Rapid identification of laboratory contamination with Mycobacterium tuberculosis using variable number tandem repeat analysis

Compared with solid media, broth-based mycobacterial culture systems have i ncreased sensitivity but also have higher false-positive rates due to cross -contamination. Systematic strain typing is rarely undertaken because the t echniques are technically demanding and the data are difficult to organize. Variable number tandem repeat (VNTR) analysis by PCR is rapid and reproduc ible. The digital profile is easily manipulated in a database. We undertook a retrospective study of Mycobacterium tuberculosis isolates collected ove r an 18-month period following the introduction of the BACTEC MGIT 960 syst em. VNTR allele profiles were determined with early positive broth cultures and entered into a database with the specimen processing date and other sp ecimen data. We found 36 distinct VNTR profiles in cultures from 144 patien ts. Three common VNTR profiles accounted for 45% of true-positive cases. By combining VNTR results with specimen data, we identified nine cross-contam ination incidents, six of which were previously unsuspected. These nine inc idents resulted in 34 false-positive cultures for 29 patients. False-positi ve cultures were identified for three patients who had previously been cult ure positive for tuberculosis and were receiving treatment. Identification of cross-contamination incidents requires careful documentation of specimen data and good communication between clinical and laboratory staff. Automat ed broth culture systems should be supplemented with molecular analysis to identify cross-contamination events. VNTR analysis is reproducible and prov ides timely results when applied to early positive broth cultures. This met tled should ensure that patients are not plated on unnecessary tuberculosis therapy or that cases are not falsely identified as treatment failures. In addition, areas where existing procedures may be improved can be identifie d.