BACKGROUND: The increase in the amount of extended spectrum beta-lactamases (ESBL)-producing gram-negative bacteria is seriously threatening human health in recent years. Therefore, it is necessary to develop a rapid and reliable method for identification of ESBLs. The purpose of this study was to establish a novel method to discriminate between ESBL-producing and non- ESBL-producing bacteria by using the matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF-MS) technique.

MATERIAL AND METHODS: We detected hydrolyzed production of cefotaxime after incubation with 69 gram-negative bacteria by using MALDI-TOF-MS. Then we established genetic algorithm (GA), supervised neural networks (SNN), and quick classifier (QC) models using several peaks to identify ESBL-producing strains. To confirm the clinical applicability of the models established, a blinded validation test was performed in 34 clinical isolated strains.

RESULTS: Using ClinPro Tools software, we identified 4 peaks (456 Da, 396 Da, 370 Da, and 371 Da) in mass spectra of cefotaxime solution that have high enough specificity to discriminate ESBL-producing from non- ESBL-producing strains. Recognition capability of models established were 97.5% (GA), 92.5% (SNN), and 92.5% (QC), and cross validation rates were 90.15% (GA), 97.62 (SNN), and 97.62% (QC). The accuracy rates of the blinded validation test were 82.4% (GA), 88.2% (SNN), and 82.4% (QC).

CONCLUSIONS: Our results demonstrate that identification of ESBLs strains by MALDI-TOF-MS has potential clinical value and could be widely used in the future as a routine test in clinical microbiology laboratories.

Keywords: Bacteria - isolation & purification, Algorithms, Cefotaxime - chemistry, Electrophoresis, Gel, Two-Dimensional, Models, Biological, Molecular Weight, ROC Curve, Reproducibility of Results, Spectrometry, Mass, Matrix-Assisted Laser Desorption-Ionization - methods, beta-Lactamases - biosynthesis