13.2.3.1 BacT/Alert 3D

The BacT/Alert system (bioMérieux, Durham, NC) is a continuously monitored closed blood-culture system for which no special fungal media are currently available. The system is based on the photochromatic detection of CO₂ as organisms grow in the broth. In 2004 Horvath et al. [139] used blood-culture bottles with blood from healthy donors and seeded with Candida spp. to compare recovery by the BacT/Alert (FA, SN, and MB+suppl) to the BACTEC 9240 (Plus Aerobic/F, Plus Anaerobic/F, and Myco/F Lytic bottles). Fifty clinical isolates, 10 of each species were used. Six bottles representing each medium received 10³ CFU (colony-forming units) of each strain of C. albicans, C. glabrata, Candida krusei, C. parapsilosis or Candida tropicalis and incubated per manufacturers’ instructions. Of the 300 bottles inoculated with Candida spp, 235 (78%) grew the corresponding strain. The remaining 65 bottles not flagged by the instruments by day 12 were subcultured and all of the corresponding organisms were grown. A breakdown of the number of positives detected by the instruments revealed that the BACTEC Myco/F, BacT/Alert FA, and BacT/Alert MB bottles recovered all isolates. This was in contrast to the BACTEC Plus Anaerobic (45 negatives), BacT/Alert SN (15 negatives) media, and the BacT/Alert SA (five negatives). Most laboratories do not include mycology media in the routine blood-culture protocol. When comparing only the aerobic and anaerobic media, the BacT/Alert performed better than the BACTEC in overall detection, time to detection, number of false-negatives, and missed episodes of simulated candidemia [139].

Although the BacT/Alert system had slightly diminished capacity for the detection of some fungi [181], use of the FAN aerobic media isolated significantly more yeasts than standard aerobic medium, with the same mean time to detection [313]. When compared to the standard anaerobic bottle, vented for aerobic use, the FAN bottle recovered significantly more Candida spp. [57]. In contrast, when the FAN anaerobe bottle was compared to standard aerobic medium, more C. glabrata isolates were recovered from the aerobic medium [117]. When the BacT/Alert system was compared to the BACTEC NR660 or 730 system, the BacT/Alert system had a noticeable but not statistically significant advantage in recovery of fungi; however, terminal subcultures of both systems added many of the fungi that otherwise would not have been recovered [322]. Saito et al. compared the FAN BacT/Alert media to standard media in a before-and-after study over a 4-year period. Although not focused on candidemia they did find that the FAN formulations had a higher percentage of positives for Candida spp. [267].

13.2.3.2 BACTEC

Various models of the BACTEC blood-cultures systems (Becton, Dickinson and Co., Sparks, MD) are used in clinical laboratories worldwide. Older models, based on the radiometric, interval-based detection of CO₂, have been replaced by nonradiometric (NR) systems and, most recently, by the BACTEC 9000 and FX systems, which use fluorescent sensor technology to continuously monitor CO₂ production. Barcode-scanning and data management systems are also included with these systems.

In a study of the BACTEC NR660 and Oxoid Signal systems (Remel), the BACTEC system proved to have the advantage in detection time. The NR660 system isolated 18 yeasts, with median times to detection of 39 h compared to 168 h for the Signal system [277]. In 1992, BACTEC PLUS high-volume resin (BP-HBV) and aerobic BP26 and anaerobic BP27 vials were compared to standard BACTEC aerobic NR6A and anaerobic NR7A vials. BACTEC PLUS vials enhanced the recovery rate for C. albicans for patients receiving antibiotics [184].

The BACTEC system has a slightly diminished capacity for detection of some fungi compared to the Isolator system. When BACTEC HBV was compared to the Isolator system and BACTEC BP26 with 93 isolates tested, HBV proved to be comparable, except for the recovery of H. capsulatum, where the Isolator system was superior. In contrast, C. glabrata was detected earlier in HBV than in the Isolator system. For HBV versus BP26, 68 positive fungal isolates were identified, and HBV was found to be superior for C. glabrata and all fungi combined. No difference was noted in the mean time to detection [321].

