The identification of biomarkers differentiating Mycobacterium tuberculosis and non-tuberculous mycobacteria via thermally assisted hydrolysis and methylation gas chromatography-mass spectrometry and chemometrics

Infections with non-tuberculous mycobacteria (NTM) are increasing, particularly among immune-compromised patients and those with damaged lungs. Mycobacterium tuberculosis complex (MTB) strains, however, remain the most common cause of mycobacterial infection. A rapid method of distinguishing MTB from NTM is required for correct diagnosis and tuberculosis management. We have developed an automated procedure based on thermally-assisted hydrolysis and methylation followed by gas chromatography–mass spectrometry (THM–GC–MS) and advanced chemometrics to differentiate MTB from NTM. We used early cultures of mycobacteria in this first step towards the direct identification of these bacteria in sputum using a hand-held portable device. To build a classification model, we used 44 strains including 15 MTB and 29 NTM. A matrix of the aligned dataset containing ~45,700 features (retention time/mass pairs) for the 44 observations was submitted to partial least squares discriminant analysis (PLS–DA). We could reduce the number of features down to 250 without compromising the accuracy of the model. Twenty different compounds were found through mass spectral interpretation of these 250 features. Some of these compounds have not been linked to tuberculosis before, others have been proposed previously as diagnostic biomarkers for this disease. We have built a final model based on our proposed biomarkers that performed with 95 % accuracy in distinguishing MTB from NTM in early cultures. Since all these biomarkers have been chemically identified, work can proceed towards the development of simpler, bed-side diagnostic tests to differentiate MTB from NTM in sputum.