ABSTRACT
BACKGROUND: Nucleic acid amplification assays allow for the rapid and accurate detection of Mycobacterium tuberculosis (MTB) directly in clinical specimens thereby facilitating diagnosis of tuberculosis (TB). With the fully automated Xpert MTB/RIF system (Cepheid) an innovative solution of TB diagnostics has been launched. We performed a direct head-to-head comparison of Xpert MTB/RIF with two widely used commercial assays, ProbeTec ET DTB (DTB) (Becton-Dickinson) and COBAS TaqMan MTB (CTM-MTB) (Roche). METHODS: 121 pre-characterized respiratory specimens (68 culture-positive for MTB complex, 24 culture-positive for non-tuberculous mycobacteria and 29 culture-negative) taken from our frozen specimen bank were tested for the presence of MTB complex by the three assays. RESULTS: Among culture-positive samples (n = 68), overall sensitivity for detection of MTB complex was 74.6%, 73.8%, and 79.1% for Xpert MTB/RIF, CTM-MTB, and DTB, respectively. Within the subgroup of smear-negative TB samples (n = 51) sensitivity was 68% for Xpert MTB/RIF and CTM-MTB and 72% for DTB. Among smear-positive TB samples (n = 17), all (100%) were detected by DTB and 94.1% and 93.3% by Xpert MTB/RIF and CTM-MTB, respectively. Specificity was best for CTM-MTB (100%) and lowest for Xpert MTB/RIF (96.2%) due to misidentification of two NTM samples as MTB complex. CTM-MTB yielded the highest rate of invalid results (4.1%) (0.8% by Xpert MTB/RIF and DTB, respectively). CONCLUSIONS: The direct comparison of Xpert MTB/RIF with CTM-MTB and DTB yielded similar overall performance data. Whereas DTB was slightly superior to Xpert MTB/RIF in terms of sensitivity, at least in the sample collection tested here, CTM-MTB performed best in terms of specificity.