RESUMO
BACKGROUND: Tuberculous meningitis (TbM) is the most severe complication of extra pulmonary tuberculosis (Tb). There is a higher frequency of positive cerebrospinal fluid (CSF) cultures for Mycobacterium tuberculosis (MTb) in samples from human immunodeficiency virus (HIV) co-infected patients than in those from HIV-negative patients. We hypothesized that real time PCR assays for MTb (MTb qPCR) using CSF would be more sensitive in HIV co-infected patients owing to a greater MTb burden. The present study aimed to verify the diagnostic performance of MTb qPCR in CSF of TbM patients who either were co-infected with HIV or were HIVnegative. METHODS: A total of 334 consecutive participants with suspected TbM were divided into two groups: HIV co-infected and HIV-negative; each group was categorized into definite TbM, probable TbM, possible TbM, and TbM-negative subgroups based on clinical, laboratory and imaging data. We evaluated the diagnostic characteristics of MTb qPCR analysis to detect TbM in CSF by comparing the results to those obtained for definite TbM (i.e., positive MTb culture) and/or probable TbM in CSF, as gold standard. RESULTS: The sensitivity of MTb qPCR in the definite and probable subgroups of the HIV coinfected participants (n = 14) was 35.7%, with a specificity of 93.8%, negative predictive value (NPV) of 94.4%, and negative clinical utility index (CUI-) of 0.89. Results of the HIV-negative group (n = 7) showed lower sensitivity (14.3%) and similar specificity, NPV, and CUI-. CONCLUSION: The findings confirmed our hypothesis, despite the low sensitivity. MTb qPCR may significantly contribute to diagnosis when associated with clinical criteria and complementary examinations.