Identification by Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry and Antifungal Susceptibility Testing of Non-Aspergillus Molds

Non-Aspergillus molds including Mucorales, Fusarium, and Scedosporium, etc. are emerging pathogens leading to higher mortality in immunocompromised patients. Fifty-two isolates of genetically confirmed non-Aspergillus molds representing 16 species from 8 genera were collected to evaluate the performance of the Bruker matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) in identification of non-Aspergillus molds. Antifungal susceptibilities were determined through the Clinical & Laboratory Standards Institute (CLSI) M38-A2 broth microdilution method and the Sensititre YeastOne colorimetric method. Bruker MALDI-TOF MS identified 57.7% (30/52) of isolates cultured in broth and 15.4% (8/52) of isolates cultured on solid agar media to the species level, respectively, according to standard interpretation criteria. Lowering the species level cut-off value (COV) from ≥2.0 to ≥1.7 could improve the MALDI-TOF MS species-level identification rate to 67.3% (38/52) for isolates cultured on solid media, with a slight increase of false identification rate of 2.6% (1/38). Amphotericin B was the most in vitro fungistatic-active agent for 98.1% (51/52) of the tested non-Aspergillus molds, with minimum inhibitory concentrations (MICs) of ≤2 μg/mL. The susceptibilities to triazoles varied, with MICs of 0.12 to >16 μg/mL among different species of non-Aspergillus molds. The correlation between the CLSI method and Sensititre YeastOne on antifungal susceptibility testing of non-Aspergillus molds was good, with essential agreement (EA) rates of >90% for triazoles and echinocandins except amphotericin B, which had a lower EA rate of 84.6%. In conclusion, a favorable performance of the Bruker MALDI-TOF MS in identification of clinical non-Aspergillus isolates directly inoculated on solid agar media could be achieved with the adoption of alternative interpretation criteria. Antifungal susceptibility testing is important for non-Aspergillus molds, especially when information on triazole susceptibility is required, and the Sensititre YeastOne is a practical and reliable method to determine antifungal susceptibilities of non-Aspergillus molds.


INTRODUCTION
In recent years, opportunistic filamentous fungi have come to pose a serious threat to immunocompromised patients with AIDS, leukemia, organ transplantation, etc. Even though Aspergillus remains the leading pathogen in invasive mold infections, infections due to non-Aspergillus molds are increasingly reported in the clinical setting, with higher mortality rates. Mucorales and Fusarium are considered the most frequently clinically encountered non-Aspergillus mold pathogens, while other less common species such as Scedosporium, Paecilomyces, and the deeply thought contaminating pathogen from environment, Penicillium, are raising serious attention as a growing number of clinical infections due to those rare species have been reported (Slavin et al., 2015;Douglas et al., 2016;Lass-Flörl and Cuenca-Estrella, 2017). Morphological identification of non-Aspergillus molds is usually less effective, especially when species-level information is desired. Molecular sequencing analysis based on informative DNA targets such as the internal transcribed spacer (ITS) region could be competent in identification of non-Aspergillus molds, but the time-consuming aspect and need for specialized molecular biological equipment limited its wide use in clinical microbiology laboratories. Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) has shown excellent performance in identification of bacteria and yeasts (Singhal et al., 2015;Cassagne et al., 2016), and could also achieve accurate identification of Aspergillus with in-house databases and adjusted interpretation criteria (Li et al., 2017;Normand et al., 2017b). However, the practicability of MALDI-TOF MS in identification of non-Aspergillus molds is somehow restricted and relevant studies are much fewer compared with those on Aspergillus.
Non-Aspergillus molds demonstrated diverse susceptibilities against the clinically available antifungal agents, and the susceptibility profiles of those molds cannot always be inferred directly from their species information, even a large number of Mucorales and Fusarium isolates are notoriously resistant to multiple antifungals including triazoles and echinocandins (Wiederhold, 2017). Although the Clinical & Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) have both put forward reference methods, namely CLSI M38-A2 and E.DEF 9.3.1 documents, against conidia-forming molds, those methods have not usually been employed in clinical laboratories because of the shortcomings of complicated operation and difficult endpoint reading. The Sensititre YeastOne YO10 system is a commercial kit of antifungal susceptibility testing based on the broth microdilution method for yeasts and Aspergillus, and good performances have been achieved, especially for yeasts (Posteraro et al., 2015;Mello et al., 2017). Although non-Aspergillus molds are not listed as detectable species in the Sensititre YeastOne manual, several studies have applied Sensititre YeastOne to determine the antifungal susceptibilities of non-Aspergillus molds because it shares the same methodology as the CLSI M38-A2 method expect for its colorimetric endpoint reading feature (Carrillo-Muñoz et al., 2006;Patel et al., 2007;Halliday et al., 2016). Owing to the limited isolates and antifungal drugs involved in previous studies, there is no widely accepted conclusion about the practicability of Sensititre YeastOne on antifugal susceptibility testing of non-Aspergillus molds.
In this study, we evaluated the performance of the Bruker MALDI-TOF MS in identification of the clinically important non-Aspergillus molds grown under different culture conditions. Further antifungal susceptibility profiles of these isolates were determined through the reference broth microdilution method and the Sensititre YeastOne system, and the capability of the use of Sensititre YeastOne for non-Aspergillus molds was assessed.

