LETTER

Roggenkamp and colleagues recently reported the results of their comparative evaluation of the MB/BacT and the BACTEC 460 (3). They observed a significantly lower recovery rate for Mycobacterium tuberculosis using the MB/BacT, and they commented that new diagnostic systems should be evaluated under conditions found in nonspecialized laboratories that have a low rate of detection of specimens positive for M. tuberculosis.

Our comparative study of these two systems (1) was performed in a nonspecialized hospital laboratory that processes approximately 5,000 specimens per year for mycobacterial culture, with 5.6% positive for mycobacteria and 1.5% positive for M. tuberculosis, similar to the numbers reported by Roggenkamp et al. We demonstrated that the MB/BacT detected the M. tuberculosis-M. avium complex at a rate that was not significantly different from that of the BACTEC 460. The numbers of specimens that were culture positive for mycobacteria in our study (44) were less than the 123 specimens positive in the study of Roggenkamp et al., and it is possible that with more specimens, differences may have been found. The decreased sensitivity of the MB/BacT system observed by Roggenkamp et al. could be due to differences in the specimens studied, the products used, and the methods. The epidemiology and presentation of patients with mycobacterial disease in Europe and the United States could differ and thereby influence the number of organisms in specimens. Our specimens were predominantly of respiratory origin, whereas the specimens in the Roggenkamp study included urine and gastric fluid, as well as samples from sterile sites. Among 23 specimens positive for M. tuberculosis in our study, 15 (65%) were smear positive, in comparison to 25 of 71 (35%) in the Roggenkamp study. Unlike Roggenkamp and colleagues, we screened all specimens with the more sensitive Auramine O fluorochrome as mandated by the College of American Pathologists (2). The MB/BacT detected all eight smear-negative M. tuberculosis-infected patients in our study. In a previous evaluation of the MB/BacT system, we had observed decreased sensitivity for nontuberculous mycobacteria and a high rate of nonmycobacterial contamination. The manufacturer subsequently revised the medium and the antibiotic reconstitution supplement. Our published study was performed with the revised reconstitution fluid and antibiotic supplement in which oleic acid was substituted for Tween 80 to enhance mycobacterial growth, and vancomycin was added to reduce overgrowth of gram-positive bacteria. Roggenkamp and colleagues used the original supplement. We maximized the recovery of mycobacteria in the MB/BacT by staining contaminated bottles weekly for the full 6 weeks of incubation. Twelve of 33 (36%) of the mycobacteria not detected in the Roggenkamp study, 6 of which were M. tuberculosis, were due to bacterial contamination that may have been improved by using the revised antibiotic supplement that is now available. The performance of this revised supplement will depend on the handling of specimens and the patient population studied.

In summary, the poor results obtained with the MB/BacT by Roggenkamp et al. might have been influenced by the type of specimens studied, how they were processed, and the growth medium and antibiotic supplement utilized. It would be interesting for them to do a study using the same type of experimental design and media as we reported in our trial of this instrument.