The BACTEC 9240 system was evaluated in comparison with the conventional BACTEC NR730 system using two media with the same formulation: NR26 for the BACTEC NR730 system and Aerobic Plus F for the BACTEC 9240 system. Simulated blood cultures were prepared for 41 strains belonging to 18 different fungal species, meant to represent fungi that are responsible for hematogenous dissemination. In a second phase of the study, two different media, HBV and NR26, were both compared on the BACTEC NR730 system. The differences between the two systems and the two media (NR26 and HBV) were not statistically significant, but use of HBV did improve the detection of fungi on the BACTEC NR730 system. In addition, the BACTEC 9240 system had a mean detection time that was significantly shorter than that of the BACTEC NR730 system [109].

In one of two 1998 studies, simulated blood cultures were used to compare the BACTEC 9240 medium (Mycosis IF/C) with the BACTEC standard medium, the Aerobic Plus F. In this study, 43 strains and 10 species of yeast were tested. The median time to detection was 29.03 ± 13.99 h for Mycosis IF/C compared to 73.92 ± 56.74 h for Aerobic Plus F medium. Mycosis IF/C reduced the time to detection [108]. In another study, specifically targeting C. neoformans, the MYCO/F medium appears to have an advantage over the Isolator system. Of ten blood cultures positive for C. neoformans, five were positive in both systems and five were positive in only the MYCO/F medium. The mean time to detection was 4 days in MYCO/F and 7 days with the Isolator system [312].

13.3.1.1 Bismuth sulfite glucose glycine yeast agar

Several studies have evaluated bismuth sulfite glucose glycine yeast (BIGGY) agar (Remel Microbiology, Lenexa, KS) for isolation and identification of yeasts. One [325] comparing CHROMagar to BIGGY agar examined 169 C. albicans, 33 C. tropicalis, 24 C. glabrata, 18 C. parapsilosis, 12 C. krusei and other Candida and yeast. They found CHROMagar sensitivity and specificity for C. albicans of 99.4%, 100% and for BIGGY 87%, 78.2%. The sensitivity for C. tropicalis, C. glabrata, and C. krusei on CHROMagar was 97%, 90.9%, and 100% and on BIGGY agar 66.6% (C. tropicalis) and 100% (C. krusei). In a different study [326] by the same authors examining 215 yeast isolates on BIGGY agar, CHROMagar, and albicansID2 found the same difference in sensitivity when comparing CHROMagar and BIGGY agar.

In 2007 a study compared BIGGY agar to albicans ID2 and Sabouraud agar with chloramphenicol for isolation and direct identification from vaginal cultures [143]. They did not find any difference in the number of yeasts isolated on each of these media. They found that BIGGY agar had a sensitivity of 86.7% and specificity of 56.3% compared to albicans ID2 sensitivity of 80% and specificity of 64.6% for C. albicans.

An additional chromogenic agar, HardyCHROM Candida (Hardy Diagnostics), is available, however, as of this writing there are no publications indicating performance relative to the other chromogenic media.

13.3.1.2 Eosin methylene blue agar

In testing with 92 isolates, all C. glabrata isolates grew better on eosin-methylene blue agar plates than on blood agar plates at 24 h. The criterion of better growth on EMB agar is used as an identification method for C. glabrata and is combined with the germ tube test as a cost-effective method for identification of C. glabrata and C. albicans [16].

13.3.2.1 Identification of unique enzymes

Traditionally, C. albicans has been identified by the germ tube test. In 1987, Perry and Miller evaluated the use of 4-methylumbelliferyl-N-acetyl-β-galactosamine (NAG) to differentiate between C. albicans and C. tropicalis [233]. Later, the enzymes L-proline aminopeptidase and β-galactosaminidase further increased the specificity of the test [234]. Candida albicans produces both enzymes, while other yeasts produce only one or neither.

The MUREX Candida albicans Test Kit (Remel Microbiology) test uses the enzyme substrates p-nitrophenyl-N-acetyl-β-D-galactosamide (NAG) and L-proline-β-naphthylamide, incorporating them onto a paper disk in a kit-provided test tube. The disk is moistened with distilled or deionized water, and a heavy inoculum of yeast is placed on it with an inoculating loop. The tube is capped and incubated at 35–37°C for 30 min, after which 0.03% sodium hydroxide is added. If one or both of the enzymes are present, the substrates will be hydrolyzed and a colored reaction will be noted. Candida albicans is positive for both enzymes; other yeasts may be positive for one or the other.