Molecular Sequencing Identification
Isolates were inoculated on Sabouraud dextrose agar (SDA) plates (Oxoid, United Kingdom) at 28 • C for 2-7 days and mycelia were collected for genomic DNA extraction. The internal transcribed spacer (ITS) region was employed as the primary sequencing gene for all isolates. As to Fusarium, the eukaryotic translation elongation factor 1α (EF-1α) gene was used additionally for species level identification (Wang et al., 2011). Sequence data were analyzed against the National Center for Biotechnology Information (NCBI) 1 or Mycobank 2 database, and sequencebased species identification was defined by 99% sequence similarity with 95% query coverage.

MALDI-TOF MS Identification
Isolates were inoculated on SDA plates at 28 • C for 2-5 days until the colonies reached a size of about 2 cm diameter. The front mycelia were collected with a sterile inoculating loop and then transferred to 1.5-mL Eppendorf tubes containing 300 µL of distilled water and 900 µL of ethanol. The suspension was centrifuged at 16,000 × g for 3 min, and the pellet was dried at room temperature for 10 min. The pellet was then resuspended in 50 µL of 70% formic acid (FA) for 5 min and an equal volume of acetonitrile was then added. Samples were subsequently centrifuged at 16,000 × g for 2 min. One microliter of supernatant was transferred to the polished steel MTP 384 target plate (Bruker Daltonik, GmbH) and allowed to dry at room temperature before being overlaid with 1 µL of a saturated a-cyano-4-hydroxycinnamic acid (HCCA) matrix solution (Bruker Daltonik, GmbH). Meanwhile, spore suspensions with an approximate size of 1 × 10 6 colonyforming units (CFUs) from the colonies on SDA plates were reinoculated with 10 mL of Sabouraud dextrose broth (SDB) at 28 • C for approximately 1 day in a rotator as the Bruker standard operating procedure suggested. Then 1 mL of funguscontaining medium was transferred to a 1.5-mL Eppendorf tube and centrifuged at 16,000 × g for 3 min, and the pellet was washed twice with 1 mL of deionized water. The remaining operation was identical to the aforementioned standard FAacetonitrile extraction method applied in this study. The acquisition and analysis of mass spectra were performed by a Bruker Autoflex TM LT mass spectrometer using the MALDI Biotyper software package (version 3.1, Bruker Daltonik, GmbH), and the Filamentous Fungi Library 1.0 (Bruker Daltonik, GmbH) was used as the analysis database. Identification scores of ≥2.0 indicated species-level identification, scores of 1.7-1.999 indicated genus-level identification, and scores of <1.7 were considered unreliable according to the Bruker manual. We also evaluated the performance of the Bruker MALDI-TOF MS in non-Aspergillus identification by lowering the specieslevel identification cut-off value (COV) from ≥2.0 to ≥1.7 and the genus-level identification COV from 1.7-1.999 to 1.4-1.699 followed by reinterpretation of the identification records.

