Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) for yeast identification is limited by the requirement for protein extraction and for robust reference spectra across yeast species in databases. We evaluated its ability to identify a range of yeasts in comparison with phenotypic methods.
MALDI-TOF MS was performed on 30 reference and 167 clinical isolates followed by prospective examination of 67 clinical strains in parallel with biochemical testing (total n = 264). Discordant/unreliable identifications were resolved by sequencing of the internal transcribed spacer region of the rRNA gene cluster.
Twenty (67%; 16 species), and 24 (80%) of 30 reference strains were identified to species, (spectral score ≥2.0) and genus (score ≥1.70)-level, respectively. Of clinical isolates, 140/167 (84%) strains were correctly identified with scores of ≥2.0 and 160/167 (96%) with scores of ≥1.70; amongst Candida spp. (n = 148), correct species assignment at scores of ≥2.0, and ≥1.70 was obtained for 86% and 96% isolates, respectively (vs. 76.4% by biochemical methods). Prospectively, species-level identification was achieved for 79% of isolates, whilst 91% and 94% of strains yielded scores of ≥1.90 and ≥1.70, respectively (100% isolates identified by biochemical methods). All test scores of 1.70–1.90 provided correct species assignment despite being identified to “genus-level”. MALDI-TOF MS identified uncommon Candida spp., differentiated Candida parapsilosis from C. orthopsilosis and C. metapsilosis and distinguished between C. glabrata, C. nivariensis and C. bracarensis. Yeasts with scores of <1.70 were rare species such as C. nivariensis (3/10 strains) and C. bracarensis (n = 1) but included 4/12 Cryptococcus neoformans. There were no misidentifications. Four novel species-specific spectra were obtained. Protein extraction was essential for reliable results.
Citation: Pinto A, Halliday C, Zahra M, van Hal S, Olma T, Maszewska K, et al. (2011) Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry Identification of Yeasts Is Contingent on Robust Reference Spectra. PLoS ONE 6(10): e25712. https://doi.org/10.1371/journal.pone.0025712
Editor: Markus M. Heimesaat, Charité, Campus Benjamin Franklin, Germany
Received: August 17, 2011; Accepted: September 8, 2011; Published: October 13, 2011
Copyright: © 2011 Pinto et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: These authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Yeast infections cause significant mortality in critically ill and immunocompromised patients. In particular, Candida species are the 4th most common cause of nosocomial bloodstream infections in the United States, and Cryptococcus neoformans, the commonest cause of fungal meningitis worldwide , . Whilst Candida albicans remains the leading pathogenic yeast, infections due to non-C. albicans species such as Candida glabrata, as well as previously rare opportunists such as Trichosporon and Geotrichum species, are increasingly reported , –. Novel pathogenic Candida species such as Candida nivariensis and Candida bracarensis have also been described . Since many non-C. albicans Candida and non-Candida yeasts are resistant or less susceptible to antifungal agents, rapid accurate species identification is central to timely, effective antifungal therapy –, , .
Conventional phenotypic-based methods for yeast identification, however, are slow (24–72 h), insensitive and often unable to identify more unusual species. Various molecular techniques including real-time PCR, DNA sequence analysis, microarray analysis, and fluorescence in-situ hybridization provide accurate identification ,  but are expensive, require substantial specimen processing time (hours to a day) and are not easily implemented as routine techniques in the clinical laboratory.
Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) has emerged as a powerful and rapid tool for the identification of bacterial and yeast pathogens –. Using MALDI-TOF MS, the protein spectral “fingerprint” of an isolate is compared to a reference spectral database for identification . Previous studies have reported species identification rates of 92–99% amongst collections of yeasts and yeast-like organisms –, , . However, taken collectively, the results may not be directly comparable since these studies have used different approaches to assign species – from comparing spectra from test organisms to reference spectra in MALDI-TOF MS databases , , to enhancing these databases, and in some instances developing study-specific databases, with “in-house” spectral signatures , . There also is a continuing need to extend the repository of reference spectra in MALDI-TOF MS databases.
We therefore undertook the present study to evaluate the utility of the MALDI Biotyper 2.0 Microflex LT spectrometer (Bruker Daltonik GmbH; Bremen, Germany) with its current spectral database in comparison with phenotypic-based methods for the identification of a broad range of yeasts in a diagnostic laboratory. The first part of the study comprised testing of reference and clinical strains from our culture collection. The second was a blinded prospective analysis of freshly-collected yeast isolates recovered during routine laboratory work flow; discrepant results between MALDI TOF MS and phenotypic methods were resolved using sequence analysis of the fungal internal transcribed spacer (ITS) regions , . Results obtained by preparation of yeasts by extraction of fungal proteins and by direct application of yeast colonies onto the MALDI TOF MS plate were also compared.