The BactiCard^TM Candida (Remel Microbiology) contains NAG and L-proline-β-naphthylamide; the latter is the same substrate found in the Murex system. These substrates are also dehydrated on a packaged test card, which is rehydrated prior to testing and inoculated with a thick paste of yeast cells. If the substrates are cleaved after 5 min of incubation at room temperature, one fluorescent reaction product, 4-methylumbelliferone, and one colorimetric reaction product will be produced. The fluorescent product is viewed under longwave UV light. The test can be performed on pure cultures that are 18–72 h old.

These tests were compared to the traditional germ tube test for presumptive identification of C. albicans [59]. The API 20 C AUX test (bioMérieux) and conventional methods were used as the gold standards. A variety of yeasts were tested, including C. albicans (303), C. glabrata (153), C. tropicalis (70), C. parapsilosis (36), Cryptococcus spp. (5), S. cerevisiae (3), and other yeasts (13). Murex C. albicans, BactiCard, and the germ tube test identified 98.7%, 99.3%, and 94.7% of C. albicans isolates, respectively. One false-positive each was observed using the Murex and BactiCard^TM Candida (Figure 13.8) tests, while two false-positives were seen with the traditional germ tube test. All of the enzymatic methods were noted to be rapid and accurate alternatives to germ tube testing of C. albicans. In another study [135) these tests were also used to test 133 yeast isolates in comparison with the germ tube test and the API 20 C AUX test and were found to correlate 100%. All of the assays generated some false-positive results for the non-Candida yeasts. The assays were compared for cost and other parameters.

The Murex C. albicans test was also compared with standard 2-h germ tube tests (n = 502). In this study, the number of C. albicans isolates tested was 316; the number of non-albicans yeasts tested was 186. Discrepant reactions were confirmed with the API 20 C AUX test and by chlamydospore production. Two yeast isolates, C. albicans and Candida lusitaniae, gave incorrect results with the Murex test. Compared to the germ tube test in fetal bovine serum and Fetal Clone II, the sensitivity and specificity of the Murex test were 94.6% and 97.8%, respectively [100].

The AlbiQuick (Hardy Diagnostics, Santa Maria, CA; Figure 13.9) is an assay that measures the presence of enzymes that will hydrolyze the proline-p-nitroanilide (PRO) and NAG compounds on a card with two reaction wells for the identification of C. albicans. No peer-reviewed studies have been reported as of this writing.

13.3.2.2 Methods for identification of C. glabrata

13.3.2.2.1 Rapid trehalose

In a 1996 study, a screening protocol for C. glabrata based on trehalose fermentation at 42°C for 24 h (Trehalose Fermentation Broth; Hardy Diagnostics) was compared to identifications by the API 20 C AUX and Microscan Rapid Yeast Identification Panel. Trehalose Fermentation Broth was found to have 97.8% sensitivity and 95.8% specificity when used for presumptive identification of C. glabrata [170]. A follow-up study [98] comparing three commercial methods for the rapid identification of C. glabrata included the 24 h Trehalose Fermentation Broth, the 3 h Remel Rapid Trehalose Assimilation Broth, and the 24 h Remel Yeast Fermentation Broth (Remel Laboratories) to the API 20C AUX. All of the methods were compared for accuracy, cost, and turnaround time. Sensitivities and specificities were 91.5 and 96.3% (Remel assimilation), 95 and 89% (Remel fermentation), and 96 and 100% respectively.

A rapid assimilation panel, GlabrataQuick (Hardy Diagnostics) includes maltose and sucrose in addition to trehalose, which reduces false-positive results (Figure 13.10).

