Low-complexity automated nucleic acid amplification tests for extrapulmonary tuberculosis and rifampicin resistance in adults and adolescents.

Background: Low-complexity automated nucleic acid amplification tests (LC-aNAATs) are molecular World Health Organization (WHO)-recommended rapid diagnostic tests widely used for simultaneous detection of Mycobacterium tuberculosis complex and rifampicin resistance in sputum. To extend our previous review on extrapulmonary tuberculosis, we performed this update to inform a WHO policy update.

Objectives: To estimate the diagnostic accuracy of LC-aNAATs for extrapulmonary tuberculosis and rifampicin resistance in adults and adolescents with presumptive extrapulmonary tuberculosis.

Search methods: We searched the Cochrane Central Register of Controlled Trials, MEDLINE, Embase, Science Citation Index, Latin American Caribbean Health Sciences Literature, Scopus, ClinicalTrials.gov, the WHO International Clinical Trials Registry Platform, the International Standard Randomized Controlled Trial Number Registry, and ProQuest, up to 11 October 2023, without language restriction. A WHO public call for data was made between 30th November 2023 and 15th February 2024 to identify unpublished studies.

Selection criteria: We included cross-sectional and cohort studies using non-respiratory specimens and eight forms of extrapulmonary tuberculosis: tuberculous meningitis and pleural, lymph node, bone or joint, genitourinary, peritoneal, pericardial, and disseminated tuberculosis. Reference standards were culture and a study-defined composite reference standard (tuberculosis detection); and phenotypic drug susceptibility testing with or without genotypic drug susceptibility testing (rifampicin resistance detection). Index tests included Xpert Ultra, Truenat assays, STANDARD M10, and Iron qPCR.

Data collection and analysis: Two review authors independently extracted data and assessed the risk of bias and applicability using the QUADAS-2 tool. For tuberculosis detection, we performed separate analyses by specimen type and reference standard using the bivariate model to estimate summary sensitivity and specificity with 95% confidence intervals (CIs). Based on a pre-defined condition, based on sample sizes and type of technology for performing class-based analysis, data for Truenat MTB Plus were not included in the meta-analyses for LC-aNAATs. Hence, we present results for Xpert Ultra and Truenat MTB Plus separately. We assessed the certainty of evidence using the GRADE approach.

Main results: We included 37 unique studies where 36 studies evaluated Xpert Ultra and three studies evaluated Truenat MTB plus. We found no eligible studies for the other index tests. Overall, the risk of bias was low for patient selection, index test, and flow and timing domains. For the reference standard, the risk of bias for included studies was low (75%) or unclear (25%). Applicability for the patient selection domain was unclear for most studies because we were unsure of the clinical settings, and the applicability concern was low for most studies for the reference standard domain. Cerebrospinal fluid Xpert Ultra (16 studies) Xpert Ultra summary sensitivity and specificity (95% CI) against a microbiological reference standard were 88.2% (83.7 to 91.6) (287 participants; high-certainty evidence) and 96.0% (86.8 to 98.9) (1397 participants; moderate-certainty evidence). Truenat MTB Plus (2 studies) There were not enough data to meta-analyze, and we have provided descriptive results for Truenat MTB Plus. The sensitivities in these two studies ranged from 95% to 100% while the specificities ranged from 55% to 100% against a microbiological reference standard. The sensitivity was 78.7% (70 to 86) and the specificity was 100% (91 to 100) against a composite reference standard from a single study. Pleural fluid Xpert Ultra (13 studies) Xpert Ultra summary sensitivity and specificity against a microbiological reference standard were 74.0% (60.8 to 83.9; 264 participants; low-certainty evidence) and 88.1% (78.8 to 93.6; 777 participants; very low-certainty evidence). Truenat MTB Plus (1 study) The sensitivity was 100% (2.5 to 100) and specificity was 100% (95.3 to 100) against a microbiological reference standard. Lymph node aspirate Xpert Ultra (6 studies) Xpert Ultra summary sensitivity and specificity (95% CI) against a composite reference standard were 71.3% (64.3 to 77.4) (243 participants; moderate-certainty evidence) and 97.4% (82.3 to 99.7) (218 participants; very low-certainty evidence). Truenat MTB Plus (1 study) The sensitivity and specificity were 77.1% (66 to 86) and 100% (88 to 100), respectively, against a microbiological reference standard. The sensitivity was 100% (81 to 100) and specificity was 56% (45 to 67) against a composite reference standard. Rifampicin resistance Xpert Ultra (13 studies) Xpert Ultra summary sensitivity and specificity were 100.0% (93.4 to 100.0; 54 participants; high-certainty evidence) and 99.4% (92.1 to 100.0; 392 participants; high-certainty evidence).

Authors' conclusions: LC-aNAATs are helpful in diagnosing extrapulmonary tuberculosis. Sensitivity varies across different extrapulmonary specimens, while for most specimens specificity is high, the tests rarely yielding a positive result for people without tuberculosis. For tuberculous meningitis, Xpert Ultra had high sensitivity against culture. Xpert Ultra also had high sensitivity and specificity for rifampicin resistance. Future research should acknowledge the concern associated with culture as a reference standard in paucibacillary specimens and consider ways to address this limitation. Additionally, there is a critical need for robust evidence on other technologies within the LC-aNAAT class.

Funding: Funded by the WHO Global Tuberculosis Program.

Registration: This is an update to the published review "Xpert MTB/RIF Ultra and Xpert MTB/RIF assays for extrapulmonary tuberculosis and rifampicin resistance in adults" via doi: 10.1002/14651858.CD012768.pub3.