A total of 264 yeast isolates were studied. These included 30 reference strains (26 species) obtained from the American Type Culture Collection (ATCC; Rockville, MD) and the Centraalbureau voor Schimmelcultures (CBS; Utrecht, The Netherlands) (Table 1) and, 167 clinical isolates (26 species) from the culture collections at the Centre for Infectious Diseases and Microbiology Laboratory Services, Westmead Hospital; and the Department of Microbiology, Liverpool Hospital, Sydney, Australia (Table 2).
For the second part of the study, 67 consecutive (57 Candida, and 10 non-Candida, yeasts) isolates (Table 2) isolated from a variety of clinical specimens over a 2-month period were prospectively identified by MALDI-TOF MS in parallel with conventional biochemical methods. All isolates were subcultured onto Sabouraud's dextrose agar (SDA; Difco Laboratories; Detroit; MI) for 48 hours at 30°C to ensure purity prior to testing.
Standard biochemical identification
All clinical isolates in the first part of the study were identified by the API ID 32C (bioMerieux, Hazelwood, MO) or RapID Yeast Plus systems (Remel Products, Lenexa, KS), with recommended additional tests where appropriate , . Phenotypic identification of isolates in the prospective study was performed using one or more of the following tests: (i) Candida isolates - germ tube test, appearance on CHROMagar™ Candida (CHROMagar™. Paris, France), RapID Yeast Plus or API ID 32 or Vitek 2 (bioMerieux) identification systems; (ii) cryptococcal isolates – urease assimilation, API ID 32 system, and appearance on canavanine glycine bromothymol blue agar . All other yeasts were identified by the RapID Yeast Plus and/or API ID 32 identification systems.
DNA extraction, PCR amplification and DNA sequencing
Sequencing of the ITS 1, 5.8S rRNA and ITS 2, or the D1–D2, regions of the 28S subunit of the rRNA gene, was performed as required to verify species identity of reference strains (Table 1) whilst ITS sequencing was undertaken in all cases of discrepant identifications or “no identifications” obtained by MALDI-TOF MS and phenotypic identification methods , . DNA extraction was performed using the High Pure DNA Template Preparation kit (Roche Diagnostics, Mannheim, Germany), according to the manufacturer's instructions.
PCR amplification of the entire ITS (ITS1, 5.8S, and ITS2) region was performed as previously described using the primers ITS1 and ITS4 . PCR products were purified using the GFX PCR DNA and Gel Band Purification kit (GE Healthcare UK Limited; Little Chalfont, Buckinghamshire) and commercially sequenced using the ITS 1 primer and the BigDye Terminator v. 3.1 cycle sequencing kit in the ABI PRISM 3100 genetic analyzer (Applied Biosystems, Foster city, CA). Sequences were edited using Chromas v. 2.23 software (Technelysium Pty. Ltd.) and entered into a BLASTN search  provided by GenBank (http://blast.ncbi.nlm.nih.gov/Blast.cgi) for species identification. A percent similarity (identity) score of ≥98% between the unknown (query) sequence and the database sequence of closest match was used as the criterion to classify an isolate to species level .
MALDI-TOF mass spectrometry
After incubation of test strains for 48 h at 30°C on SDA, 2–3 yeast colonies were suspended in 300 µL molecular grade deionized water and vortexed. Next, 900 µL of 100% ethanol (Sigma-Aldrich, St Louis, MO) was then added, vortexed, and centrifuged at 15 000 g for 2 min; the supernatant was decanted and the centrifugation repeated once. The pellet was dried at 25°C and reconstituted in equal volumes of 70% formic acid (Sigma-Aldrich) and acetonitrile (Sigma-Aldrich) (in 50 µL volumes), mixed thoroughly, followed by centrifugation (15 000 g) for 2 min. One microliter of supernatant was spotted onto a 96-spot steel plate (Bruker Daltonik) and allowed to dry at room temperature before addition of 1 µL of MALDI matrix (a saturated solution of α-cyano-4-hydroxycinnamic acid (Bruker Daltonik)). Each sample was tested in quadruplicate to ensure reproducibility of spectra. An Escherichia coli protein extract test standard (Bruker Daltonik) was used to calibrate the instrument and a negative extraction control included for each run. For a subset of clinical isolates (n = 88) in the first part of the study, the direct colony method was also used whereby yeasts grown on SDA were applied directly as a thin film to the steel plate and allowed to dry at room temperature before MALDI-TOF MS analysis.