13.3.2.3 Identification of C. dubliniensis

Identification of C. dubliniensis is a relatively new diagnostic dilemma, and few commercial systems have been tested. In one such study [112], 66 isolates of C. dubliniensis and 100 isolates of C. albicans were tested with the API 20 C AUX and the Vitek Yeast Biochemical Card (YBC) systems. The results were specifically focused on the abilities of the respective systems to identify xylose and α-methyl-D-glucoside (MDG) utilization by C. albicans and lack of utilization by C. dubliniensis. Growth at 45°C was also examined. The results suggested that clinical laboratories could use the lack of growth at 45°C and negative xylose utilization with either the API 20 C AUX or Vitek system to presumptively identify C. dubliniensis. Less correlation existed with a negative MDG test on either system, and misclassification may occur if this result is used, especially with the API 20 C AUX system. Ells et al. published a review of the reliability of methods for distinguishing C. dubliniensis from C. albicans [77].

Methyl blue Sabouraud agar contains 0.01% methyl blue as a supplement to Sabouraud agar [119]. This medium has been reported to distinguish between C. albicans and C. dubliniensis based on the yellow fluorescence of C. albicans under a Woods lamp [275].

13.3.2.4 Identification of C. neoformans and C. gattii

In 1962 Staib observed that when yeasts were grown on media containing Guizotia abyssinica seeds that Cryptococcus neoformans formed brown colonies while other yeasts were nonpigmented [287]. Subsequently it was determined that the seeds contain caffeic acid (3,4-dihydrocinnamic acid), which is the melanin-producing component. Phenol oxidase activity results in the brown pigment of C. neoformans and C. gattii colonies. Incubation requires up to 7 days before considering it negative. An attempt to circumvent this limitation was undertaken using a disk containing caffeic acid, which is placed on a glass slide, moistened with water or a nonglucose-containing agar, and inoculated with 5–6 colonies and incubated at 35°C in the dark for up to 4 h. The production of a brown pigment should be considered presumptive and the identification verified by another test. Also, it does not differentiate between C. neoformans and C. gattii. Both the caffeic agar and disks are available from Hardy Diagnostics and Remel.

The majority of clinical disease in immunocompromised patients is due to C. neoformans var. neoformans and C. neoformans var. grubii. The recognition of C. gattii as an emerging pathogen in immune competent individuals makes the differentiation of the two species important. Kwon-Chung et al. in 1982 reported a selective medium which differentiated between C. neoformans and C. gattii based on the ability of C. gattii to grow in the presence of canavanine, and uses glycine as a sole carbon and nitrogen source resulting in a shift in pH detected by bromothymol blue (CGB agar; Figure 13.11) [168]. This did not receive broad acceptance in clinical diagnostic laboratories outside areas where C. gattii had been initially described, e.g., tropical and subtropical regions [167]. More recent studies utilizing mostly molecular methods have demonstrated a much wider distribution of C. gattii in North America [167], making it more important to differentiate the species. In 2009 Klein et al. [163] compared the ability of CGB to differentiate between C. neoformans and C. gattii and a number of other yeasts. They found that 100% of C. gattii grew and turned the media blue (positive) while 100% of C. neoformans did not. They also found that other Cryptococci, Trichosporon spp., and some Candida spp. also would give a positive reaction on CGB agar.

A recent study evaluated Remel rapid urea broth and BD BBL urease test broth, caffeic acid disk, CGB, and internal transcribed spacer (ITS) sequencing in 147 reference and clinical strains of Cryptococcus spp., Candida spp., Trichosporon spp., and other non-Cryptococcus species [192]. All of the cryptococci were urease positive as well as Rhodotorula and Trichosporon spp. Addition of the caffeic acid result eliminated all of the noncryptococci with 85 of 86 (99%) C. neoformans positive and 26 of 27 (96%) C. gattii positive. Utilizing these two rapid tests, identification of C. neoformans/gattii was made in 4 h. Within 48 h further separation of the species was achieved using CGB and/or ITS sequencing. A positive CGB was found in 25 of 27 (93%) of C. gattii and none of the C. neoformans. It required ITS sequence to find the urea positive, caffeic-acid positive, and CGB negative C. gattii. The CGB agar is available from Hardy Diagnostics and Remel.

