Evaluierung und Etablierung von Analyseverfahren zur Differenzierung der Spezies V. cholerae Non-O1/-O139 und V. mimicus mittels MALDI-TOF MS in einem Routinelabor der amtlichen Lebensmittelüberwachung

Evaluation and establishment of analytical methods for the differentiation of the species V. cholerae Non-O1/-O139 and V. mimicus by means of MALDI-TOF MS in a routine laboratory of official food monitoring

The aim of this work was to differentiate the closely related species V. cholerae and V. mimicus quickly and reliably, using MALDI-TOF MS and the available standard softwares (flexAnalysis, MALDI Biotyper OC) and their tools. Later on it was possible to get an additional software (ClinPro Tools) and an extra Vibrio-specific database (VibrioBase) to also include them into the tests.
First, 20 (ten per species) of 62 Vibrio-Isolates were selected to examine them for potentially species-discriminating peaks in the flexAnalysis software. An internal calibration of the spectra to a genus common peak was profitably, because they were constistently lined up and scatter was minimized. Noticeable peaks had to pass the selection criteria to become evaluated as potentially species-discriminating. Then these peaks were validated using the previous 20 and another 42 isolates. A sensitivity of 97% and a specificity of 100% for V. cholerae were calculated and 93% and 97% for V. mimicus.
With the software tools of MALDI Biotyper OC no discriminating peaks could be detected. But the species could easily be distinguished with the dendrogram- and PCA-function, if control isolates were included. In parallel, spectra of the 42 above-mentioned isolates were created and compared using direct transfer (BRUKER DALTONIK GmbH, 2013b) and formic acid extraction (BRUKER DALTONIK GmbH, 2013a). It quickly became apparent that the direct transfer saves an important amount of work, resources and time.
Later it was discovered with the software ClinPro Tools that the previously found discriminating peaks were determined much faster. Especial using the function to generate your own models, in the future direct transfer could result in the possibility of obtaining a result for a questionable V. cholerae-mimicus isolate (from agar plate to result) within about 15 minutes. However, ClinPro Tools had the limitation, that spectra of the same isolates or of very closely related species, e.g. V. metoecus, were not noticed and could not be differentiated. With this problem, felxAnalysis turned out to be time-consuming, but because of the very precise spectrum analysis it turned out to be more suitable.
At a later point in time, eleven additional isolates were provided by the BfR, for which contrary species results have been obtained in previous examinations. These isolates were examined in the same way as the previous ones. In all three used softwares and their applications they got the result V. mimicus. This was later confirmed by the BfR via genome sequence analysis and it supports the applicability of the differentiation, used in this work, of closely related bacterial species by MALDI-TOF MS.
Finally, the VibrioBase database was added to MALDI Biotyper OC. The aim was totest, if it could be used to differentiate the investigated species better than the previously available databases (BDAL and SR). V. cholerae isolates were identified 100% correctly with both the VibrioBase (+ SR) and Bruker BDAL databases (+SR). V. mimicus isolates were identified 89% correctly and 11% incorrectly as V. cholerae or questionable. While the BDAL database (+SR) was only able to differentiate about 20% of the V. mimicus isolates correctly. The isolates of the species V. metoecus found by chance in this work always received the wrong result V. mimicus, because this species was not deposited in the databases.
To rapidly differentiate routine isolates in the future, it is recommended to first process V. cholerae-mimicus questionable isolates by direct transfer and matched with Vibrio-specific databases, e.g. VibrioBase. If the user does not have such databases available, they can create their own models in ClinPro Tools using the peaks at 4025 Da, 7969 Da, and 9122 Da for V. cholerae and 7958 Da, 9141 Da, 9477 Da and 9981 Da for V. mimicus and classify the isolates in a short time. Otherwise, these peaks can be searched in flexAnalysis after the spectra have been internally calibrated to the peak at 4279 Da. The species V. metoecus could not be differentiated because only two isolates were available. However, an experienceduser may be able to recognize the different nature of these spectra compared to V. cholerae and V. mimicus spectra.
To determine specific peaks for other species with very similar spectra, e.g. V. metoecus, in the future, it is recommended to grow the examined isolates under standardized conditions and to process them with the formic acid extraction. The search for specific peaks should first be performed in ClinPro Tools e.g. using the Total Average Spectrum, the Gelview function, ROC analysis or a combination of them. Appropriate models can then be generated. A rapid check in flexAnalysis should still be performed to identify unusual isolates. If ClinPro Tools is not available, visual comparison in flexAnalysis for stand-alone peaks or peak shifts is possible. An internal calibration of all spectra, e.g. to a genus-specific peak, should be performed.