Antifungal Susceptibility Testing
Susceptibility testing was carried out using the broth microdilution method according to the CLSI M38-A2 document. The drugs tested were amphotericin B [AMB; National Institute for Food and Drug Control (NIFDC), China], itraconazole (ITR; NIFDC, China), voriconazole (VOR; NIFDC, China), posaconazole (POS; Merck, United States), caspofungin (CAS; Merck, United States), and micafungin (MF; Astellas, Japan). Drug concentration ranges for AMB, ITR, VOR, and POS were 0.015-16 µg/mL; and 0.008-8 µg/mL for CAS and MF. Candida parapsilosis (ATCC 22019) and Candida krusei (ATCC 6258) were included as the quality control strains for each run as positive controls of the antifungals' potency. Minimum effective concentrations (MECs) of echinocandins and minimum inhibitory concentrations (MICs) of the other tested agents were taken visually after 24 h for Zygomycetes; for Fusarium and the remaining species, 48-h or 72-h reading were needed as the CLSI protocol recommended. The ranges of MICs or MECs were calculated for each species-antifungal agent combination. We also evaluated the practicability of the Sensititre YeastOne YO10 panel in antifungal susceptibility testing of non-Aspergillus molds, and the agreement between these two methods in susceptibility determination of non-Aspergillus molds was analyzed. Sensititre YeastOne testing was carried out according to the manual instructions. For AMB and triazoles, the colorimetric MICs were taken as the lowest concentration well when the antifungal solution changed from red (growth) to blue (100% growth inhibition) at 24 h or 48 h depending on the respective species. MECs of echinocandins were defined as the lowest drug concentration that allowed the growth of small, rounded, and degenerated hyphae at 24 h or 48 h regardless of the solution color change in the wells.
Essential agreement (EA) between the Sensititre YeastOne and CLSI M38-A2 on susceptibility determination of non-Aspergillus molds was considered when the MICs/MECs obtained with the two methods fell within 2 dilutions of the twofold dilution scheme. EA values of >90% were considered acceptable (Lamoth and Alexander, 2015).

Statistical Analysis
Statistical calculations were done using IBM SPSS statistics software, version 19 (SPSS Inc., Chicago, IL, United States). Differences between the tested COVs for MALDI-TOF MS identification were evaluated using the Wilcoxon signed-rank test, and P < 0.05 was considered statistically significant (Schulthess et al., 2014).

MALDI-TOF MS Identification
Applying the standard interpretation criteria recommended by the manufacturer, the Bruker MALDI-TOF MS identified 57.7% (30/52) of isolates cultured in SDB to the species level with scores of ≥2.0 and 34.6% (18/52) to the genus level with scores of 1.7-1.999, whereas for isolates cultured on SDA plates, only 15.4% (8/52) of isolates could achieve species-level identification and 57.7% (30/52) to genus level. Correct identification rates at the species level could be reached to 88.5% (46/52) for SDBcultured isolates and 76.9% (40/52) for SDA-cultured isolates by the Bruker MALDI-TOF MS regardless of the score values obtained after comparison with the sequencing identification results. There were 11.5% (6/52) of SDB-cultured isolates and 21.1% (11/52) of SDA-cultured isolates identified correctly at the genus level by Bruker MALDI-TOF MS, and most of them had lower scores of <1.7. Only one isolate of Mucor hiemalis cultured on SDA was totally misidentified at the genus level, with a score of 1.316. After lowering the interpretation criteria, i.e., the species-level COV from scores of ≥2.0 to ≥1.7, the species-level identification rates increased significantly to 92.3% (48/52) (Z = −3.169, P = 0.002) and 73.1% (38/52) (Z = −3.224, P = 0.001) for SDB-and SDA-cultured isolates, respectively. For isolates directly grown on SDA plates, the reanalyzed species level identification rate with the lowered COV was comparable to the rate of isolates cultured in broth with the original criteria, namely the Bruker recommended procedure (73.1% vs. 57.7%). Unfortunately, the adjusted COV also brought species-level false Mucor hiemalis (n = 1), M. indicus (n = 1), Rhizomucor variabilis (n = 1), and Rhizopus azygosporus (n = 1) lack reference spectra in the Bruker library. Numbers in italic and underlined refer to isolates misidentified at the species level. a Bruker criteria for the species level is a score ≥2.0; genus-level score is 1.7-1.999. b Adjusted criteria for the species level is a score ≥1.7; genus-level score is 1.4-1.699. identification cases; 6.3% (3/48) of isolates cultured in SDB and 2.6% (1/38) of isolates on SDA with scores of ≥1.7 turned out to be falsely identified at the species level but with the correct genus information, and most of those misidentified cases were from species without reference spectra in the Bruker database (Table 1).

