Tuberculosis (TB) is one of the world’s most important infectious causes of morbidity and mortality among adults. An estimated one-third of the world’s population is infected with TB. In 2010, there were 8.8 million new, and 12.0 million prevalent cases of TB; 1.1 million deaths occurred among HIV-uninfected people and an additional 0.35 million deaths among HIV-infected people (WHO 2011a). Of the total new TB cases, an estimated 12% to 14% occurred among people living with HIV. Approximately one million TB cases occur in children younger than 15 years old annually (Marais 2010a). In 2010, the three countries with the highest number of new TB cases were India, China, and South Africa (WHO 2011a). In 2010, there were an estimated 650,000 cases of multidrug-resistant TB (MDR-TB), defined as resistance to at least isoniazid and rifampicin, the two most powerful first-line anti-TB drugs (WHO 2011a). TB is a treatable and curable disease; according to the World Health Organization (WHO), up to six million lives were saved with the expansion of Directly Observed Treatment Short-Course (DOTS; the basic package that underpins the Stop TB Strategy) (WHO 2010a). Early and accurate diagnosis and effective treatment is the cornerstone of TB care and control (Dye 2010). A basic tenet of early and accurate diagnosis is the identification of smear-negative disease (associated with HIV infection) and MDR-TB (WHO 2010b).
In 2011, poor diagnosis remains an obstacle to TB care and control, partly because TB diagnosis continues to rely on century-old tests. In most TB endemic countries, TB diagnosis is based on direct Ziehl-Neelsen sputum microscopy or chest radiography, tests with known shortcomings, especially in people with HIV infection (Harries 2004; Perkins 2007; Steingart 2006). Mycobacterial cultures are time consuming and have biosafety and training requirements, and culture capacity is limited in underserved and remote areas. Nucleic acid-amplification (NAA) tests, including polymerase chain reaction (PCR) tests, can rapidly detect Mycobacterium tuberculosis (M. tuberculosis), but studies of conventional NAA tests have shown relatively low sensitivity in sputum smear-negative patients (Flores 2005; Greco 2006; Ling 2008) and these assays can only be performed in laboratories with specialized equipment and expertise. The disappointing performance of these tools is compounded by the lack of access to health services with diagnostic laboratories (WHO 2010a). A substantial proportion (˜35%) of the estimated TB cases worldwide remain undiagnosed, including a staggering proportion (˜85%) of patients with MDR-TB (WHO 2011a). In addition, because of delays in diagnosis, TB patients have often been symptomatic for months, leading to increased illness and mortality, secondary drug resistance, and ongoing transmission. Simple and rapid diagnostic tests at the point of treatment in high-TB burden countries are urgently needed.
Target condition being diagnosed
TB is an airborne disease caused by the bacterium, M. tuberculosis, and is spread primarily by droplet nuclei expelled by a person who has infectious active TB. Although TB most commonly affects the lungs, any organ or tissue may be involved. Signs and symptoms of pulmonary TB include cough for at least two weeks, fever, chills, night sweats, weight loss, haemoptysis (coughing up blood), and fatigue; symptoms of extrapulmonary TB depend on the site of disease. From 1970 to 1986, wide-scale international multicentre randomized controlled trials conducted by the British Medical Research Council and its collaborators established the pivotal role of rifampicin in TB treatment leading to modern short-course TB therapy (Fox 1999). TB, even when resistant to rifampicin, can be cured; however,to be effective, TB treatment regimens must contain multiple drugs to which the organisms are susceptible. International guidelines for the treatment of TB are issued by a WHO Expert Group and are regularly updated. The current WHO treatment guidelines are based on evidence assessed according to the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach for developing health care recommendations (Guyatt 2008; WHO 2009).
The Xpert MTB/RIF© test is an automated PCR test utilizing the GeneXpert© platform. Xpert MTB/RIF can detect TB as well as rifampicin resistance in less than two hours with minimal hands-on technical time. Xpert MTB/RIF is considered to be ground-breaking because the test comes close to meeting the niche for a TB point-of-care test. Unlike conventional NAA tests, this test is unique because all steps involved in the PCR are completely automated and self-contained, allowing the technology to be taken out of the laboratory, into the clinic setting, where it can be used nearer to the patient (Small 2011).