13.3.3.1 Uni-Yeast-Tek

The Uni-Yeast-Tek (Remel) system was the first widely used commercial system for yeast identification. It consists of a sealed multiwell plate, with each well containing medium for testing carbohydrate utilization, nitrate utilization, urease production, and morphology on cornmeal agar. The kit also includes supplementary tests, such as the C-N screen, glucose beef extract, and sucrose assimilation medium tubes. Evaluations of this product have shown it to be generally satisfactory for the identification of commonly isolated yeasts [254]. However, the system may require up to 7 days for complete identification.

13.3.3.2 API 20 C AUX

The API 20 C AUX (bioMérieux) test is a micromethod for carbohydrate assimilation of clinically important yeasts. Yeasts are inoculated into basal medium in test strips containing dehydrated substrates for assimilation testing. The API 20 C AUX system consists of a strip that contains 20 microcupules; all but two contain dehydrated substrates for determining substrate utilization profiles. As the yeasts utilize various substrates, the microcupules will appear cloudy (Figure 13.12). The reactions are compared to the first reaction cupule, which does not contain a carbohydrate substrate. The test strips are incubated at 30°C for 24–72 h. The results of assimilation reactions are read and converted to a seven-digit biotype profile number. In addition, the system requires that the microscopic morphology be documented on cornmeal agar containing 1% Tween 80 and that this information be used in conjunction with the substrate utilization data for yeast identification [254]. This yeast identification system is perhaps the most widely evaluated and well published of all the current commercial systems. It is frequently used as the gold standard against which comparisons are made.

In early studies [328] performed to compare the API 20 C AUX test with conventional methods, 98% correlation was documented (n = 45). The system was considered to be an acceptable substitute for conventional carbohydrate fermentation and assimilation tests. In a more recent study [276], 178 yeasts were tested. After 72 h, correct identification occurred for 86.5% of those isolates tested by the API 20 C AUX system compared to 86% with Auxacolor (Sanofi Diagnostics Pasteur), 68% with Mycotube (Roche Diagnostics), and 51.1% with Candifast (International Microbio, Milan, Italy). When 100 common yeasts were tested, a correct identification was obtained for 98% of the isolates with Vitek YBC (bioMérieux) and 100% with the API20 C AUX [126]. In another study, 123 common and 120 rare clinical yeast isolates were tested, and the API 20 C AUX test correctly identified 97 and 88%, respectively [256]. A further evaluation reported 90.8% of 120 yeast and yeast-like isolates were correctly identified by the API 20 C AUX test [132].

13.3.3.3 RapID Yeast Plus

The RapID^TM Yeast Plus system (Innovative Diagnostic Systems; sold as IDS YeastPlus Panel from Remel/ThermoFisher) is a qualitative method that uses conventional and chromogenic substrates to identify carbohydrate utilization, hydrolysis of fatty acid esters, and constitutive enzymes of yeast and yeast-like organisms. It combines conventional tests and single-substrate chromogenic tests with 4-h incubation. A suspension of pure yeast culture is made in the RapID inoculation fluid, turbidity is standardized, and the suspension is used to rehydrate the biochemicals in the RapID plastic reaction panel (Figure 13.13). With a one-step inoculation, the suspension will disperse in the panel by way of the plastic inoculating trough. After a 4-h incubation at 30°C, the RapID^TM Yeast Plus system is interpreted. Some reactions require additional reagents to be added to the reaction panel. The resulting pattern of positive and negative reactions is used to compile a profile number based on the organism’s individual reactions in the panel. The RapID^TM Yeast Plus Code Compendium supplies species identification for the biochemical profile numbers, which are generated based on the results of biochemical reactions. The RapID^TM Yeast Plus method correctly identified 94.1% of 304 clinical yeast isolates within 5 h [162].

In a study of 201 yeasts [311], when no further tests were performed, identification to the species level was 96%, 83%, and 86% for the RapID^TM Yeast Plus system, the API 20 C AUX system and the Vitek YBC respectively. The API 20 C AUX system did not correctly identify any C. krusei isolate without supplemental testing, and this organism accounted for most of the differences seen between the API 20 C AUX system and the RapID^TM Yeast Plus system.

