobacterium tuberculosis and a meta-analysis. Infect Dis (Lond) 2020;1–13.
resistance to rifampin in an assay suitable for point-of-care testing. mBio [39] Chen YZ, Sun LC, Wen YH, et al. Pooled analysis of the Xpert MTB/RIF
2017;29:8. assay for diagnosing tuberculous meningitis. Biosci Rep 2020;31:40.