Table 1.
The 56 mycobacterial strains from The Netherlands obtained via The National Institute for Public Health and the Environment (RIVM).
Table 2.
103 primary isolates from Stellenbosch, Cape Town, South Africa.
Table 3.
Compounds identified as relevant for the differentiation of NTM and MTB strains.
Figure 1.
Fingerprint patterns of the normalized areas of the 20 markers in samples from the Netherlands.
The graph shows the fingerprint patterns of the normalized areas of 20 marker compounds in different mycobacteria from the Netherlands with (A) M. tuberculosis, (B) M. gordonae, (C) M. marinum, and (D) and (E) two M. kansasii type I strains. Case (A) belongs to the MTB group; cases (B–E) belong to the NTM group. Compounds 1–20 are identified in Table 3.
Table 4.
Results of the analysis of 56 mycobacterial strains from patients in the Netherlands using THM-GC-MS and the 20-compound model.
Figure 2.
Representative examples of the THM-GC-MS chromatograms.
Chromatograms are shown for A: M. tuberculosis, B: M. avium, C: M. marinum, D: M. kansasii.
Figure 3.
Fingerprint patterns of the normalized areas of the 20 markers in samples from South Africa.
The graph shows the fingerprint patterns of the normalized areas of 20 marker compounds in different mycobacteria from South Africa with two representative M. tuberculosis strains (A & B), one representative NTM strain M. intracellulare (C), five misclassified M. tuberculosis strains (D–H). Note the high value of compound 3 and the low values of compound 17–20. Compounds 1–20 are identified in Table 3.
Table 5.
Results of the analysis of 103 primary isolates from patients from Stellenbosch, South Africa using THM-GC-MS and the 20-compound model.
Figure 4.
Algorithm derived using manual inspection of 103 samples from South Africa.
The graph shows a decision tree for the differentiation of M. tuberculosis complex and non-tuberculous mycobacteria.
Figure 5.
(A) used 44 samples from the Netherlands as training and validation sets and 159 samples as test sets (Testset-1, 56 Dutch samples and Testset-2, 103 samples from South Africa); (B) used all 203 samples as training and validation sets.