In another study [135] the RapID^TM and API 20 C AUX systems were compared with 57 C. albicans isolates, 26 C. tropicalis isolates, 23 C. glabrata isolates, and 27 isolates from other species. The RapID^TM system correctly identified 125 (94%) of all yeasts tested and 99% (105 of 106) of the isolates of the three most common species, C. albicans, C. tropicalis, and C. glabrata. Discrepancies were resolved with Vitek 2 Yeast ID cards (bioMérieux), germ tube production, and microscopic morphology and confirmed that the RapID^TM system compared favorably with the API 20 C AUX system but had a more rapid identification process. In another study, the identification rate for yeasts tested with this system was 78–84% [38].

In a similar study, good correlation was observed between the two systems, with 95.7% agreement with the API 20 C AUX system for Candida and Cryptococcus. Lower agreement was noted with emerging Candida spp. and with other, yeast-like pathogens (79.1 and 75.2%) [85]. In another study, the RapID^TM Yeast Plus system was compared with the Uni-Yeast-Tek system (Remel). A total of 117 fresh and frozen yeast isolates were tested. The bioMérieux Vitek system and morphology were used to resolve discrepant results. The RapID^TM system correctly identified 96.6% of the isolates in 4 h [207].

13.3.4.1 MALDI-ToF/MS

Traditional microbiological techniques have long relied on the phenotypic characteristics of an organism including morphology, biochemical profiles, and culture patterns. Over the past several years, molecular techniques including genetic probes, polymerase chain reaction (PCR), and sequencing have become increasingly prevalent in microbial identification algorithms. Recently, MS has been adapted to determine the physical protein expression profile of an organism and compare it to known profiles in a database to identify organisms in patient samples. Mass spectrometry was previously used for bacterial identification in 1975 in a process that focused on lipid analysis. In the 1990s, MALDI-ToF/MS was used to identify surface proteins of bacteria. A change of matrix composition allowed the analysis of bacterial ribosomal proteins that is considered more reliable for microbial identification [274]. MALDI-ToF works by embedding analytes in a crystalline matrix. A laser is utilized to vaporize the analytes, which are then ionized and detected. MALDI-ToF/MS measures the mass-to-charge ratio of a given sample and compares it to a library of reference spectra to establish identification. Results can be obtained either from subculture or directly from blood cultures, and a protein extraction step may be utilized depending on the platform, especially for fungal organisms. This technology is useful for the identification of cultures containing a single species but is currently insufficient to detect mixtures of organisms [74]. The sample is placed with the matrix on a conductive metal plate that can then be inserted directly into the mass spectrometer. Obtained spectra results are compared to reference spectra within the database, thereby leading to identification of the organism. Currently, two commercially available MALDI-ToF/MS platforms have received the Conformité Européenne (CE) mark for in vitro diagnostic use in Europe. The Bruker Biotyper system, utilizing the Microflex LT mass spectrometer and Biotyper database (Figure 13.14; Bruker Daltonics), and the bioMérieux Vitek MS system (Figure 13.15), utilizing the MS-ID database (bioMérieux, previously developed by AnagnosTec and Shimadzu Corporation), are currently available for the identification of bacterial and fungal pathogens [307]. An additional Andromas database (Andromas SAS, Paris, France) can be utilized in conjunction with the Bruker system. Interpretation of MS data is conducted by algorithms such as the Bruker Main Spectrum analysis or the bioMérieux Advanced Spectra Classifier for the profiling of the presence or absence of peaks. At the time of this writing, the majority of published studies relating to fungal identification utilize the Bruker MALDI-ToF platform coupled to the Biotyper database. The Sepsityper system (Bruker Daltonics) can be utilized to identify organisms directly from blood culture [148]. For fungal organisms, a protein extraction step substantially improves identification for the Bruker system in combination with the Biotyper database; however a protein extraction step is not required for the Vitek MS system [144]. At the time of this writing, the bioMérieux Vitek MS system has received FDA approval for the diagnosis of 193 clinical bacterial and yeast organisms, while the Bruker platform has FDA approval for an aerobic Gram-negative organism database including 40 species and species complex (100 total identifications). Advantages of MALDI-ToF/MS include low cost of consumables, fast analysis turnaround times, and easy interpretation of identifications. The reader should consult current publications and vendor web sites for up-to-date information since this is an extremely rapidly changing field of study.