Mass spectra were generated with the Microflex LT mass spectrometer (Bruker Daltonik GmbH) using the manufacturer's protocol. Briefly, ≈240 laser shots per sample spot were acquired using the recommended instrument settings for organism identification (linear positive mode, 60 Hz laser frequency, 20 kV acceleration voltage, 16.75 kV IS2 voltage, 170 ns pulsed ion extraction delay, and 2,000 to 20,137 m/z range). Manual and automated spectrum processing and species identification were performed with the MALDI-Biotyper 3.0 application and software (version 220.127.116.11; Bruker Daltonik GmbH). The software compares acquired sample spectra to reference spectra in the provided database and compiles a list of best matching database records. Identification scores were interpreted according to the manufacturer's recommended criteria: a score of ≥2.0 indicated species-level identification, a score of 1.70–1.999 indicated identification to the genus level, and a score of <1.70, was interpreted as “no reliable identification”.
For each organism, three of four MS tests within the same category of score were required for a definitive result. Failures (ie. if < three tests were of the same score, or if “no reliable identification result” was obtained) were retested. A major error was whenever the resolved final genus identification differed from that proposed by MALDI-TOF MS while a result was a minor error when the genus identification was concordant but the species name was incorrect.
Initial experiments on 88 archived clinical isolates showed that extraction of yeast proteins (see Methods) was essential for accurate MALDI-TOF MS identifications. Using the direct colony preparation method, correct species-, and genus-level identifications were obtained for 14% and 69% of C. albicans isolates (total tested, n = 36), 35% and 74% of C. glabrata (n = 23 tested) and 13% and 38% of C. lusitaniae (n = 8 tested). Analysis of the remaining isolates inclusive of nine C. kefyr and 12 C. parapsilosis strains produced no reliable spectra. Protein extraction of yeasts was subsequently always performed.
Correct species-level identification according to the manufacturer's criteria (spectral score ≥2.0) was obtained for 20 of 30 reference strains (16 species) including all common Candida species - C. albicans, C. glabrata, Candida tropicalis, Candida parapsilosis and Candida krusei , ) -, C. neoformans and Cryptococcus gattii (Table 1). One isolate each of Candida colliculosa, Candida orthopsilosis, Candida rugosa and Pichia ohmeri was resolved to “genus-level” ie. score of >1.70 (but <2.0) yet yielded the correct final species designation. MALDI-TOF MS distinguished between C. parapsilosis, Candida metapsilosis and C. orthopsilosis although identification of the last was to “genus-level” (Table 1).
There were no minor errors and no misidentifications (major errors). However, five species (n = 6 isolates) produced spectral scores <1.70 with “no reliable identification”. Species which provided the closest spectral-based match for these isolates are shown in Table 1. Reference spectra for three species, Candida ciferii, Pichia kluyveri and Debaromyces carsonii, were not present in the Biotyper software. The remaining two species (Arxiozyma telluris, Candida zeylanoides) failed to match their respective species-specific reference spectra. However, for A. telluris CBS 2685 (Table 1), the closest spectral match identified the strain as Candida sloofii; both A. telluris and C. sloofii are species reduced to synonymy .
Overall, measured against current reference spectra in the Biotyper version 18.104.22.168 library (Bruker Daltonik GmbH), concordance with DNA sequence analysis in identification to species-, and genus-level was 67% (20/30 isolates) and 80% (24/30 isolates), respectively. For species included in the database, MALDI-TOF MS resolved 74% (20/27) of isolates to species-level identification.
Clinical isolates: Retrospective study
(i) MALDI-TOF MS identifications.
One hundred and forty of 167 archived isolates (84%) yielded spectra which fulfilled the criteria for species-level identification, and 160 (96%) isolates, for genus-level identification (Table 2). Collectively, Candida pathogens yielded a higher percentage of spectral scores of ≤2.0 (127 isolates; 86%) than non-Candida yeasts (68%).
