Accuracy of cobas MTB and MTB-RIF/INH for Detection of Mycobacterium tuberculosis and Drug Resistance

This study evaluated the performance of cobas MTB and cobas MTB-RIF/INH for the diagnosis of tuberculosis and detection of rifampicin (RIF) and isoniazid (INH) resistance. Adults presenting with pulmonary tuberculosis symptoms were recruited in South Africa, Moldova, and India. Performance of cobas MTB was assessed against culture, whereas cobas MTB-RIF/INH was assessed using phenotypic drug susceptibility testing and whole-genome sequencing as composite reference standards. Xpert MTB/RIF (Xpert) or Xpert MTB/RIF Ultra (Ultra) was used as a comparator. The overall sensitivity and specificity of cobas MTB were 95% (95% CI, 93%–96%) and 96% (95% CI, 95%–97%). Among smear-negatives, the sensitivity of cobas MTB was 75% (95% CI, 66%–83%). Among participants tested with both cobas MTB and Xpert, sensitivity was 96% (95% CI, 94%–97%) for cobas MTB and 95% (95% CI, 93%–97%) for Xpert. Among participants tested with both cobas MTB and Ultra, sensitivity was 88% (95% CI, 81%–92%) for cobas MTB and 89% (95% CI, 83%–93%) for Ultra. Sensitivity and specificity of cobas MTB-RIF/INH for RIF and INH detection were 90% (95% CI, 84%–94%) and 100% (95% CI, 99%–100%), and 89% (95% CI, 84%–93%) and 99.5% (95% CI, 98%–100%), respectively. The cobas MTB and cobas MTB-RIF/INH assays exhibited high performance in a diverse population and present a suitable option for molecular detection of tuberculosis and RIF and INH resistance. This study evaluated the performance of cobas MTB and cobas MTB-RIF/INH for the diagnosis of tuberculosis and detection of rifampicin (RIF) and isoniazid (INH) resistance. Adults presenting with pulmonary tuberculosis symptoms were recruited in South Africa, Moldova, and India. Performance of cobas MTB was assessed against culture, whereas cobas MTB-RIF/INH was assessed using phenotypic drug susceptibility testing and whole-genome sequencing as composite reference standards. Xpert MTB/RIF (Xpert) or Xpert MTB/RIF Ultra (Ultra) was used as a comparator. The overall sensitivity and specificity of cobas MTB were 95% (95% CI, 93%–96%) and 96% (95% CI, 95%–97%). Among smear-negatives, the sensitivity of cobas MTB was 75% (95% CI, 66%–83%). Among participants tested with both cobas MTB and Xpert, sensitivity was 96% (95% CI, 94%–97%) for cobas MTB and 95% (95% CI, 93%–97%) for Xpert. Among participants tested with both cobas MTB and Ultra, sensitivity was 88% (95% CI, 81%–92%) for cobas MTB and 89% (95% CI, 83%–93%) for Ultra. Sensitivity and specificity of cobas MTB-RIF/INH for RIF and INH detection were 90% (95% CI, 84%–94%) and 100% (95% CI, 99%–100%), and 89% (95% CI, 84%–93%) and 99.5% (95% CI, 98%–100%), respectively. The cobas MTB and cobas MTB-RIF/INH assays exhibited high performance in a diverse population and present a suitable option for molecular detection of tuberculosis and RIF and INH resistance. Drug-resistant tuberculosis (TB) continues to threaten global TB control.1WHO: Global Tuberculosis Report 2023. World Health Organization, Geneva, Switzerland2023Google Scholar From 2018 to 2022, 825,000 individuals globally with rifampicin (RIF)–resistant or multi–drug-resistant TB [resistant to at least RIF and isoniazid (INH)] have been reported to have initiated treatment, which is only 55% of the World Health Organization (WHO) target of 1.5 million for the period.1WHO: Global Tuberculosis Report 2023. World Health Organization, Geneva, Switzerland2023Google Scholar Delays or failures in the detection of drug-resistant TB lead to increased risk of patient mismanagement, drug resistance amplification, and ongoing disease transmission.2Georghiou S.B. Schumacher S.G. Rodwell T.C. Colman R.E. Miotto P. Gilpin C. Ismail N. Rodrigues C. Warren R. Weyer K. Zignol M. Arafah S. Cirillo D.M. Denkinger C.M. Guidance for studies evaluating the accuracy of rapid tuberculosis drug-susceptibility tests.J Infect Dis. 2019; 220: S126-S135Crossref PubMed Scopus (10) Google Scholar Improving accuracy and speed of diagnosis is key to improving patient outcomes and preventing spread of TB and drug-resistant TB. In 2021, the WHO endorsed moderate complexity nucleic acid amplification tests for initial diagnosis of TB and detection of RIF and INH resistance.3WHO: WHO Consolidated Guidelines on Tuberculosis: Module 3: Diagnosis: Rapid Diagnostics for Tuberculosis Detection, 2021 Update. World Health Organization, Geneva, Switzerland2021Google Scholar Tests included in this class are largely fully automated and less complex than conventional culture-based methods [such as phenotypic drug susceptibility testing (pDST) and whole-genome sequencing (WGS)] and offer potential for rapid high-throughput testing. With the addition of INH susceptibility detection, they overcome the limitations of Xpert MTB/RIF (Xpert; Cepheid, Sunnyvale, CA) and Xpert MTB/RIF Ultra (Ultra; Cepheid), which only detect RIF susceptibility. The cobas MTB and MTB-RIF/INH assays (Roche Diagnostics International AG, Rotkreuz, Switzerland) have been endorsed by WHO in the moderate complexity nucleic acid amplification test class.3WHO: WHO Consolidated Guidelines on Tuberculosis: Module 3: Diagnosis: Rapid Diagnostics for Tuberculosis Detection, 2021 Update. World Health Organization, Geneva, Switzerland2021Google Scholar cobas MTB is an automated, qualitative nucleic acid amplification test for Mycobacterium tuberculosis complex (MTBC) detection from raw, digested, or decontaminated sputum and bronchoalveolar lavage samples. cobas MTB-RIF/INH is a reflex assay for the detection of RIF resistance-associated mutations in rpoB and INH resistance-associated mutations in katG and the inhA promoter region. Both cobas assays are performed on the cobas 6800 or 8800 closed platforms. In previous studies, cobas MTB exhibited high sensitivity against a culture reference standard,4Scott L. David A. Govender L. Furrer J. Rakgokong M. Waja Z. Martinson N. Elsenberg G. Marlowe E. Stevens W. Performance of the Roche cobas® MTB assay for the molecular diagnosis of pulmonary tuberculosis in a high HIV burden setting.J Mol Diagnostics. 2020; 10: 1225-1237Abstract Full Text Full Text PDF Scopus (8) Google Scholar,5Nadarajan D. Hillemann D. Kamara R. Foray L. Conteh O.S. Merker M. Niemann S. Lau J. Njoya M. Kranzer K. Somoskovi A. Maurer F.P. Evaluation of the roche cobas MTB and MTB-RIF/INH assays in samples from Germany and Sierra Leone.J Clin Microbiol. 2021; 59e02983-20Crossref PubMed Scopus (5) Google Scholar and performance of cobas MTB-RIF/INH was sufficient for WHO endorsement.5Nadarajan D. Hillemann D. Kamara R. Foray L. Conteh O.S. Merker M. Niemann S. Lau J. Njoya M. Kranzer K. Somoskovi A. Maurer F.P. Evaluation of the roche cobas MTB and MTB-RIF/INH assays in samples from Germany and Sierra Leone.J Clin Microbiol. 2021; 59e02983-20Crossref PubMed Scopus (5) Google Scholar However, further evaluation of these assays in geographically diverse settings is required to inform global and national policy decision-making. The objective of the current study was to evaluate the clinical diagnostic accuracy of cobas MTB and cobas MTB-RIF/INH in settings of intended use, including those with a high prevalence of individuals living with HIV. This multicenter, cross-sectional, clinical diagnostic accuracy study evaluated the performance of cobas MTB for MTBC detection and cobas MTB-RIF/INH for RIF and INH resistance detection (NCT04147676; https://www.clinicaltrials.gov, last accessed March 14, 2024). Xpert and Ultra were included in the evaluation as comparator tests as both assays are WHO endorsed and implemented for routine patient care in their intended setting. Xpert was used in Moldova and India, whereas Ultra was used in South Africa. The reference standard for MTBC detection was liquid culture confirmed by a WHO-recommended rapid or molecular test. The reference standard for RIF and INH resistance was a composite of pDST and WGS. The study was approved by relevant institutional review boards and independent ethics committees. All participants provided written informed consent. Participants were prospectively recruited sequentially between May 17, 2019, and November 30, 2021 at three sites: University of the Witwatersrand (Johannesburg, South Africa); Phthisiopneumology Institute “Chiril Draganiuc” (Chisinau, Moldova); and Indian Council of Medical Research–Regional Medical Research Center (Bhubaneswar, India). Adults (≥18 years of age) presenting with symptoms or characteristics consistent with pulmonary TB, in outpatient clinic or inpatient hospital settings, were eligible for inclusion. Criteria indicating clinical suspicion of pulmonary