Although, Xpert MTB/RIF provides testing for both M. tuberculosis and rifampicin resistance, it is really only one test. The test uses five molecular beacons. Molecular beacons are nucleic acid probes that recognize and report the presence or absence of the normal, susceptible, 'wild type' sequence of the RNA polymerase (rpoB) gene of M. tuberculosis. When a beacon fluoresces or 'lights up', this indicates the presence of the gene sequence which is characteristic of rifampicin-susceptible M. tuberculosis. The number of positive beacons allows the test to distinguish among the following results: 'No TB detected' (none of the five beacons is positive); 'TB detected, rifampicin resistance detected' (from two to four beacons are positive); 'TB detected, no rifampicin resistance detected' (five beacons are positive); and an 'invalid result' (one beacon is positive). Xpert MTB/RIF occupies one physical unit and uses the same sputum sample to provide results for both detection of M. tuberculosis and rifampicin resistance. One cannot switch rifampicin resistance testing off and only do TB testing, although it is possible for the laboratory to omit results for rifampicin resistance testing when reporting to the healthcare provider.
The physical unit for Xpert MTB/RIF has two main components: 1) a plastic cartridge containing liquid sample processing buffers and PCR buffers and reagents; and 2) a device that controls the fluidics inside the cartridge and performs real-time PCR analysis (Helb 2009). The Xpert MTB/RIF assay amplifies a sequence of the rpoB gene specific to members of M. tuberculosis complex. In greater than 95% of M. tuberculosis clinical isolates, resistance to rifampicin has been associated with single amino acid alterations in a limited region (81 base pairs (bp) also called rifampicin-resistance determining region of the rpoB gene (Telenti 1993). Rifampicin-susceptible isolates show no mutations in this region. With Xpert MTB/RIF, the limit of detection for M. tuberculosis DNA was found to be 4.5 genomes per PCR reaction (95% confidence interval (CI) 3.3 to 9.7) and for M. tuberculosis cells in clinical sputum samples, 131 colony-forming units (cfu)/mL (95% CI 106.2 to 176.4) Helb 2009). In investigations using clinical or experimental samples, all 23 81-bp mutants were correctly identified as rifampicin-resistant (Helb 2009).
The most widely used test for TB diagnosis in low- and middle-income countries is Ziehl-Neelsen smear microscopy. Sensitivity is low (from 50 to 60% on average) and variable (from 20 to 80%), and microscopy does not detect smear-negative TB which may account for 24% to 61% of all pulmonary cases in HIV-infected individuals (Steingart 2006; Getahun 2007).
Chest radiography for TB diagnosis is limited by high inter- and intra-observer differences in reporting of radiographs, and absence of typical findings in people living with HIV who have advanced immunosuppression (Harries 2004).
Improved diagnostic tests such as mycobacterial culture and NAA tests are available in high-income countries, but are often too expensive and complex for routine use by TB control programmes in resource-limited TB-endemic settings. Lack of access to diagnostic services in resource-limited settings presents an additional barrier to using these tests.
