Abstract
BACKGROUND: Laboratory confirmation of the diagnosis of tuberculous meningitis (TBM) has always been problematic. Using the uniform case definition suggested by Marais et al., we determined the sensitivity of a variety of laboratory tests. METHODS: Human immunodeficiency virus (HIV)-seropositive patients suspected of having subacute meningitis were included in the study. Using the uniform case definition, patients were divided into possible and probable cases of TBM. The following specific tests were done on the cerebrospinal fluid (CSF): layered Ziehl-Neelsen (ZN) staining, CSF culture and a panel of nucleic acid amplification tests (NAAT) consisting of the GenoType MTBDRplus assay, Cepheid Xpert MTB/RIF, the MTB Q-PCR Alert (Q-PCR) and the loop-mediated isothermal amplification (LAMP) assay. The sensitivity of each test was compared to the case definition and to each other. RESULTS: A total of 68 patients were evaluated. Using the uniform case definition only, without any of the specific laboratory tests, there were 15 probable cases (scores > 12) and 53 possible cases (scores 6-11) of TBM. When the uniform case definition was tested against any laboratory test, 12 of the 15 (80%) probable cases and 26 of the 53 (49.1%) possible cases had laboratory confirmation. When each test was compared to any other test, the sensitivities for the Xpert MTB/RIF, GenoType MTBDRplus, CSF culture, Q-PCR, LAMP and ZN layering were 63.2 (46.0-78.2), 76.3 (59.8-88.6), 65.7 (47.8-80.9), 81.1 (64.8-92.0), 70.3 (53.0-84.1) and 55.6 (38.1-72.1), respectively. CONCLUSION: In this study, the GenoType MTBDRplus and the Q-PCR tests performed better than the Xpert MTB/RIF. Because the Xpert MTB/RIF is not good enough to 'rule out' TBM, a negative result should be followed up by another NAAT, such as the GenoType MTBDRplus or Q-PCR. The LAMP assay may be considered as the first test in resource-poor settings. At the time of the study, we did not have access to the Xpert MTB/RIF Ultra, which has now been recommended by the World Health Organization as the test of first choice. However, even this test has a similar limitation as the Xpert MTB/RIF, with two recent studies showing variable results.