TB were cough of ≥2 weeks’ duration (≥1 week for HIV co-infected individuals in South Africa) and one or more of the following: fever, malaise, recent weight loss, night sweats, hemoptysis, chest pain, loss of appetite, or contact with an individual with active TB. Participants were required to provide sputum samples of adequate volume (two ≥3-mL samples in Moldova and India, and four ≥2-mL samples in South Africa) over one or two visits, depending on the site protocol. Receipt of any dose of TB treatment before 6 months of enrollment resulted in exclusion. Retrospectively collected frozen raw sputum samples from the FIND Specimen Bank (https://www.finddx.org/what-we-do/cross-cutting-workstreams/biobank-services, last accessed March 14, 2024) were also used, to achieve accurate performance estimates through a diverse collection of resistance-conferring mutations. To that end, a random selection of samples that tested positive for RIF- or INH-resistant TB in previous phenotypic and genotypic investigations was included in the study. Evaluations were performed at reference laboratories at the recruitment sites in South Africa and India. Samples from Moldova were delivered weekly to the WHO Supranational Reference Center for Mycobacteria (Borstel, Germany), for testing. In India and Germany, the two samples from each participant were randomized on arrival at the laboratory. The first sample was homogenized and used for direct cobas MTB, cobas MTB-RIF/INH, and Xpert testing, acid-fast bacilli smear microscopy, and N-acetyl-l-cysteine/sodium hydroxide decontamination before BACTEC Mycobacteria Growth Indicator Tube 960 (Becton Dickinson, Franklin Lakes, NJ) (MGIT; liquid) or Löwenstein-Jensen (solid) culture. The second sample was N-acetyl-l-cysteine/sodium hydroxide decontaminated before all the aforementioned tests (Figure 1A). In South Africa, the two samples collected on day 1 were not randomized (Figure 1B). Sample 1 was tested with Ultra per national standard of care, whereas sample 2 was homogenized for direct cobas MTB, Ultra, and smear testing. Samples 3 and 4, collected on day 2, were randomized and decontaminated before all testing. Both samples 3 and 4 were used for smear microscopy and MGIT; sample 3 was also used for cobas MTB and Ultra testing, and sample 4 was used for GenoType MTBDRplus (Bruker Hain Lifescience, Nehren, Germany) molecular testing per national standard of care. At all cobas MTB was performed as a reflex test on all cobas samples. To MTBC all positive were further tested with the TB assay or GenoType pDST was performed using the (Becton with of 1 and for RIF and respectively. was performed at by of using an and was using the from the laboratories in Germany and South Africa. in the inhA and were and per the WHO mutations second of in Mycobacterium Tuberculosis and with World Health Organization, Geneva, Switzerland2023Google Scholar and reference standard were to the of the and sufficient sample volume was cobas MTB and cobas MTB-RIF/INH testing was performed for the confirmed culture positive and cobas MTB culture and cobas MTB pDST and cobas MTB-RIF/INH or and pDST and cobas MTB-RIF/INH or pDST was also performed for and test were not included in the The outcomes were clinical sensitivity and specificity of cobas MTB for MTBC detection, and of cobas MTB-RIF/INH for RIF and INH resistance detection, with the reference standards. estimates and 95% were using the Both outcomes were also assessed for Xpert and Ultra, and were evaluated overall and by smear Both sample methods were for the of these Sample 2 from South Africa was from overall for cobas MTB as reference standard was was included in of Ultra performance culture from samples 3 or 4 as the reference the in participant on HIV recruitment TB and sample The and of for cobas MTB and cobas MTB-RIF/INH as drug resistance MTB were also sample was with an and INH-resistant samples from the FIND Specimen P. E. L. C. D. R. Zignol M. Gilpin C. Niemann S. Denkinger C.M. J. Warren D. J. S. S. Rodrigues C. D.M. M. D. L. C. K. R. R. M. Cirillo D.M. M. Rodwell T.C. for the between mutations and phenotypic drug resistance in Mycobacterium J. PubMed Scopus Google Scholar The sample was with the of a prevalence of for TB, and for TB all TB, using the by in Google Scholar in with at and at The population was as all