Existing diagnostic tools for TB are not accurate, or cannot be done rapidly at the point-of-care; therefore, there is an urgent need for a diagnostic tool that is accurate, simple to operate, and can be placed nearer to the patient in facilities such as health posts and microscopy centres. While rapid, simple point-of-care tests exist for infections like HIV and malaria, there is currently no such option for TB. Such a tool would have significant impact on TB control through interruption of transmission and potentially earlier, more efficient diagnosis of TB, including the detection of smear-negative disease and MDR-TB. In December 2010, WHO announced its endorsement of the Xpert MTB/RIF test (WHO 2011). The WHO decision to endorse this test was based on a large multicentre study (Boehme 2010) and other preliminary data, reviewed by an Expert Group (WHO 2010b). These studies showed that Xpert MTB/RIF has a comparable level of diagnostic performance to mycobacterial culture, even in resource-limited settings. A subsequent implementation study (Boehme 2011) demonstrated high sensitivity of Xpert MTB/RIF in smear-negative patients (a concern in HIV co-infected individuals, and in children) (Mugusi 2006; Perkins 2006; Perkins 2007), while conventional NAA tests have low sensitivity in smear-negative patients (Greco 2006; Ling 2008). The WHO now recommends that Xpert MTB/RIF should be used as the initial diagnostic test in individuals suspected of MDR-TB or HIV-associated TB (WHO 2010b). Furthermore, Xpert may be used as an add-on test to microscopy in settings where MDR-TB or HIV, or both, are of lesser concern, especially in smear-negative patients. While several studies of Xpert MTB/RIF have shown excellent performance, this assay has not been fully validated in all settings. Since the WHO’s endorsement, several new studies have been published on Xpert MTB/RIF, and several others are expected to be published shortly. At the time of this writing, a systematic review on the diagnostic accuracy of this new test has not been published.
Review question A: To obtain summary estimates of the diagnostic accuracy of Xpert MTB/RIF for the diagnosis of pulmonary TB, using solid or liquid culture as a reference standard.
Purpose of index test:
A.1 Xpert MTB/RIF as a replacement test for smear microscopy.
A.2. Xpert MTB/RIF as an add-on test after smear microscopy.
Review question B: To obtain summary estimates of the diagnostic accuracy of Xpert MTB/RIF for detection of rifampicin resistance, using solid or liquid culture as a reference standard (WHO 2008).
Purpose of index test:
B.1 Xpert MTB/RIF as a replacement test for WHO-approved tests for detection of rifampicin resistance.
- To summarize evidence on time to treatment initiation.
- To summarize evidence on time to diagnosis.
Although these secondary outcomes will not be systematically reviewed, we will extract data when present in the included accuracy studies.
Investigation of sources of heterogeneity
We will investigate whether HIV-infection status; sputum smear status; country income status; setting; or storage conditions of specimen can explain the expected heterogeneity in estimates of test sensitivity and specificity.
Criteria for considering studies for this review
Types of studies
We will include primary studies that compared the results of the index test with the reference standard. Diagnostic accuracy studies are typically cross sectional in design. However, we will also include randomized controlled trials and cohort studies.
Only studies that report data from which we can extract true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN) will be included.
We will exclude case-control studies. We will exclude studies that are reported in abstracts.
Included participants for review question A.1 will be adults (15 years and older) suspected of having pulmonary TB or MDR-TB, from all settings and all countries.
Included participants for review question A.2 will be adults suspected of having pulmonary TB or MDR-TB that are determined to be microscopy smear negative. In this diagnostic strategy, smear microscopy may be performed prior to, or concurrently with, Xpert MTB/RIF test.
Included participants for review question B will be adults suspected of having pulmonary TB or MDR-TB, from all settings and all countries.
Patients suspected of having MDR-TB may include patients with a history of TB, patients on TB treatment for pulmonary TB without sputum conversion, and symptomatic contacts of patients with known MDR-TB.
Three categories of participants will be classified:
- Definite TB - culture positive.
- Probable TB (clinically considered as TB, culture negative).
- Non-TB (clinically not considered as non-TB, culture negative).
We will include studies that evaluated the Xpert MTB/RIF assay.
For review question A, we will include studies that evaluated smear microscopy as a comparator.
Review question A: Active pulmonary TB.
Review question B: Rifampicin resistance.
Review question A: The reference standard for the diagnosis of active pulmonary TB is solid or liquid mycobacterial culture.
Review question B: The reference standard for detection of rifampicin resistance is 1) solid culture, or 2) a commercial liquid culture system (BACTEC 460, MGIT 960, and MGIT Manual System, Becton Dickinson, USA; BacT/ALERT MP, Biomerieux, France; VersaTREK, Trek Diagnostic Systems, USA) (Canetti 1963; Laszlo 1997; WHO 2008).
Search methods for identification of studies
We will attempt to identify all relevant studies regardless of language or publication status (published, unpublished, in press, and ongoing).