The mean best (ie. closest)-match score values for Candida species in general, showed little intra-species variation. Table 3 summarises these data for the major Candida species (all with mean score values of ≥2.0). C. nivariensis had a lower mean score value (1.81+/−0.31; Table 3). Amongst Candida isolates, four Candida utilis, three Candida dubliniensis, two C. glabrata, two Candida lusitaniae, two C. nivariensis, and a single isolate each of Candida gulliermondii and C. parapsilosis had spectral scores of 1.7–2.0 yet the resolved species identification was correct. Thus, 142/148 or 96% of isolates were resolved to correct species at a score of ≥1.7. For non-Candida yeasts, MALDI-TOF MS enabled correct identification of Cryptococcus spp. and other yeasts including Geotrichum and Lodderomyces elongisporus. Five of 19 (26.3%) isolates including four Cryptococcus isolates were resolved to the correct species but with scores of 1.72–1.79 (Table 2).
(ii) MALDI-TOF MS versus conventional identification.
For Candida spp., MALDI-TOF MS and phenotypic-based identification provided correct species-level identification for 127 (85.8%) and 112 (75.6%) isolates, respectively (data not shown). Of 17 discordant identifications, 16 were resolved in favour of MALDI-TOF MS using ITS sequencing. The MALDI Biotyper was able to identify C. nivarineisis (n = 5 isolates), Candida famata (n = 2), C. lusitaniae (n = 2) and Candida haemulonii (n = 1), with scores of ≥2.0, where biochemical methods yielded a result of “Candida spp.” Biochemical methods also erroneously classified three strains of C. parapsilosis as C. lusitaniae, and a single strain each of C. guilliermondii, C. glabrata and Candida sphaerica as other Candida species. Conversely, they identified a C. parapsilosis isolate that yielded no identification by MALDI-TOF.
Concordance with biochemical identification to genus level was present for 26/27 (97%) for non-Candida isolates (data not shown). Phenotypic methods were unable to identify Geotrichum candidum, Geotrichum silvicola and L. elongisporus, all successfully identified by MALDI-TOF MS, yet speciated all six Cryptococcus spp., of which only two yielded spectral scores of >2.0.
(iii) Incorrect and no identification.
There were no major errors by MALDI-TOF MS. Six isolates (3.6%) yielded a result of “no identification” (score of<1.7). These were three C. nivariensis, one C. bracarensis, one C. guilliermondii and one A. telluris isolates (Table 2). Reference spectra for C. bracarensis are not present in the Bruker database version 22.214.171.124 (Bruker Daltonik).
Clinical isolates: Prospective study
Table 2 summarises the identification results for 67 yeast isolates (13 species) by conventional methods, MALDI-TOF MS and the final species designation as determined by ITS sequencing. Biochemical methods identified all 67 isolates. There were no major errors in identification by MALDI-TOF MS.
Four of the six cryptococcal isolates studied were not identified by MALDI-TOF MS criteria (scores 1.3–1.5; Table 2). Assignment to species level at a score of ≥2.0 was achieved for 53 (79%) isolates (100% concordant with biochemical identification methods). Ten isolates including four C. glabrata, one C. albicans and one C. parapsilosis, had spectral scores of >1.70 but <2.0; single isolates of Candida kefyr (mean score 1.93), Rhodotorula mucilaginosa (mean score 1.96) and Saccharomyces cerevisiae (mean score 1.97) also yielded similar spectral scores (Table 2). All genus-level identifications resolved isolates to their correct species designations.
The time to issuing a definitive identification result by phenotypic tests after recovery of the isolate on culture was 24–72 h whilst the average hands-on time per isolate for MALDI-TOF MS identification after culture isolation was approximately one hour including sample preparation.
Whilst the utility of the MALDI-TOF MS is established for the efficient, routine identification of bacteria , , its position in assisting with, or replacing, phenotypic identification of yeasts in the clinical mycology laboratory is less well defined. The present study re-affirms and extends the results of previous reports of the ability of the Bruker Microflex LT (Bruker Daltonik) MALDI-TOF system to provide fast and accurate identification of a broad range of yeasts. Key findings were that (a) identification was genus-as well as species- dependent, (b) there were no major errors, including in the absence of reference spectra for a species, resulting in misidentifications, and (c) isolates with spectral scores between 1.70 and 1.90 were all correctly resolved.
