RESUMO
Identification of microorganisms by MALDI-TOF MS has been widely accepted in clinical microbiology. However, for Mycobacterium spp. and Nocardia spp. such identification has not yet reached the optimal level of routine testing. Here we describe the development of an identification tool for 49 and 15 species of Mycobacterium spp. and Nocardia spp., respectively. During database construction, a number of ambiguous reference identifications were revealed and corrected via molecular analyses. Eventually, more than 2000 individual mass spectra acquired from 494 strains were included in a reference database and subjected to bio-statistical analyses. This led to correct species identification and correct combination of species into several complexes or groups, such as the Mycobacterium tuberculosis complex. With the Advanced Spectrum Classifier algorithm, class-specific bin weights were determined and tested by cross-validation experiments with good results. When challenged with independent isolates, overall identification performance was 90% for identification of Mycobacterium spp. and 88% for Nocardia spp. However, for a number of Mycobacterium sp. isolates, no identification could be achieved and in most cases, this could be attributed to the production of polymers that masked the species-specific protein peak patterns. For the species where >20 isolates were tested, correct identification reached 95% or higher. With the current spectral database, the identification of Mycobacterium spp. and Nocardia spp. by MALDI-TOF MS can be performed in routine clinical diagnostics although in some complicated cases verification by sequencing remains mandatory.
