Background: In Africa, tuberculous meningitis (TBM) is an important opportunistic infection in HIV-positive patients. Current diagnostic tools for TBM perform sub-optimally. In particular, the rapid diagnosis of TBM is challenging because smear microscopy has a low yield and PCR is not widely available in resource-poor settings.
Methods: We evaluated the performance outcome of a novel standardized lipoarabinomannan (LAM) antigen-detection assay, using archived cerebrospinal fluid samples, in 50 African TBM suspects of whom 68% were HIV-positive.
Results: Of the 50 participants 14, 23 and 13 patients had definite, probable and non-TBM, respectively. In the non-TB group there were 5 HIV positive patients who were lost to follow-up and in whom concomitant infection with Mycobacterium tuberculosis could not be definitively excluded. The test sensitivities and specificities were as follows: LAM assay 64% and 69% (cut-point 0.22), smear microscopy 0% and 100% and PCR 93% and 77%, respectively.
Conclusion: In this preliminary proof-of-concept study, a rapid diagnosis of TBM could be achieved using LAM antigen detection. Although specificity was sub-optimal, the estimates provided here may be unreliable because of a classification bias inherent in the study design where it was not possible to exclude TBM in the presumed non-TBM cases owing to a lack of clinical follow-up. As PCR is largely unavailable, the LAM assay may well prove to be a useful adjunct for the rapid diagnosis of TBM in high HIV-incidence settings. These preliminary results justify further enquiry and prospective studies are now required to definitively establish the place of this technology for the diagnosis of TBM.