UNLABELLED

Identification of uncommon fungi using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-ToF-MS) remains challenging. Its performance depends on the protein extraction method, instrument, algorithm, and database. This study compared two established mass spectrometers (VITEK MS [bioMérieux] and Microflex [Bruker]) to the new VITEK MS PRIME (bioMérieux) and four databases, including MSI-2, a non-commercial database constructed with spectra acquired with Bruker instruments. Isolates of species, rare molds, and uncommon yeasts, previously identified by sequencing, were analyzed. After a two-step protein extraction, MALDI-ToF-MS identification was performed, and spectra were submitted to diagnostic (IVD) (KB3.2 and KB3.3 [bioMérieux]), research use only (RUO) (FilFungi V5 [Bruker]) databases and MSI-2. A total of 169 isolates (61 spp., 72 rare molds, and 36 uncommon yeasts), representing 33 genera and 96 species, were included. Identification rates at species level for all fungi were similar between VITEK MS and VITEKMS PRIME (79% vs. 77%). The main difference lies in the rates of non-analyzable spectra, being 15% (26 strains) for VITEK MS PRIME versus only 3% (5 strains) for VITEK MS. For and rare molds, MSI-2 performed equally well on VITEK MS and Microflex (92% vs. 91%), indicating a spectra compatibility. For uncommon yeasts, all databases performed equally at the species level. For and rare molds, FilFungi V5 performed poorly (23% and 21%), while MSI-2 was best (77% and 82%) due to broader species coverage. Misidentifications mostly involved cryptic species. MALDI-ToF-MS is a powerful tool for identifying rare fungi. Improvements are needed in completing commercial databases and optimizing acquisition systems for fungal spectra.

IMPORTANCE

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-ToF-MS) is key for fungal identification nowadays. We present here the largest comparison of MALDI-ToF-MS instruments and databases focusing on rare yeasts and molds identification, challenging the existing instruments and both commercial and academic databases. The strains were collected in Saint Louis Hospital (Paris) and identified using bar-code sequencing. We showed that commercial databases are efficient for the identification of the main fungal species tested, whereas the academic database outperforms them for the identification of cryptic species.