Comparative usefulness of PCR in the detection of Mycobacterium tuberculosis in different clinical specimens.

The role of the polymerase chain reaction (PCR) in the diagnosis of tuberculosis in clinical practice remains to be defined; most results have been based on sputum samples. This study systematically compared the relative sensitivity and specificity of a single simplified method for different clinical samples. A wide range of clinical samples, including sputum, bronchoalveolar lavage fluid, cerebrospinal fluid, pleural fluid, gastric aspirate, pus and tissues (both fresh and paraffin-embedded) was tested. This method did not require routine DNA extraction before PCR, and consisted of an optimised single tube PCR amplification designed with different sets of time and temperature profiles. A total of 398 samples from 293 patients was studied. The sensitivity was 100% for all types of specimens, while the specificity ranged from 95% for sputum to 88% for bronchoalveolar lavage fluid and pleural fluid and to 85% for non-pulmonary specimens. This study showed that it was possible to employ a single simplified method with minor modifications for a wide range of specimens in clinical practice without loss of sensitivity and specificity.