participants provided written informed and any test The population participants were to and those with or or reference standard provided were for and 2 were of sample (Figure 2 and further were of culture, or participants in the RIF and INH resistance detection, a further samples from the FIND Specimen Bank were The was were were HIV and a of TB In were culture positive and were smear were between the three The site in South Africa the at with and for Moldova and India, respectively. The site in Moldova the at Phthisiopneumology Institute “Chiril Draganiuc” is a national TB with also the at with and for South Africa and India, and Africa to to to to of TB, All sample All sample and culture All sample MTB/RIF or Ultra TB All sample MTB All sample TB, in a TB, of sample cobas MTB and culture these were culture positive for were smear positive and culture and were smear and culture all MTBC samples from were as they an reference the of or samples from participants In the of both sample overall sensitivity of cobas MTB was 95% (95% CI, 93%–96%) (Figure Among participants tested with both cobas MTB and Xpert and sensitivity was 96% (95% CI, 94%–97%) for cobas and 95% (95% CI, 93%–97%) for Xpert Among participants tested with both cobas MTB and Ultra sensitivity was 88% (95% CI, 81%–92%) for cobas MTB and 89% (95% CI, 83%–93%) for Ultra In sensitivity of cobas MTB was (95% CI, with four samples as MTBC positive by the Xpert three of the four samples were as MTBC positive by cobas of cobas MTB to Ultra in South Africa estimates (95% CI, 100% (95% CI, In sensitivity of cobas MTB was 75% (95% CI, 66%–83%). From the cobas MTB were by Xpert or Ultra Among participants tested with both cobas MTB and Xpert, sensitivity was (95% CI, for cobas MTB with (95% CI, for Xpert the sensitivity estimates for cobas MTB and Ultra in South Africa were (95% CI, 55% (95% CI, of cobas MTB in the of both sample methods was 96% (95% CI, with for both Xpert and Ultra (Figure samples on cobas MTB, were also positive on Xpert or Ultra. the cobas MTB that were on Xpert or Ultra, all were culture on the second sample from the of were of 3 treatment and 2 treatment and and not treatment and to In overall sensitivity of cobas MTB was (95% CI, and specificity was (95% CI, of sensitivity was to that of Xpert Ultra in South Africa whereas to Xpert is not of the sample of individuals in Moldova and India. recruitment sensitivity and specificity estimates for cobas MTB were for Moldova and India of 96% at both and specificity of (95% CI, and 96% (95% CI, with sensitivity and specificity in South Africa (95% CI, 81%–92%) and (95% CI, performance was cobas MTB testing was performed on raw and decontaminated of the testing site (95% CI, (95% CI, for Moldova and India and (95% CI, 88% (95% CI, for South Africa. Sensitivity of cobas MTB was 95% (95% CI, 93%–97%) in participants with of TB and (95% CI, in participants with a of TB In the of both sample sensitivity of cobas MTB-RIF/INH for RIF resistance detection was 90% (95% CI, with 96% (95% CI, and 100% (95% CI, for Xpert and Ultra, (Figure with cobas MTB-RIF/INH to detect RIF resistance in of samples. these mutations in the rpoB and 1 was by mutations were by All cobas MTB-RIF/INH from the FIND Specimen Bank mutations in the rpoB four are in the WHO as with RIF and three are as with to a that with of the from Moldova the rpoB whereas the sample was by pDST and by WGS. 100% of the rpoB for this and that the between pDST and WGS. of the samples were Xpert RIF indicating that were by the molecular assays testing was on the direct of was as with the cobas MTB of cobas MTB-RIF/INH to detect rpoB mutations was high The specificity of cobas MTB-RIF/INH for RIF resistance detection was 100% (95% CI, The sensitivity of cobas MTB-RIF/INH for detection of INH resistance in the of both sample methods was 89% (95% CI, 84%–93%) (Figure The assay to detect INH resistance in of samples. All from the FIND Specimen Bank were to INH and mutations in katG at acid or in the or promoter of the mutations are not in the WHO mutations of the three from Moldova mutations in katG and The sample was for katG and of this sample by cobas a Both from South Africa were in the assessed by 100% of the katG with that the The two samples were from the of the a Both from India the in the inhA promoter region. The two samples