13.3.4.2 MALDI biotyper

A 2009 study compared MALDI-ToF/MS (Bruker Daltonics, Bremen, Germany) versus conventional identification methods including the API ID 32C (bioMérieux) for identification of 267 clinical yeast isolates composed of Candida, Cryptococcus, Saccharomyces, Trichosporon, Geotrichum, Pichia, and Blastoschizomyces species subcultured on Sabouraud agar [187]. Utilizing the provided Biotyper software, MALDI-ToF identified 247 isolates (92.5%) on the first round compared to 244 isolates in the first round for the API ID 32C. Following the addition of complementary reference spectra to the database for the remaining isolates, MALDI-ToF was able to identify all 267/267 isolates while the API ID 32 C identified 256/267 isolates in the second round. Another study in 2009 found that MALDI-ToF/MS showed promise in identifying fluconazole resistance in C. albicans isolates by assaying the minimum concentration of fluconazole that induces a detectable mass spectrum profile change. Results correlated with the Clinical Laboratory Standards Institute (CLSI) methodology of minimum inhibitory concentration (MIC) by broth microdilution [185]. Further studies need to be conducted to better define the efficacy and usefulness of MALDI-ToF in this context.

A 2010 study [308] compared MALDI-ToF/MS (Bruker Daltonics) to conventional techniques (Vitek2, API, and biochemical ) in the identification of 80 subcultured yeast isolates (19 retrospective and 61 prospective) covering seven genera and 12 species. In the prospective study, MALDI-ToF achieved 85.2% species identification and 96.7% correct genus identification compared to 77.0% species and 98.4% genus identification for routine biochemical methods. Misidentifications for MALDI-ToF were largely the result of an incomplete database that could be resolved with the addition of appropriate reference spectra. Stevenson et al. [288] conducted a study in 2010 where 192/194 (99%) of subcultured clinical yeast isolates were correctly identified using a Bruker MALDI-ToF with BioTyper software. One C. rugosa isolate (1/7) and one Cryptococcus neoformans isolate (1/3) could not be identified in this study. Low MALDI-ToF scores requiring repeat testing were observed commonly for isolates of C. guilliermondii, C. krusei, C. lusitaniae, and C. neoformans. Another 2010 study established a proof of concept study where various Candida species could be identified directly from blood cultures with appropriate sample preparation, thus allowing identification 1–2 days more quickly than if the isolates were subcultured [186]. Ferroni et al. [102] conducted a 2010 study in which the Bruker MALDI-ToF/MS system combined with the Andromas database (Andromas SAS) successfully identified 20/20 (100%) spiked blood cultures composed of 10 C. albicans isolates and 10 non-albicans Candida species in addition to 11/11 patient blood cultures containing C. albicans. Also in 2011, Bader et al. [13] analyzed 1192 clinical yeast and yeast-like isolates and compared two MALDI-ToF/MS platforms, the Bruker system/Biotyper database versus the AXIMA Assurance system/Saramis database (AnagnosTec/Shimadzu) along with traditional biochemical techniques (API 20 C AUX and ID 32 C systems from bioMérieux). With the Biotyper database, 97.6% of the 1192 isolates were correctly identified compared to 96.1% with the Saramis database and 96.9% for classic methods. The two MALDI platforms performed comparably and had lower rates of misclassification (Biotyper: 8 misclassifications, 21 unknown; Shimadzu: 2 misclassifications, 44 unknown) than classical techniques (30 misclassifications, 7 unknown). With supplementation of the MALDI databases, the rate of misidentifications and unknown assignments could be decreased. The MALDI-ToF/MS was much better at resolving C. orthopsilosis/metapsilosis/parapsilosis and C. glabrata/bracarensis than traditional biochemical approaches. Analysis with the Biotyper database required a protein extraction step while analysis using the Saramis database did not. Alanio et al. [2] constructed a MALDI database including the reference spectra of 28 clinically relevant species from seven Aspergillus sections comprised of five common and 23 unusual species. They then utilized the database to identify 140 previously characterized isolates (124 clinical and 16 environmental) following processing on a Microflex MALDI-ToF/MS spectrometer (Bruker Daltonics) with flex control software along with comparative data analysis using the Andromas SAS software. Out of the 140 samples, 127 were identified with one run, 11 were identified with two runs, and two samples could not be identified. Identified species included Aspergillus species from the Fumigati, Flavi, Terrei, Nigri, Nidulantes, Usti, and Circumdati sections. The two unidentified samples were ultimately identified as atypical Aspergillus fumigatus isolates (absence of conidiogenesis or atypical sporulation) by multilocus sequencing. With no misidentifications, MALDI was able to correctly identify 98.6% of tested isolates with 100% sensitivity utilizing the constructed database.

