Identification of Mycobacterium tuberculosis and M. avium complex directly from smear-positive sputum specimens and BACTEC 12B cultures by high-performance liquid chromatography with fluorescence detection and computer-driven pattern recognition models.

A high-performance liquid chromatography method that utilized fluorescence detection (HPLC-FL) of mycolic acid 6,7-dimethoxycoumarin esters was developed to identify Mycobacterium tuberculosis (MTB) and M. avium complex (MAC) directly from fluorochrome stain smear-positive sputum specimens and young BACTEC 12B cultures. HPLC-FL chromatograms from a training set that included 202 smear-positive clinical sputum specimens and 343 mycobacterial cultures were used to construct a calibrated peak-naming table and computer-based pattern recognition models for MTB and MAC. Pattern recognition model performance was measured with an evaluation set of samples that included 251 smear-positive clinical sputum specimens and 167 BACTEC 12B cultures. Evaluation sputum specimens were culture positive for MTB (n = 132) and MAC (n = 48). With evaluation sputa, the MTB and MAC models were 56.8 and 33.3% sensitive, respectively. Evaluation set BACTEC 12B cultures were culture positive for MTB (n = 97) and MAC (n = 53). The sensitivities of the MTB and MAC models for identification of BACTEC 12B cultures were 99.0 and 94.3%, respectively. The specificity of both models was 100% for both types of evaluation samples. The average times from BACTEC 12B inoculation to cell harvest were 10.2 and 7.4 days for MTB and MAC, respectively. HPLC-FL can identify MTB and MAC in 1 day from many smear-positive sputa. Rapid and sensitive identification of MTB and MAC from young BACTEC 12B cultures was achieved.