Diagnostic performance of the Xpert MTB/RIF assay in BAL fluid samples from patients under clinical suspicion of pulmonary tuberculosis: a tertiary care experience in a high-tuberculosis-burden area

Guilherme Machado Xavier de Brito, Thiago Thomaz Mafort, Marcelo Ribeiro-Alves, Larissa Vieira Tavares dos Reis, Janaína Leung, Robson Souza Leão, Rogério Rufino, Luciana Silva Rodrigues

To assess the diagnostic performance of the Xpert MTB/RIF assay, a rapid molecular test for tuberculosis, comparing it with that of AFB staining and culture, in BAL fluid (BALF) samples from patients with clinically suspected pulmonary tuberculosis (PTB) who are sputum smear-negative or produce sputum samples of insufficient quantity. This was a retrospective study of 140 cases of suspected PTB in patients who were smear-negative or produced insufficient sputum samples and were evaluated at a tertiary teaching hospital in the city of Rio de Janeiro, Brazil. All of the patients underwent fiberoptic bronchoscopy with BAL. The BALF specimens were evaluated by AFB staining, mycobacterial culture, and the Xpert MTB/RIF assay. Among the 140 patients, results for all three microbiological examinations were available for 73 (52.1%), of whom 22 tested positive on culture, 17 tested positive on AFB staining, and 20 tested positive on the Xpert MTB/RIF assay. The sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy for AFB staining were 68.1%, 96.1%, 88.2%, 87.5%, and 87.6%, respectively, compared with 81.8%, 96.1%, 90.0%, 92.4%, and 91.8%, respectively, for the Xpert MTB/RIF assay. The agreement between AFB staining and culture was 82.3% (kappa = 0.46; p < 0.0001), whereas that between the Xpert MTB/RIF assay and culture was 91.8% (kappa = 0.8; p < 0.0001). In BALF samples, the Xpert MTB/RIF assay performs better than do traditional methods, providing a reliable alternative to sputum analysis in suspected cases of PTB. However, the rate of discordant results merits careful consideration.