We first validated MALDI-TOF MS for yeast identification using well-characterised reference strains in comparison with ITS sequencing. To more accurately represent the utility of the MALDI Biotyper 2.0 instrument with its current database, we did not add new spectral entries to the database. Although not all isolates achieved species-level identification using manufacturer-recommended criteria, all common Candida species ,  were unambiguously assigned (Table 1). Uncommon Candida spp. with genus-level identifications were also resolved to their correct species and importantly, “misidentifications” only associated with scores of <1.70 and therefore not given an identification. Whilst we cannot explain why C. zeylanoides CBS 619 (spectra contained in the Bruker library version 126.96.36.199), was not identified, lower scores may result from poor extraction technique, inadequate drying of the yeast pellet or inadvertent inclusion of agar into the extract. However, the scores were reproducible between runs and high scores (≥2.0) were achieved for common yeasts (Tables 1, 2, 3) analysed within the same run using identical protein extraction. Lower scores for uncommon yeasts have been noted by others –. The ability to provide species-level identification is dependent on the number of entries per species in the manufacturer's database, usually being higher for common isolates. A higher number of entries for the same species (as for common yeasts) will better reflect diversity within the species due to possible variation in protein expression between strains ; such intraspecies variation in fungi which may impact on protein spectral profiles warrants further study.
Examination of a large number of clinical strains (n = 167) confirmed the observations for reference strains. When measured against ITS sequence analysis, MALDI-TOF MS performed better than biochemical tests (84% of isolates correctly identified with spectral scores of ≥2.0 vs. 76.4%). Other studies examining yeast profiles following protein extraction have noted that lowering the score threshold for species-level identification to ≥1.80 or even ≥1.70 can still be used to provide reliable identification , . In our study, a score cut off of ≥1.70 allowed correct identification of 96% of isolates, and Candida isolates, lowering the threshold from ≥2.0 to ≥1.70 increased species-level identification by an additional 12%. Despite these findings, however, analyses of large numbers of different yeast species across multiple centres using standardised extraction methods are essential prior to defining a consensus cut-off score. Mean score values for common Candida species (eg. C. albicans, C. glabrata) were ≥2.0 with small standard deviations reflecting good result reproducibility, and likely, the number of spectra from these species already contained in the database. C. nivariensis yielded lower mean score values with greater variance (1.81+/−0.31); reproducibility is likely to improve with additions of reference spectra into the database (currently, n = 2 in the database). Five species of non-Candida yeasts were profiled with scores of ≥2.0 yet four of six Cryptococcus strains had scores of <2.0 (≥1.70; see below).
The Bruker database (version 188.8.131.52) used in our study was a more recent version containing a larger number of spectral entries compared with that in three other studies – and logically, may be expected to provide better identification rates. Possible explanations for the apparent reduced ability of MALDI-TOF MS to identify isolates to “species level (84%) in comparison with previous reports (87–100%) include differences in the range or mix of yeast species, and nature of the database employed. In studies where complementation of the manufacturer's database with in-house spectra or where the reference database was specifically constructed for the purpose of the study, species-level identification rates have ranged from 92.5–100% , , . Others using the Bruker database without supplementation have reported similar identification rates to those observed herein , .
A novel finding was the ability of MALDI-TOF MS to distinguish within the C. glabrata clade ie. between C. glabrata and C. nivariensis and C. bracarensis, which presently rely on molecular methods for species separation , , . Spectra from C. nivariensis were clearly distinguished from those of C. glabrata. Seven of 10 clinical isolates had spectral scores of >1.8 (five were >2.0; Table 2); possible reasons for the lower scores for the remaining three include suboptimal protein extraction despite every care with the manufacturer's extraction protocol, or variability in protein expression within this species. Spectra from the only isolate of C. bracarensis were novel. Members of the C. parapsilosis species complex, indistinguishable by biochemical methods , were also resolved by MALDI-TOF MS, including C. parapsilosis sensu stricto, C. metapsilosis and C. orthopsilosis , , . Quiles-Melero et al reported that these three species as assigned by pyrosequencing, yielded MALDI-TOF MS results with 100% concordance . We noted that a further member of the C. parapsilosis group, L. elongisporus, also yielded species-level identification. Differentiation of species within the C. parapsilosis and C. glabrata complexes is important since there are species-specific differences in antifungal susceptibilities and ability to form biofilms in vitro , , .
We further assessed in real time the efficiency of MALDI-TOF MS for routine yeast identification. Although only 79% of yeasts achieved species-level identification, all genus-level identification results (94%) yielded the correct final species designation. There were no major errors. Biochemical methods provided 100% species identification in this arm of the study as they identified four C. neoformans isolates that produced a score of <1.70 by MALDI-TOF (the “closest match” spectra still assigned the correct species). Unreliable identifications for C. neoformans have been reported by others (identification rates 50–66%) ,  and may be due to the polysaccharide capsule of this pathogen, rendering extraction and solubilisation of proteins difficult or due to insufficient database entries to enable spectral matches. We observed for Cryptococcus spp., that colonies with mucoid morphology posed technical difficulties with extraction, often requiring repeat analysis. Although McTaggart et al. reported species-level identification rates of 100% for Cryptococcus spp.  this was only after complementation of Bruker software with home-generated spectra.
