A comparative study of the polymerase chain reaction and conventional procedures for the diagnosis of tuberculous pleural effusion.

Preliminary reports by ourselves and others suggest that amplification of mycobacterial DNA by the polymerase chain reaction (PCR) is a sensitive and rapid diagnostic test for tuberculosis. We recently described a PCR assay with a 336 bp repetitive sequence specific for Mycobacterium tuberculosis as the DNA target, which gave encouraging results in culture-positive smear-negative clinical specimens. In the present prospective study of patients with pleural effusions we compared PCR of the pleural fluid with conventional procedures. 84 adult patients with pleural effusions were divided into 4 groups. In group A (44 patients), M. tuberculosis was detected by culture of pleural fluid, pleural biopsy or extrapleural source. In group B (6 patients), tuberculous infection was confirmed by histology (group A excluded). Group C (3 patients) had clinical evidence of tuberculosis. Group D (31 patients) had no evidence of active M. tuberculosis infection. Analysis of the pleural fluid confirmed a sensitivity for PCR of 81%. The sensitivity of pleural fluid culture, culture of pleural biopsy, and histology of biopsy was 52.8%, 69.8% and 77.3% respectively. There were however 7 PCR positive results within group D; 6 of these were in patients with malignant effusions. We conclude that for the diagnosis of M. tuberculosis PCR is more sensitive than laboratory culture as determined by the analysis of pleural fluids. Positive PCR results among patients with malignant effusions may be false-positives or the result of latent tuberculous infections. PCR should remain an investigational procedure until prospective studies in high and low prevalence areas have critically evaluated the specificity of the assay.