Evaluation of the Roche AMPLICOR™ MTB assay for the detection of Mycobacterium tuberculosis in sputum specimens from prison inmates

Abstract

The reliability of the Roche Mycobacterium tuberculosis polymerase chain reaction (PCR) assay (AMPLICOR™ MTB) for the diagnosis of pulmonary tuberculosis was evaluated by testing expectorated sputum specimens from 187 inmates in Texas state prisons and comparing the results to culture and medical history. Of the 80 specimens that were culture positive for mycobacteria, 36 specimens from 16 patients grew M. tuberculosis. Forty-six specimens were smear positive for acid-fast bacilli (AFB), and of these, M. tuberculosis was isolated from 24. On initial testing, 52 specimens were PCR positive. Thirty-one of these 52 were culture positive for M. tuberculosis, and 21 were culture negative, resulting in a PCR sensitivity and specificity of 86.1 and 96.1%, respectively. After resolving discrepancies by review of the medical history and repeat testing, PCR sensitivity, specificity, and positive and negative predictive values, respectively, were 92.8, 99.8, 98.1, and 99.2%. For AFB smear-positive specimens, the sensitivity, specificity, and positive and negative predictive values, were 95.8, 100, 100, and 93.3, respectively; whereas, for AFB smear-negative specimens, these values were 87.5, 99.7, 95.5, and 99.4%, respectively. These results confirm the reliability of the AMPLICOR™ MTB assay for direct detection of M. tuberculosis in AFB smear-positive sputum specimens and suggest a potential role in evaluating AFB smear-negative sputum specimens.