To identify all relevant studies, we will search the following databases using the search terms and strategy described in Appendix 1: Cochrane Infectious Diseases Group Specialized Register; MEDLINE; EMBASE; ISI Web of Knowledge; MEDION; LILACS; BIOSIS; and SCOPUS. We will also search the metaRegister of Controlled Trials (mRCT) and the search portal of the WHO International Clinical Trials Registry Platform (www.who.int/trialsearch), to identify ongoing trials.
Searching other resources
We will review reference lists of included articles and any relevant review articles identified through the above methods. We will contact the test manufacturer (Cepheid Inc.) to identify unpublished studies. We will also handsearch WHO reports on Xpert MTB/RIF. We will contact researchers at the Foundation for Innovative New Diagnostics (FIND), members of the Stop TB Partnership's New Diagnostics Working Group, and other experts in the field of TB diagnostics for information on ongoing or unpublished studies.
Data collection and analysis
Selection of studies
Two independent reviewers (HS and KRS) will first look at titles and abstracts identified by electronic literature searching to identify potentially eligible studies. Any citation identified by either reviewer during this screen (screen 1) will be selected for full-text review. Two independent reviewers (HS and KRS) will then review full-text papers (screen 2) for study eligibility using the predefined inclusion and exclusion criteria. In screen 2, any discrepancies will be resolved by discussion between the reviewers, or if they are unable to resolve, by decision of a third reviewer (MP). A list of excluded studies and their reasons for exclusion will be maintained.
Data extraction and management
Two independent reviewers (HS and KRS) will extract a set of data from each study using a piloted data extraction form. Based on the pilot, the extraction form will be finalized. Two independent reviewers (HS and KRS) will then extract data on the following characteristics:
- Details of study: first author; publication year; case country of residence; World Bank country income status; setting (outpatient, inpatient, laboratory); study design; manner of participant selection; number participants enrolled; number participants for whom results available; industry sponsorship.
- Characteristics of participants: HIV status; smear status; history of TB.
- Target conditions: review question A: pulmonary TB; review question B: rifampicin resistance.
- Reference standards: review question A: culture type (solid or liquid); percentage of contaminated cultures; review question B: the name and manufacturer of the reference standard.
- Details of index test: software version of test.
- Details of comparator: type of microscopy: light or fluorescence; type of smear: direct or concentrated; number of smears used to determine smear positivity.
- Details of sputum specimen: type (such as expectorated sputum, induced sputum, bronchoalveolar lavage); condition (fresh or frozen); definition of a positive smear.
- Details of outcomes: the number of true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN); number of missing or unavailable test results.
Time to treatment initiation is defined as the time from specimen collection until patient starts treatment.
Time to diagnosis is defined as the time from specimen collection until there is an available TB result in lab or clinic, if the test was performed in a clinic.
Country income status will be classified as low/middle-income or high-income, according to the World Bank List of Economies. Country income status will serve as a surrogate indicator for TB incidence.
Review question A: Additional data about (a) proportion of indeterminate results, and (b) discordant results between Xpert MTB/RIF and the reference standard (Xpert positive/culture negative results) will be recorded.
Review question A: other possible test results are invalid, error, or no result. These results will be combined and considered as 'indeterminate'.
The proportion of indeterminate results will be the number of tests classified as 'invalid', error', or 'no result' divided by the total number of tests performed.
Review question B: RIF indeterminate means that the M. tuberculosis concentration was very low and resistance could not be determined. Proportion of indeterminate results will be the number of tests classified as 'indeterminate' divided by the total number of tests performed.
Culture contamination will be defined as 1) growth of non-TB bacteria or fungi or 2) cross-contamination as determined after a standard laboratory evaluation. Percentage of contaminated cultures will be the (number of contaminated cultures/total number of cultures performed) x 100.
We will contact authors of primary studies for missing data or clarifications. All data will be entered into a database manager. A draft data extraction form is included in Appendix 2..
