The manual MGIT system for the detection of M. tuberculosis in respiratory specimens: an experience in the University Malaya Medical Centre

A prospective study was conducted on 510 respiratory specimens for the presence of M. tuberculosis detected by direct acid-fast bacilli (AFB) smear examination, culture in the Manual Mycobacteria Growth Indicator Tube (BBL MGIT, Becton-Dickinson) and culture on Lowenstein-Jensen (LJ) medium. From positive BBL MGIT tubes, Ziehl-Neelsen and Gram stains were performed and subcultures were put up on LJ medium. A total of 101 (19.8%) specimens were positive by the BBL MGIT, 60 (11.8%) by primary LJ medium culture, 31 (6.1%) by direct smear examination and 29 (5.7%) by all three methods. Using primary LJ culture as the gold standard, the sensitivity and specificity of the BBL MGIT were 90% and 89.6% respectively but the sensitivity of AFB smear microscopy was only 48.3%. About half (51.1%) of the BBL MGIT false positives were due to contamination by non-AFB bacteria. The remaining false positives comprised specimens that were AFB microscopy positive but LJ culture negative. Of the AFB isolates obtained on LJ primary and sub-cultures, almost all (93.3%) were identified as Mycobacterium tuberculosis complex. The mean time-to-detection was significantly shorter (p < 0.0001) for the BBL MGIT than for LJ culture. For the former, positive results were available within 14 days for both AFB smear-positive and AFB smear-negative specimens. On the average, positive results were obtained 1.8 days earlier for direct AFB smear-positive samples than for AFB smear-negative samples. On the other hand, positive growth on LJ medium appeared after at least 33 days of incubation. These findings suggest that the BBL MGIT system will be a suitable alternative to LJ culture for the routine diagnosis of pulmonary tuberculosis, but a combination of liquid and solid cultures is still required for the highest diagnostic accuracy.