ABSTRACT
PCR is, in principle, a simple and rapid test for use in the detection of Mycobacterium tuberculosis. However, virtually no data are available on the reliability and reproducibility of the method. In order to assess the validity of PCR for the detection of mycobacteria in clinical samples, seven laboratories participated in a blinded study of 200 sputum, saliva, and water samples containing either known numbers of Mycobacterium bovis BCG cells or no added organisms. Each laboratory used its own protocol for pretreatment, DNA extraction, and detection of the amplification product. Insertion sequence IS6110 was the target for DNA amplification. Several participating laboratories reported high levels of false-positive PCR results, with rates ranging from 3 to 20% and with one extreme value of 77%. The levels of sensitivity also ranged widely among the different participants. A positive PCR result was reported for 2 to 90% of the samples with 10(3) mycobacteria. Although most participants did include control tests to check the sensitivity and specificity of the PCR, the sequence of operations from sample pretreatment to purification of DNA from bacteria was not always monitored adequately. During these procedures cross-contaminating DNA was introduced and/or bacterial DNA was lost. The results of the study show that the implementation of an effective system for monitoring sensitivity and specificity is required before the PCR can be used reliably in the diagnosis of tuberculosis.