Assessment of methodological quality
Two reviewers will independently assess study quality with the modified version of Quality Assessment of Diagnostic Accuracy Studies (QUADAS) (Reitsma 2009). Items 1 to 11 from the QUADAS list will be scored as yes, no, or unclear. Disagreements will be resolved by discussion between the reviewers or by a third reviewer (MP). Results will be described in the text and presented graphically. Appendix 3 describes the criteria that need to be met for each study to be rated as yes, no, or unclear for each of the QUADAS items.
Statistical analysis and data synthesis
The first step in data analysis will be a descriptive analysis of the results of the primary studies. For both review question A and review question B, the results will be based on categorical (binary) test results. For both review questions, the index test results are automatically generated and the user is provided with a printable test result. Examples are:
1. MTB detected; RIF resistance not detected.
2. MTB detected; RIF resistance detected.
3. MTB detected; RIF resistance indeterminate.
4. MTB not detected; RIF not detected.
Descriptive analysis will be performed using Stata 2009 and key study characteristics will be displayed in tables. For each study, sensitivity and specificity of the test along with the 95% CI will be calculated using exact methods, and forest plots generated using Review Manager 5 (RevMan). The primary analysis will be performed for participants with definite TB (culture positive) and non-TB (culture negative). Secondary analyses will estimate accuracy using definite TB and probable-TB patients combined for sensitivity calculations, and non-TB and probable TB patients combined for specificity calculations.
For review question B, we will sort the studies in RevMan by "year of study" to look for a trend with time, since software and cartridge changes have been made to improve the specificity for rifampicin detection.
Where sufficient data are available, meta-analyses will be carried out to estimate sensitivity and specificity of the index test (primary objective). The exact method used, either bivariate or HSROC, will depend on data provided by the included studies. For review question A, detection of M. tuberculosis, since the test uses a common threshold for a positive result, we will use the bivariate random effects regression model (Macaskill 2010; Reitsma 2005). The model will be estimated using a Bayesian approach with nonsubjective prior distributions and implemented using WinBUGS (Version 1.4.1) (Spiegelhalter 2004).
For review question B, detection of rifampicin resistance, if most studies report one or two and the same threshold (the same amplification cycle used for detection of rifampicin resistance), meta-analysis can also be done by using the bivariate method. However, where the studies report several different thresholds, it may be more appropriate to use the Hierarchical Summary Receiver Operating Characteristic (HSROC) model (Rutter 2001; Macaskill 2010).
We will also compare the sensitivity and specificity of Xpert MTB/RIF with that of smear microcopy. For smear microscopy, the results will be based on categorical results, either positive or negative. If most studies report one and the same threshold (for example, smear positive is greater than or equal to one acid-fast bacillus), we will pool results using the bivariate method, but if studies report several different thresholds, it may be more appropriate to use the HSROC model (Macaskill 2010; Rutter 2001).
Our approach to Xpert MTB/RIF used as an add-on test to smear microscopy will be as follows: in patients found to be smear negative/culture positive for pulmonary TB, we will consider sensitivity and specificity estimates to be a proxy for Xpert MTB/RIF used as an add-on test to microscopy, even if the primary study objective was not explicit for this outcome and both tests were run concurrently.
Approach to indeterminate index test results
For both review questions A and B, we will exclude indeterminate index test results from the main analysis.
For both review questions A and B, we will determine pooled estimates and the predicted interval for indeterminate index test results.
For both review questions A and B, where data are available, we will perform sensitivity analyses to determine the potential impact of indeterminate index test results considered to be false/true positives or false/true negatives. In the discussion section, we will discuss the consequences of an indeterminate index test result considered to be a true negative result (may lead to missed/delayed diagnosis, with potential for increased morbidity, mortality, and TB transmission), or considered to be true positive result (may lead to unnecessary treatment with adverse effects and increased anxiety).
For both review questions A and B, we will first stratify studies by country income status (low- and middle-income versus high-income). We expect the majority of studies to report TP, TN, FP, and FN stratified by smear status and/or HIV status. For these subgroups, we will compare summary estimates of accuracy in HIV infected and uninfected subgroups, and smear positive and smear negative subgroups. If there are sufficient studies, we will also determine summary accuracy estimates for smear-negative, HIV-infected versus smear-positive, HIV-infected subgroups.
