Comparison of two automated DNA amplification systems with a manual one-tube nested PCR assay for diagnosis of pulmonary tuberculosis.

Eighty-four specimens of respiratory secretions culture positive for mycobacteria (70 positive for Mycobacterium tuberculosis and 14 positive for nontuberculous mycobacteria) and 120 culture-negative specimens were evaluated by three DNA amplification techniques: a manual in-house single-tube nested PCR (nPCR) and two commercial automated assays (the Cobas Amplicor System [aPCR-h] from Roche Diagnostic Systems and the Abbott LCx Probe System [aLCx-p] from Abbott Laboratories). The overall diagnostic sensitivities of the nPCR, aPCR-h, and aLCx-p were 77.1, 84.3, and 77.1%, respectively, and the sensitivities were 57.9, 57.9, and 36.8%, respectively, for smear-negative specimens. Specimens culture positive for nontuberculous mycobacteria were negative by all three assays. Eight culture-negative specimens which were positive by one or more assays had previously been documented by culture to be positive for M. tuberculosis and were taken from patients who were treated with antituberculosis agents. Retesting of specimens negative by one assay by the other two assays revealed that each test had its unique group of negative specimens. When considering the DNA extraction and amplification steps of these assays separately, it was found that extracts from aPCR-h and aLCx-p were compatible with nPCR amplication, while the two automated assays could only amplify extracts processed with their own reagents. Limiting dilution analysis revealed that the order of analytical sensitivity was nPCR, followed by aLCx-p and then aPCR-h. Comparison of the work flow of each assay revealed that although the aPCR-h demands the least specimen handling, the turnaround time of aLCx-p is the most favorable.