Clinical validity of an automated DNA amplification system for diagnosis of pulmonary tuberculosis

Objective - The Cobas Amplicor(TM) PCR system used to detect Mycobacterium tuberculosis and diagnose pulmonary tuberculosis was compared to the cultur e results of this bacterium. Antituberculosis treatment was taken into acco unt to assess the clinical validity of a fast automated PCR based diagnosis . Material and methods - From February 1997 to January 1998, 407 respiratory specimens from 205 patients were examined, including 237 sputa, 110 bronchi al aspirates, 24 bronchoalveolar lavages, and 36 gastric aspirates. Patient s were classified clinically as having a life-threatening disease (group 1) , high probability of tuberculosis (group 2), or low probability of tubercu losis (group 3). Most patients (62 %) belonged to the second group. Results - The specificity of the Cobas Amplicor(TM) PCR system cached 100% compared to culture or treatment, and the sensitivity of direct staining an d PCR was 54% and 88% respectively compared to culture. Three patients had PCR-negative, culture-positive specimens without inhibitors. Inhibitory sub stances were removed by freezing and/or heating DNA extracts in 7/8 specime ns. The impact of PCR results on the decision to treat was low in smear-neg ative patients (even when tuberculosis was highly suspected) and greater in smear-positive patients. Comments - PCR may help confirm the diagnosis of tuberculosis more rapidly than culture, especially in severely ill patients or those with a suspected disease. (C) 2000 Editions scientifiques et medicales Elsevier SAS.