AFST
The distributions of MICs and MECs of 52 isolates of non-Aspergillus molds against six antifungals by the CLSI method are shown in

Agreement Between the CLSI Method and Sensititre YeastOne
The essential agreement (EA) data between the CLSI method and the Sensititre YeastOne on antifungal susceptibility determination of non-Aspergillus molds are presented for each genus-drug combination in Table 3. EA values were >90% for the majority of genus-drug combinations; however, the total EA between the CLSI and Sensititre YeastOne to AMB was 84.6%, because the MIC values obtained by the Sensititre YeastOne were generally higher than the CLSI ones especially in Mucorales. For triazoles, acceptable EA values of >90% were reached between those two methods for all the three tested drugs. EA values for echinocandins were excellent in this study, at 100%.

DISCUSSION
Identification and antifungal susceptibility testing of non-Aspergillus molds are rarely performed as routine tests in most microbiology laboratories. In this study, we identified 52 isolates of non-Aspergillus molds from clinical samples through molecular sequencing analysis and MALDI-TOF MS. According to the Bruker MALDI-TOF MS recommended procedure, 57.7% (30/52) of isolates cultured in broth could be identified to the species level with scores of ≥2.0. In contrast, for isolates directly grown on solid agar media (e.g., SDA), a widely used culture method in microbiological laboratories, only 15.4% (8/52) of isolates could achieve species-level identification. This discrepancy of the identification capacities of the Bruker MALDI-TOF MS might be attributed to the different elements grown under those two culture methods, as short-term broth culture mainly yielded mycelia while solid agar culture generated abundant conidia and hyphea within a medium-sized colony. A previous study indicated that young colonies and mature colonies of the identical mold isolate could present some obvious differences in MS spectra (Alanio et al., 2010). It is supposed here that MS spectra of non-Aspergillus molds from colonies on solid agar might not be well matched with the reference spectra in the Bruker Filamentous Fungi Library created with the mycelia from liquid broth culture. A short incubation time of about 2 days and sampling of the front hyphea from the young colonies will be favorable for the identification of non-Aspergillus molds grown on solid agar media by the Bruker MALDI-TOF MS system. After lowering the species-level identification COV from ≥2.0 to ≥1.7, the species-level identification rate was increased significantly to 73.1% (38/52) for solid agar-cultured isolates and only one isolate (2.6%, 1/38) was misidentified at the species level, which is comparable with the rate (57.7%) of the Bruker MALDI-TOF MS recommended parameters, namely the broth culture of isolates and the standard COV of ≥2.0. Most of the remaining isolates that failed to be identified at the species level under the adjusted COV, due to the lack of the speciesspecific reference spectra in the Bruker database, were correctly identified at the genus level. Many studies have established inhouse databases expanded with the rare species lacking reference spectra in the Bruker commercial database and improved the identification performance of MALDI-TOF MS remarkably. The mass spectrometry identification (MSI) platform is an outstanding public database featured with the online MALDI-TOF MS identification application and an extensive in-house reference database comprising 938 fungal species (Normand et al., 2017a;Stein et al., 2018;Dupont et al., 2019;Imbert et al., 2019). Notice should be paid that the sample preparation should be in accordance with the recommended method when using these public databases.
Besides the optimization of the comprehensive databases and the interpreting COVs, there were other studies concerning the simplification of the conventional protein extraction method in MALDI-TOF MS identification of filamentous fungi. Most of them have applied the bead grinding procedure plus a short incubation time, with the extraction solution containing formic acid and acetonitrile in one step, which provided labor-and time-saving advantages without increasing the misidentification rates (Luethy and Zelazny, 2018). Attention does need to be paid to biosecurity, especially when performing the bead-based grinding of molds.

