MALDI-TOF Mass Spectrometry Based Identification of Clinically Important Microorganisms

摘要

Introduction:Identification of microorganisms in the clinical laboratory is based traditionally on morphology and biochemical characteristics. For many species, identification may take days so a rapid species-specific method would improve patient care. Nucleic-acid based assays are sensitive and accurate methods for species identification but are relatively slow and labor intensive. We evaluated MALDI-TOF mass spectrometry for the rapid identification of microorganisms using Bruker Daltonics MALDI BioTyper software and either an in-house generated spectral library of 121 yeast reference strains or the manufacturer's library consisting mainly of bacteria. We correctly identified 85percent of 117 yeast, 184 bacterial, and 47 blood isolates. The MALDI-TOF based method was found to be highly reproducible and accurate with low consumable cost and minimal preparation time. Methods: 105 yeast reference strains (8 genera, 40 species), 115 yeast clinical isolates (6 genera, 11 species), 319 bacterial clinical isolates (59 species), and 131 positive blood culture isolates (59 species) were assayed. A single colony from an agar culture plate or concentration of organisms from a positive blood culture broth was washed in ethanol, dissolved in formic acid and acetonitrile, and spotted 4 times on a steel target with saturated alpha-cyano-4-hydroxycinnamic acid as the matrix. MALDI-TOF spectra were accumulated in positive mode from each sample across the 2000 to 20,000 m/z range on a Bruker UltraFlex TOF-TOF. The spectra were evaluated using the MALDI BioTyper software. Total time from colony to results was <30 minutes for a single isolate. Preliminary Data: Using the MALDI BioTyper software from Bruker Daltonics, a yeast library was developed using 115 clinically important reference strains. When multiple isolates of a species were available, we verified that all isolates had nearly identical spectra. When challenged with 115 known clinical isolates, 109 isolates (93percent) were correctly identified. Of the 6 discrepancies, 5 were identified correctly but scored below significance levels, three were identified correctly only at the genus level, and one did not give usable spectra. MALDI-TOF spectra of 319 bacterial clinical isolates spanning 59 species (including Gram positive and Gram negative organisms) were evaluated using the manufacturer supplied library. Identification was correct for 267 isolates (83percent). Twenty-nine isolates (9percent) were identified correctly but scored too low to justify identification at the species level. Ten isolates were not identified due to poor spectra and two organisms were not in the library. Thirteen isolates gave mixed or incorrect results. Of 131 positive blood cultures evaluated against the manufacturers library, 75 (57percent) were identified correctly using a score of >2.0 and 92 (70percent) were identified correctly using a score of >1.8. There were four samples that were shown by biochemical methods to be mixed cultures. In all four cases, the MALDI-TOF method successfully identified one organism but failed to identify the second organism. Three cultures gave incorrect identification at the species level. The remaining cultures failed to produce usable spectra, most likely due to either ion suppression by free hemoglobin or low organism count. Although 10percent of the 434 bacteria and yeast samples did not produce sufficiently detailed spectra to give acceptable scores for identification purposes, there were only thirteen false identifications (3percent). This combination of 97percent specificity and 86percent sensitivity compares favorably to current techniques but MALDITOF is considerably faster and cheaper, thus making MALDI-TOF identification of microorganisms a preferred method for the routine identification of microorganisms. Considerations for using this method as a primary method for organism identification in clinical samples will be discussed.