The results of the present study have also provided novel spectra for four pathogens - C. ciferi, C. bracarensis, P. kluyveri, and D. carsonii - for future integration into the Bruker library. Spectra were acquired by testing each isolate in quadruplicate to ensure good reproducibility and on two separate occasions (data not shown) but require further independent validation by testing larger numbers of isolates.
Preparation for MALDI-TOF MS analysis can be undertaken in two ways. Although the direct colony method is used to identify bacteria, our attempts to adapt this for yeasts resulted in no spectra or unacceptably low scoring spectra in a large proportion of cases including for the most common pathogen, C. albicans, and there is now evidence that even for bacteria, extraction methods may be necessary . By adhering to the manufacturer's protocol for fungal cell wall disruption, reliable spectra were achieved in most instances. Other clues in assessing MALDI-TOF results include noting that the highest scores for species-level identification were seen when there was a large gap in score values or “score jump” between the most likely identification to the next most likely result (, this study).
MALDI-TOF MS offers considerable savings in time and cost over standard yeast identification methods. A single isolate may be identified in 45–60 mins including extraction time; up to 96 isolates may be identified within 3 h at AUD0.50 per sample and the protein extraction procedure is readily integrated into routine workflow. Phenotypic identification methods cost ≈ AUD10 per sample and take up to 48 h for definitive results. Costs associated with ITS sequencing where phenotypic methods cannot provide identification are ≈AUD9 per isolate (turn-around-time 48 h). The limitations of MALDI-TOF MS, however, are acknowledged. Robust reference spectra are required with regular curation. Other limitations are the high capital costs (≈AUD 250000) and inability to detect mixed cultures.
In conclusion, the rapid, accurate identification of pathogenic yeasts by MALDI-TOF MS is contingent on optimal protein extraction and upon robust reference spectra. Candida species were reliably identified, with superior performance of MALDI-TOF MS over conventional methods. However, MALDI-TOF MS was less reliable in distinguishing between C. neoformans and C. gattii compared with biochemical methods.
We thank Mr. Leith Fremlin (Bruker Daltonik, Australia) for helpful technical advice and Ms. Sue Sleiman for assistance with phenotypic isolate identifications.
Conceived and designed the experiments: AP SC-AC SvH WM JRI TO. Performed the experiments: AP CH MZ KM. Analyzed the data: AP SC-AC CH SvH JRI. Contributed reagents/materials/analysis tools: AP SC-AC KM CH SvH MZ TO. Wrote the paper: AP SC-AC CH SvH. Editing and approval of final version of manuscript: AP CH MZ SvH KM TO JRI WM SC-AC.
- 1. Pfaller MA, Diekema DJ (2010) Epidemiology of invasive mycoses in North America. Crit Rev Microbiol 36: 1–53.
- 2. Park BJ, Wannemuehler KA, Marston BJ, Govender N, Pappas PG, et al. (2009) Estimation of the current global burden of cryptococcal meningitis among persons living with HIV/AIDS. AIDS 23: 525–530.
- 3. Chen S, Marriott D, Playford G, Nguyen Q, Ellis D, et al. (2009) Candidaemia with uncommon Candida species: predisposing factors, outcome, antifungal susceptibility, and implications for management. Clin Microbiol Infect 15: 662–669.
- 4. Girmenia C, Pagano L, Martino B, D'Antonio D, Fanci R, et al. (2005) Invasive infections caused by Trichosporon species and Geotrichum capitatum in patients with haematological malignancies: a retrospective multicentre study from Italy and review of the literature. J Clin Microbiol 43: 1818–1828.
- 5. Atkinson BJ, Lewis RE, Kontoyiannis DP (2008) Candida lusitaniae fungaemia in cancer patients: risk factors for amphotericin B failure and outcome. Med Mycol 46: 541–546.
- 6. Ruan SY, Chien JY, Hsueh PR (2009) Invasive trichosporonosis caused by Trichosporon asahii and other unusual Trichosporon species at a medical centre in Taiwan. Clin Infect Dis 49: 11–17.
- 7. Cuenca-Estrella M, Gomez-Lopez A, Isla G, Rodriguez D, Almirante B, et al. (2011) Prevalence of Candida bracarensis and Candida nivariensis in a Spanish collection of yeasts: comparison of results from a reference centre and from a population-based surveillance study of candidemia. Med Mycol 49: 525–9.
