RESUMO
OBJECTIVES: Tuberculosis is one of the most prevalent infections in humans. Although culture is the reference for diagnosis, its sensitivity is compromised, especially in paucibacillary samples. Because polymerase chain reaction (PCR) amplifies mycobacterial DNA, it is more sensitive than culture for the diagnosis of Mycobacterium tuberculosis (Mtb). However, its performance can be affected by intrinsic sample inhibitors and by the extraction/detection techniques used. METHODS: We evaluated the influence of preanalytical conditions on Mtb detection in samples of sputum (SPU), bronchoalveolar lavage (BAL), and pleural fluid (PF) using combinations of extraction/detection methods. Respiratory samples were prepared to contain different concentrations of red blood cells and nucleated cells to which increasing amounts of Mtb colonies were inoculated and submitted to PCR. RESULTS: Up to 102 CFU/ml of Mtb were detected in the SPU in all methods, except for the Roche extraction/detection method, regardless of the preanalytical sample condition. In BAL samples, medium and high concentrations of cells and high concentrations of red blood cells contributed to a lower Mtb detection, regardless of the extraction method used. In PF, red blood cells were the variable that most interfered with Mtb detection, with better recovery (102 CFU/ml) observed with the Qiagen/Nanogen combination. CONCLUSION: The choice of Mtb extraction and detection method is of fundamental importance for PCR analytical sensitivity, especially when paucibacillary samples and/or samples containing potential PCR inhibitors are analyzed.