RESUMO
BACKGROUND: The diagnosis of tuberculosis (TB) in developing countries, such as Bangladesh, is based mainly on microscopic detection of acid-fast bacilli in smears from clinical specimens. On the other hand, the detection of TB by polymerase chain reaction (PCR) is quite new in Bangladesh. In this study, we compared the molecular method with the conventional diagnosis procedures, where Lowenstein-Jensen medium culture results have been used as the "gold standard." METHODS: A total of 135 sputum samples were collected from clinically suspected patients with pulmonary TB. A direct smear was made from each sputum specimen and stained by the Ziehl-Neelsen (Z-N) method. The sputum samples were then processed, and the pellet was used for both Z-N (concentration) and auramine O fluorescence staining or resuspended in phosphate buffered saline to inoculate Lowenstein-Jensen medium or processed for PCR detection of Mycobacterium tuberculosis. RESULTS: The direct smear staining yielded 44 (32.6%) sputum samples that were acid-fast positive, which increased after concentration, yielding 60 (44.4%) acid-fast-positive samples. Fluorescence microscopy using auramine O staining further increased the number of positive samples to 67 (49.6%). The biochemical tests showed 75 (55.6%) sputum samples to be culture positive, and the MB/BacT system increased the recovery up to 90 (66.7%) culture positives. On the other hand, PCR yielded 93 (68.9%) positive results, 20 (21.5%) of which were culture-negative sputum specimens. CONCLUSION: It is suggested that the Z-N direct microscopy on its own is the best method (with high specificity) for confirming the diagnosis of acid-fast bacilli. Although the PCR diagnosis of TB appears to be a rapid and sensitive method, the results should be interpreted with care in the clinical settings.