- 8. Hsu DL, Nguyen M, Nguyen L, Law A, Wong-Beringer A (2010) A multicentre study to evaluate the impact of timing of caspofungin administration on outcomes of invasive candidiasis in non-immunocompromized adult patients. J Antimicrob Chemother 65: 1765–70.
- 9. Garey KW, Rege M, Pai MP, Mingo DE, Suda KJ, et al. (2006) Time to initiation of fluconazole therapy impacts mortality in patients with candidemia: a multi-institutional study. Clin Infect Dis 43: 25–31.
- 10. Lau A, Chen SCA, Sleiman S, Sorrell TC (2009) Current Status and Future Perspectives on Molecular and Serological Methods in Diagnostic Mycology. Future Microbiol 4: 1185–1222.
- 11. Shepard JR, Addison RM, Alexander BD, Della-Latta P, Gherna M, et al. (2008) Multicentre evaluation of the Candida albicans/Candida glabrata peptide nucleic acid fluorescent in situ hybridisation method for simultaneous dual-colour identification of C. albicans and C. glabrata directly from blood culture bottles. J Clin Microbiol 46: 50–55.
- 12. Eigner U, Holdelder M, Oberdorfer K, Betz-Wild U, Bertsch D, et al. (2009) Performance of a matrix-assisted laser desorption ionization time-of flight mass spectrometry system for the identification of bacterial isolates in the clinical routine laboratory. Clin Lab 55: 289–296.
- 13. Sauer S, Kliem M (2010) Mass spectrometry tools for the classification and identification of bacteria. Nat Rev Microbiol 8: 74–82.
- 14. Stevenson L, Drake S, Shea Y, Zelazny A, Murray P (2010) Evaluation of matrix-assisted laser desorption ionization-time of flight mass spectrometry for identification of clinically important yeast species. J Clin Microbiol 48: 3482–3486.
- 15. Marklein G, Josten M, Klanke U, Muller E, Horre R, et al. (2009) Matrix- assisted laser desorption ionization time of flight mass spectrometry for fast and reliable identification of Clinical Yeast Isolates. J Clin Microbiol 47: 2912–2917.
- 16. Dhiman N, Hall L, Wohfiel SL, Buckwalter SP, Wengenack N L (2011) Performance and cost analysis of MALDI-TOF mass spectrometry for routine identification of yeast. J Clin Microbiol. Available: doi: 10.1128/JCM.02381-10. Accessed 2011 Jul 20.
- 17. Bader O, Weig M, Taverne-Ghadwal L, Lugert R, Groß U, et al. (2010) Improved clinical laboratory identification of human pathogenic yeasts by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. Clin Microbial Infect. Available: doi: 10.1111/j.1469-0691.2010.03398.x. Accessed 2011 Jun 01.
- 18. Saffert RT, Cunningham SA, Ihde SM, Monson Jobe KJE, Mandrekar J, et al. (2011) Comparison of Bruker Biotyper MALDI-TOF mass spectrometer to BD Phoenix automated microbiology system for identification of gram negative bacilli. J Clin Microbiol 49: 887–892.
- 19. Ferroni A, Suarez S, Beretti JL, Daophin B, Bille E, et al. (2010) Real-time identification of bacteria and Candida species in positive blood culture broths by matrix-assisted laser desorption ionization-time of flight mass spectrometry. J Clin Microbiol 48: 1542–1548.
- 20. van Veen SQ, Claas ECJ, Kujiper EJ (2010) High throughput identification of bacteria and yeast by matrix-assisted laser desorption ionization-time of flight mass spectrometry in conventional medical microbiology laboratories. J Clin Microbiol 48: 900–907.
- 21. Leaw SN, Chang HC, Sun HF, Barton R, Bouchara JP, et al. (2006) Identification of medically important yeast species by sequence analysis of the internal transcribed spacer regions. J Clin Microbiol 44: 693–699.
- 22. White TJ, Burns T, Lee S, Taylor JW (1990) Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis MA, Gelfand DH, Sninsky JJ, White TJ, editors. PCR Protocols: A Guide to Methods and Applications. New York: Academic Press, Inc. pp. 315–322.
- 23. Hazen KC, Howell SA (2007) Candida, Cryptococcus, and other yeasts of medical importance. In: Murray PR, Barron EJ, Jorgensen JH, Landry ML, Pfaller MA, editors. Manual of Clinical Microbiology. Herndon, VA: ASM Press. pp. 1762–1788.
