Reliability of MALDI-TOF mass spectrometry in the identification of anaerobic bacteria

Aim of the study: MALDI-TOF mass spectrometry (MS) is becoming a major resource in the Clinical Microbiology laboratory. Results on some groups of microorganisms are still controversial. We have studied the reliability of MALDI-TOF MS for the identification of anaerobic clinical isolates was studied compared to conventional biochemical methods, with rRNA 16S sequencing being used as a reference when discrepancies arose. Material and methods: A total of 126 anaerobic bacteria clinical isolates were studied by using API20A kits (bioMerieux, Marcy l'Etoile, France) and MALDI-TOF MS (Autoflex II, Bruker Daltonics, Germany), and using the data library BioTyper 2.0 (Bruker Daltonics, Germany). When discrepancies arose, or MALDI-TOF MS was not able to identify any microorganism, rRNA 16S sequencing was used as the reference standard. Results: The biochemical method and MALDI-TOF MS agreed in identifying 60.9% of isolates at species level, and 20.3% of isolates at genus level. Among the 48 discrepancies observed, rRNA 16S sequencing supported MALDI-TOF MS identification, at species level, in 32 isolates (66.7%), and in 8 isolates (16.7%) at genus level. rRNA 16S sequencing supported biochemical identification in only two isolates (4.2%) at species level, and in 26 isolates (54.2%) at genus level. The eight isolates for which MALDI-TOF MS did not manage to identify, or the identification obtained was rejected by sequencing, belonged to species that are still not added to the BioTyper II data library. Conclusions: Results obtained in this study show that, overall, MALDI-TOF MS identification of anaerobic bacteria is more reliable than identification obtained by conventional biochemical methods (24% more correct identifications at species level). The number of major errors (incorrect identification at the genus level) is also 2.5-times lower. Moreover, all the major errors obtained by MALDI-TOF MS were due to the absence of some species in the data library. Thus, when data libraries are more complete, reliability differences between both methods will probably be even higher. (c) 2011 Elsevier Espana, S.L. All rights reserved.