were from the of cobas MTB-RIF/INH to detect katG and promoter mutations was high of cobas MTB-RIF/INH for detection of INH resistance was 99.5% (95% CI, samples from the participant in were INH with cobas The of for cobas MTB was of for raw sputum samples and of for decontaminated for cobas MTB-RIF/INH in raw sputum and decontaminated were of and of for RIF resistance and of and of for INH resistance and RIF resistance from raw sputum and decontaminated and were smear and INH resistance and were smear respectively. were between sites for raw sputum and in India for decontaminated and of with cobas MTB by Sample and sputum tested with cobas cobas MTB cobas MTB decontaminated tested with cobas cobas MTB cobas MTB to N-acetyl-l-cysteine/sodium in a and of with cobas MTB-RIF/INH by Sample and sputum cobas MTB positive tested with RIF RIF RIF INH INH INH decontaminated cobas MTB positive tested with RIF RIF RIF INH INH INH to N-acetyl-l-cysteine/sodium in a to N-acetyl-l-cysteine/sodium to N-acetyl-l-cysteine/sodium This was the first diagnostic accuracy study for cobas MTB and cobas cobas MTB performance to Xpert and Ultra in a diverse clinical with a of The assay WHO target sensitivity and specificity for diagnosis of active for Tuberculosis Report of a World Health Organization, Geneva, for at World Health Organization, Geneva, Switzerland2021Google Scholar and were consistent with previous L. David A. Govender L. Furrer J. Rakgokong M. Waja Z. Martinson N. Elsenberg G. Marlowe E. Stevens W. Performance of the Roche cobas® MTB assay for the molecular diagnosis of pulmonary tuberculosis in a high HIV burden setting.J Mol Diagnostics. 2020; 10: 1225-1237Abstract Full Text Full Text PDF Scopus (8) Google Scholar,5Nadarajan D. Hillemann D. Kamara R. Foray L. Conteh O.S. Merker M. Niemann S. Lau J. Njoya M. Kranzer K. Somoskovi A. Maurer F.P. Evaluation of the roche cobas MTB and MTB-RIF/INH assays in samples from Germany and Sierra Leone.J Clin Microbiol. 2021; 59e02983-20Crossref PubMed Scopus (5) Google Scholar cobas MTB-RIF/INH high accuracy for RIF and INH resistance detection, to a previous D. Hillemann D. Kamara R. Foray L. Conteh O.S. Merker M. Niemann S. Lau J. Njoya M. Kranzer K. Somoskovi A. Maurer F.P. Evaluation of the roche cobas MTB and MTB-RIF/INH assays in samples from Germany and Sierra Leone.J Clin Microbiol. 2021; 59e02983-20Crossref PubMed Scopus (5) Google Scholar with from to on sample and setting. This that the cobas assays suitable to Xpert and Ultra, with the of INH resistance detection and the for high-throughput with previous sensitivity of cobas MTB and the Xpert assays was in L. David A. Govender L. Furrer J. Rakgokong M. Waja Z. Martinson N. Elsenberg G. Marlowe E. Stevens W. Performance of the Roche cobas® MTB assay for the molecular diagnosis of pulmonary tuberculosis in a high HIV burden setting.J Mol Diagnostics. 2020; 10: 1225-1237Abstract Full Text Full Text PDF Scopus (8) Google Scholar,5Nadarajan D. Hillemann D. Kamara R. Foray L. Conteh O.S. Merker M. Niemann S. Lau J. Njoya M. Kranzer K. Somoskovi A. Maurer F.P. Evaluation of the roche cobas MTB and MTB-RIF/INH assays in samples from Germany and Sierra Leone.J Clin Microbiol. 2021; 59e02983-20Crossref PubMed Scopus (5) Google L. W. J. W. of the Xpert MTB/RIF test for the diagnosis of pulmonary tuberculosis in to hospital in South a Infect Dis. Full Text Full Text PDF PubMed Scopus Google S. C. Evaluation of the MTB/RIF assay for rapid diagnosis of tuberculosis and detection of resistance in pulmonary and Clin Microbiol. PubMed Scopus Google M. N. Schumacher S.G. M. Xpert Ultra Xpert MTB/RIF for pulmonary tuberculosis and rifampicin resistance in with pulmonary 2021; Google Schumacher S.G. D. P. S. Cirillo D.M. N. N. Stevens W. L. Rodrigues C. M. L. W. R. R. A. M. Denkinger C.M. study MTB/RIF Ultra for detection of Mycobacterium tuberculosis and rifampicin a diagnostic accuracy Infect Dis. Full Text Full Text PDF PubMed Scopus Google Scholar the WHO of The sensitivity of cobas MTB and both Xpert assays in participants with of in the sputum of individuals living with HIV. In a previous study in South Africa, sensitivity of cobas MTB was by HIV L. David A. Govender L. Furrer J. Rakgokong M. Waja Z. Martinson N. Elsenberg G. Marlowe E. Stevens W. Performance of the Roche cobas® MTB assay for the molecular diagnosis of pulmonary tuberculosis in a high HIV burden setting.J Mol Diagnostics. 