- 24. Kurtzman CP, Fell JW (1998) The Yeasts, A Taxonomic study. 4th Edition. B.V. Amsterdam, The Netherlands: Elsevier Science. pp. 77–99.
- 25. Kwon-Chung KJ, Polacheck I, Bennett JE (1982) Improved diagnostic medium for separation of Cryptococcus neoformans var. neoformans (serotypes A and D) and Cryptococcus neoformans var. gattii (serotypes B and C). J Clin Microbiol 15: 535–7.
- 26. Ciardo DE, Schar G, Bottger EC, Altwegg M, Bosshard PP (2006) Internal transcribed spacer sequencing versus biochemical profiling for identification of medically important yeasts. J Clin Microbiol 44: 77–84.
- 27. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, et al. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25: 3389–3402.
- 28. Petti CA, Bosshard PP, Brandt ME, Clarridge JE, Feldblyum TV, et al. (2008) Interpretive Criteria for Identification of Bacteria and Fungi by DNA Target Sequencing; Approved Guideline. pp. 54–59.
- 29. James SA, Collins MD, Roberts IN (2001) Phylogenetic analysis of the psychrophobic yeast Arxiozyma telluris and the reinstatement of Candida pintolopesii (van Uden) Meyer et Yarrow and Candida slooffii van Uden et doCarmo Sousa. Int J of Syst and Evol Microbiol 51: 1917–1925.
- 30. Carbonelle E, Mesquita C, Bille E, Day N, Duaphin B, et al. (2011) MALDI-TOF mass spectrometry tools for bacterial identification in clinical microbiology laboratory. Clinical Biochem 44: 104–109.
- 31. Lartigue MF, Gery-Arnaud G, Haguenoer E, Domelier AS, Schmit PO, et al. (2009) Identification of Streptococcus agalactiae isolates from various phylogenetic lineages by matrix-assisted laser desorption ionization-time of flight mass spectrometry. J Clin Micrbiol 47: 2284–2397.
- 32. McTaggart L, Lei E, Richardson SE, Hoang L, Fothergill A, et al. (2011) Rapid identification of Cryptococcus neoformans and Cryptococcus gattii by matrix assisted laser desorption ionization-time-of-flight mass spectrometry (MALDI-TOF MS). J Clin Microbiol 2011; 49: 3050–3053.
- 33. Correia A, Sampaio P, James S, Pais C (2006) Candida bracarensis sp. nov., a novel anamorphic yeast species phenotypically similar to Candida glabrata. Int J Syst Evol Microbiol 56: 313–317.
- 34. Borman AM, Petch R, Linton CJ, Palmer MD, Bridge PD, et al. (2008) Candida nivariensis, an emerging pathogenic fungus with multidrug resistance to antifungal agents. J Clin Microbiol 46: 933–938.
- 35. Tavanti A, Davidson AD, Gow NAR, Maiden MCJ, Odds FC (2005) Candida orthopsilosis and Candida metapsilosis spp. nov. To Replace Candida parapsilosis Groups II and III. J Clin Microbiol 43: 284–292.
- 36. Quiles-Melero I, Garcia-Rodriguez J, Gomez-Lopez A, Mingorance J (2011) Evaluation of matrix-assisted laser desorption.ionisation time-of-flight (MALDI-TOF) mass spectrometry for identification of Candida parapsilosis, C. orthopsilosis and C. metapsilosis. Eur J Clin Microbiol Infect Dis.
- 37. van Asbeck E, Clemons RV, Martinez M, Tong AJ, Stevens DA (2008) Significant differences in drug susceptibility among species in the Candida parapasilosis group. Diagn Microbiol Infect Dis 62: 106–109.
- 38. Alatoom A, Cunningham S, Ihde S, Mandrekar J, Patel R (2011) Comparison of Identification of Gram Positive Cocci Using Direct Colony versus Extraction Methods by Bruker Biotyper MALDI-TOF Mass Spectrometry. J Clin Microbiol. Available: doi:10.1128/JCM.00506-11. Accessed 2011 Jul 20.
- 39. Iranzo Tatay A, Frasquet Artes J, Fagundez Machiain G, Gobernado Serrano M (2011) MALDI-TOF assay in diagnosis of fungal infections. 21st ECCMID and 27th ICC Abstract No 750. Available: http://onlinelibrary.wiley.com/doi/10.1111/j.1469-0691.2011.03558.x/pdf. Accessed: 2011 Jul 20.