Investigations of heterogeneity
Heterogeneity between diagnostic accuracy studies is to be expected (Harbord 2007). Initially, heterogeneity will be addressed by pre-specifying subgroups. We will assess heterogeneity by visual inspection of forest plots. We will further investigate potential sources of heterogeneity by adding covariates in the models. All covariates will be study level, dichotomous.
Review question A
Condition of specimen (categorical covariate): 1. fresh; 2. frozen.
TB prevalence (categorical covariate): 1. low; 2. high.
Setting in which Xpert was used (categorical covariate): 1. clinical (outpatient or inpatient); 2. laboratory.
Review question B
Software version (amplification cycle thresholds of the test, dichotomous): 1. Versions 1 and 2 (cycle threshold 3.5); 2. Versions 3 (cycle threshold 5.0) and 4.
MDR-TB prevalence (categorical covariate): 1. low; 2. high.
If sufficient studies are available, we will perform sensitivity analysis for QUADAS items to explore whether the results we found are robust with respect to methodological quality of the studies.
Assessment of reporting bias
Data included in this review will not allow for formal assessment of publication bias using methods such as funnel plots or regression tests because such techniques have not been found to be helpful for diagnostic test accuracy studies (Tatsioni 2005 Macaskill 2010). However, being a new test for which there is going to be considerable attention and scrutiny, we believe reporting bias will be minimal.
We will summarize, if feasible, evidence on outcomes important to Xpert users, including time to diagnosis and time to treatment initiation. We will also summarize hands-on time for specimen processing and work-flow, instrument ease-of-use, and user satisfaction. This information will be excluded from the formal protocol. We will address these outcomes in a section of the discussion and present summary data in additional tables. In addition, if data are available, we will prepare a qualitative description in the discussion section of Xpert positive/culture negative (false positive) patients followed longitudinally, and report the number and percent of these patients who become culture positive during follow up.
We are grateful to Vittoria Lutje, Liverpool School of Tropical Medicine, for help with the search strategy. The editorial base for the Cochrane Infectious Disease Group is funded by the Department for International Development (DFID), UK, for the benefit of low- and middle-income countries.
Appendix 1. Detailed search strategies
1. Xpert*.tw .
4. near* patient ti, ab
5. or 1-4
7. exp Tuberculosis, Pulmonary/
8. exp Tuberculosis, Multidrug-Resistant/
9. Mycobacterium tuberculosis/
13. 5 and 12
14. limit 13 to yr="2007 -Current"
This is the preliminary search strategy for MEDLINE. It will adapted for other electronic databases. All search strategies will be reported in full in the final version of the review.
Appendix 2. Data extraction form
* Clinical TB (describe criteria used by authors)
* Follow-up test included, circle all that apply
** Repeat culture, DNA sequencing, GenoType® MTBDRplus test, other, describe
Appendix 3. Assessment of methodological quality
Protocol first published: Issue 1, 2012
Contributions of authors
MP had the original idea for the review. The protocol was written by KRS and MP, with input from HS and ND. The search strategy was written by LAK. HS and KRS will review articles for relevance, and extract and analyse data. HS, MP, ND, CCB, and KRS will interpret data and write the manuscript.
Declarations of interest
KRS serves as Co-ordinator of the Evidence Synthesis and Policy Subgroup of Stop TB Partnership’s New Diagnostics Working Group. MP is a recipient of a New Investigator Award from the Canadian Institutes of Health Research (CIHR). MP serves as an external consultant for the Bill & Melinda Gates Foundation. CCB is employed by the Foundation for Innovative New Diagnostics FIND) and has conducted studies and published on Xpert MTB/RIF as part of a collaborative project between FIND, a Swiss non-profit, Cepheid, a US company, and academic partners. The product developed through this partnership was developed under a contract that obligated FIND to pay for development costs and trial costs and that obligated Cepheid to make the test available at specified preferential pricing to the public sector in developing countries. The authors have no financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the protocol apart from those disclosed.