2020; 10: 1225-1237Abstract Full Text Full Text PDF Scopus (8) Google Scholar The for the between studies is to or or the testing used volume of sample sensitivity of cobas MTB in participants to Ultra. sensitivity in is the high of TB co-infected with HIV in settings as Global Tuberculosis Report 2023. World Health Organization, Geneva, Switzerland2023Google Scholar of Ultra to cobas MTB in the samples tested in South Africa. The in performance between Xpert and cobas MTB in samples Moldova and to cobas MTB for MTBC detection, whereas Xpert only a In the in specificity estimates to the of cobas MTB to detect from bacilli for which detection by Xpert is Ultra and Xpert performance estimates as the tests were not on the were evaluated in settings with TB South Africa a high TB and HIV and used sample and testing The sensitivity for both Ultra and cobas MTB in from South Africa with as estimates are of the sample Sensitivity of cobas MTB was in Moldova and India, in study assay This the for diagnostic accuracy studies using WHO-recommended assays as comparator The sensitivity of cobas MTB-RIF/INH for RIF resistance detection was with 96% and 100% for Xpert and Ultra, respectively. The of have been as the included assay from both sample and whereas and were only performed on one sample per sample was with cobas MTB-RIF/INH, whereas for four of the samples from Moldova, Xpert also to detect RIF resistance. This the of with a selection that were by direct testing of the molecular The sensitivity of cobas MTB-RIF/INH for INH resistance detection was also high the of samples from the FIND Specimen with katG mutations not by the cobas MTB-RIF/INH is an for to detect INH as individuals with INH-resistant TB a of the standard treatment In to detect INH resistance with the of RIF resistance mutations not by molecular assays to the spread and of multi–drug-resistant TB E. M. C. A. L. C. G. P. Niemann S. M. L. M. P. E. of tuberculosis in South Africa by an Infect Dis. Full Text Full Text PDF PubMed Google Scholar The high for cobas MTB-RIF/INH are a of the of detection for with cobas MTB, which cobas MTB-RIF/INH samples were smear in this study the in sample and comparator test for South Africa Moldova and India. pDST was using the WHO-recommended RIF of 1 at the of the study. However, study WHO were a of for Report on for of and the and World Health Organization, Geneva, Switzerland2021Google Scholar in this pDST have any samples with rpoB mutations between and 1 samples have been as the was by the on study is to have been of samples used in this study to evaluate the performance of the cobas MTB-RIF/INH assay to detect RIF and INH resistance were from the FIND Specimen the from this study not that these samples of have been a selection in the rpoB and katG mutations that were in this study. In cobas MTB and cobas MTB-RIF/INH high diagnostic performance in a diverse clinical assays suitable to molecular tests for the detection of TB and drug in settings high-throughput testing is and are of and were by FIND at the of the study. and are of India. from Research Center (Borstel, Germany) to Roche Switzerland) the of the study in and are the study participants and their for in this the study site for and in the study and with the of the and the Health and the Health Center for recruitment of participants and study Medical the of the was provided by by and with E. L. D. A. L. K. for PubMed Scopus Google Scholar was provided by of the Institute of Institute Council of Medical Research–Regional Medical Research Health Research of Health University of South Geneva, Geneva, Kranzer and WHO Supranational Reference Center for Research Center Germany), Geneva, Institute University of and of Health Research of Health University of and Institute and specificity of cobas MTB, Xpert, and Ultra for Mycobacterium tuberculosis complex detection by recruitment site of both sample reference standard reference standard and specificity of cobas MTB, Xpert, and Ultra for Mycobacterium tuberculosis complex detection by sample reference standard reference standard South and specificity of cobas MTB, Xpert, and Ultra for Mycobacterium tuberculosis complex detection by of tuberculosis reference standard reference standard South with with with with