Pneumologie 2012; 66 - P351
DOI: 10.1055/s-0032-1302666

Direct comparison of GeneXpert MTB/RIF (Cepheid) with probeTEC (Becton-Dickinson) and Taqman MTB (Roche) for detection of TB bacteria

H Hoffmann 1, S Hofmann-Thiel 1, L Turaev 2, A Esenalieva 3, M Abdullaeva 4, K Mirazim 2, U Antonenka 1
  • 1IML Red GmbH, Synlab Bayern, Asklepios Fachklinken, WHO Supranationales Referenzlabor für Tuberkulose, Gauting
  • 2National Institute of Tuberculosis, National Reference Laboratory of Tuberculosis, Tashkent, Republic of Uzbekistan
  • 3National Reference Laboratory of Tuberculosis, Bishkek, Republic of Kyrgyzstan
  • 4National Reference Laboratory of Tuberculosis, Machiton, Republic of Tajikistan

Background: With the GeneXpert MTB/RIF system, Cepheid and FIND have launched an innovative solution of TB diagnostics. It is easy to use, bio-safe, resistant to DNA-contamination, fulfilling many POCT criteria. As to detection of TB bacteria (MTBC) in respiratory specimens, a recent multi-centre-evaluation-study (Böhme et al. 2010) described 98.2% sensitivity in smear positive (ss+) sputa, 72.5% in smear negative (ss-) ones, while specificity reached 99.2%. These values are higher than most ones reported for commercial assays so far.

Scope: Comparing Xpert head-to-head to probeTecTM MTBC (Becton-Dickenson) and COBAS® Taqman® MTB (Roche).

Results: 1ml each of 122 decontaminated sputa were selected from the pre-characterized specimen bank of IML Gauting. 69 samples grew TB bacteria in culture. 51 were ss-, 18 ss+. 53 samples were TB negative, 23 of them grew non-tuberculous mycobacteria (NTM) in culture. Xpert, probeTec and Taqman were performed following the manufacturers' instructions. Xpert, probeTec, and Taqman yielded 1, 1, and 5 indeterminate results, respectively. Overall sensitivity values were 73.5%, 77.9%, 72.7% for Xpert, probeTec, and Taqman, respectively. Sensitivities for ss- samples were higher for probeTec (72.0%) than for Xpert (70.0%) or Taqman (68.0%). With ss+ samples, Xpert had the lowest detection rate (83.5%, probeTec 94,4%, Taqman 87.5%). Specificity was best for Taqman (100%) and lowest for Xpert (92.5%) due to misidentification of four NTM as MTBC.

Conclusions: Directly comparing Xpert with two commercial assays, the system did not perform better. Well known strengths and weaknesses (e.g. high specificity but relatively lower sensitivity of Taqman) of conventional assays were confirmed in the present study. Sample collection was not representive for routine diagnostics. Therefore positive and negative predictive values need to be statistically adjusted and will be discussed in the presentation.