Optimization of MALDI-ToF mass spectrometry for yeast identification: a multicenter study. (3rd October 2019)

• Record Type: Journal Article
• Authors: Normand, Anne-Cécile; Gabriel, Frédéric; Riat, Arnaud; Cassagne, Carole; Bourgeois, Nathalie; Huguenin, Antoine; Chauvin, Pamela; De Geyter, Deborah; Bexkens, Michiel; Rubio, Elisa; Hendrickx, Marijke; Ranque, Stéphane; Piarroux, Renaud
• Is Part Of: Medical mycology. Volume 58:Number 5(2020)
• Page Start: 639
• Page End: 649
• Publication Date: 2019-10-03
• Subjects: MALDI-TOF mass spectrometry -- Yeast identification -- human -- hospital -- Microflex; Medical mycology -- Periodicals; Veterinary mycology -- Periodicals; Mycology -- Periodicals; Mycoses -- Periodicals; Pathogenic fungi -- Periodicals; 616.969005
• DOI: 10.1093/mmy/myz098
• Languages: English
• ISSNs: 1369-3786

Abstract

Abstract: Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-ToF MS) is routinely used in mycology laboratories to rapidly identify pathogenic yeasts. Various methods have been proposed to perform routine MS-based identification of clinically relevant species. In this study, we focused on Bruker technology and assessed the identification performance of three protocols: two pretreatment methods (rapid formic acid extraction directly performed on targets and full extraction using formic acid/acetonitrile in tubes) and a direct deposit protocol that omits the extraction step. We also examined identification performance using three target types (ground-steel, polished-steel, and biotargets) and two databases (Bruker and online MSI [biological-mass-spectrometry-identification application]) in a multicenter manner. Ten European centers participated in the study, in which a total of 1511 yeast isolates were analyzed. The 10 centers prospectively performed the three protocols on approximately 150 yeast isolates each, and the corresponding spectra were then assessed against two reference spectra databases (MSI and Bruker), with appropriate thresholds. Three centers evaluated the impact of the targets. Scores were compared between the various combinations, and identification accuracy was assessed. The protocol omitting the extraction step was inappropriate for yeast identification, while the full extraction method yielded far better results. Rapid formic acid extraction yielded variable results depending on the target, database and threshold. Selecting the optimal extraction method in combination with the appropriate target, database and threshold may enable simple and accurate identification of clinically relevant yeast samples. Concerning the widely used polished-steel targets, the full extraction method still ensured better scores and better identification rates.