±2
dil.  (100) 21 (100) 21 (100) 21 (100) Paecilomyces spp.  In spite of the absence of authorized clinical breakpoints or epidemiological COV for antifungal susceptibility testing of non-Aspergillus molds, the antifungal susceptibility results could provide a certain guide for clinical treatment. In this study, the antifungal susceptibilities of 52 isolates of non-Aspergillus molds were analyzed by the reference CLSI broth microdilution method, and some underlying profiles were detected even with the limited isolates of each species available for testing. For Mucorales, all isolates presented lower MICs of ≤1 µg/mL to AMB, which suggests good treatment effects according to the suggestion that a cut-off of 0.5 µg/mL for AMB among Mucorales is associated with better outcomes (Lamoth et al., 2016). The tested triazole agents have various susceptibilities, with MICs range from 0.12 to >16 µg/mL for the non-Aspergillus molds isolates in this study, and these diverse susceptibility profiles could be observed in isolates from the same species, which supports the importance of testing each individual isolate to guide clinical therapy. Almost all the Mucorales isolates had MICs of >16 µg/mL to VOR and relatively higher MICs of >2 µg/mL to ITR and POS, indicating the difficulty in treatment of mucormycosis. The newly developed extended-spectrum triazole isavuconazole (ISA) is recommended as a salvage therapy of invasive mucormycosis, even the extreme MICs of >16 µg/mL were also detected in some isolates according to the previous research (Miceli and Kauffman, 2015).

Mucorales
A rapid test of the intended antifungal susceptibility of a pathogenic fungus will be very helpful in patient treatment. Some MALDI-TOF MS-based assays offered new perspectives in defining the susceptibility pattern of Candida in a fast way, while controversy remains when it comes to molds such as Aspergillus (Gitman et al., 2017;Delavy et al., 2019). Since the laborsaving and commercially available Sensititre YeastOne panel holds the same operating principle as the CLSI M38-A2 method in antifungal susceptibility testing except for the endpoint reading patterns, the possibility of determining susceptibility of non-Aspergillus molds by Sensititre YeastOne has been discussed. In our study, the essential agreements within 2 dilutions between the CLSI M38-A2 and Sensititre YeastOne results on triazoles and echinocandins against non-Aspergillus molds were favorable, with rates of >90%, which met the given criterion. For AMB, the EA rate was lower, 84.6%, between the two methods because Sensititre YeastOne gave higher AMB MICs than the CLSI method, and this phenomenon has been discovered in another study on Aspergillus (Wang et al., 2018). In the matter of providing suggestive clues for the clinical treatment of infection with non-Aspergillus molds, the Sensititre YeastOne system could be competent, especially when information on the triazole susceptibility is required.

CONCLUSION
The Bruker MALDI-TOF MS has favorable performance in identification of non-Aspergillus molds directly from young colonies on solid agar media with the adoption of proper adjusted interpretation criteria and the additionally expanded MS database. The Sensititre YeastOne YO10 panel is a simple and reliable alternative to the reference broth microdilution