A 2011 study [284] compared the MALDI Biotyper MALDI-ToF MS system (Bruker Daltonics) to conventional techniques directly from blood cultures with the utilization of a protein extraction step. MALDI-ToF correctly identified 95.9% of C. albicans and 86.5% of non-albicans Candida species. MALDI exhibited difficulty in the identification of C. guilliermondii, C. krusei, and C. lusitaniae isolates as well as polyfungal bloodstream infections. Dhiman et al. [68] conducted a 2011 study comparing a Bruker MALDI-ToF to traditional phenotypic methods using 138 common and 103 archived yeast strains. MALDI-ToF correctly identified 96.3% of common and 84.5% of uncommon isolates using a score threshold of 1.8. Unreliable identifications were noted for C. guilliermondii, Aureobasidium/Hormonema, C. neoformans, Blastoschizomyces capitatus, C. colliculosa, C. inconspicua, C. albidus, and Trichosporon asahii. Low identification scores for Cryptococcus neoformans were thought to result at least partially from the organism’s thick capsule, and it was noted that reference spectra for C. gattii were not in the database. McTaggart et al. [193] addressed the issue of identifying Cryptococcus species in a 2011 study that analyzed 137 Cryptococcus strains and 23 non-Cryptococcus strains subcultured on Inhibitory Mold Agar by MALDI-ToF using intergenic spacer DNA sequencing as the reference method. Following supplementation of the Biotyper 2.0.1 database with C. gattii reference spectra, 27/27 C. gattii isolates were correctly identified. Overall, all 137 isolates were correctly identified to species level. A single C. neoformans var. neoformans (1/4) isolate was misidentified to subspecies level as C. neoformans var. grubii. Another 2011 study conducted by Pinto et al. [251] utilized the Bruker Biotyper MALDI-ToF/MS system to retrospectively analyze 30 reference and 167 clinical isolates as well as prospectively analyzing 67 clinical isolates subcultured to Sabouraud’s dextrose agar in parallel with biochemical methods, where discrepancies were addressed by sequencing of the internal transcribed spacer region. MALDI-ToF/MS successfully identified 80% of the 30 isolates to species level. Out of 167 clinical isolates, 140/167 (84%) were identified to species level (threshold score > 2.0), and 160/167 (96%) were identified to genus level (threshold score > 1.70). Prospective analysis of 67 clinical isolates yielded successful species-level identification for 79% of isolates. Protein extraction of yeasts was found to be necessary for the generation of reliable mass spectra as compared to the direct colony preparation method. MALDI-ToF exhibited difficulty with identification of C. ciferrii, C. zeylanloides, C. nivariensis, C. bracarensis, Pichia kluyveri, Arxiozyma telluris, Debaryomyces carsonii, and Cryptococcus neoformans. Another study conducted in 2011 [324] investigated the Bruker MALDI Sepsityper kit (Bruker Daltonics) designed for the processing of blood cultures with a dedicated lysis step that disrupts red blood cells while leaving the bacterial and yeast cell walls intact. A total of 42 blood cultures containing yeast were isolated with correct identification of all samples. Serial dilutions of a C. parapsilosis culture revealed the limit of detection of this MALDI-ToF system is 5.9 × 10⁵ colony-forming units (CFU).

In 2011 Bille et al. [24] investigated the Bruker MALDI-ToF/MS system along with the Andromas SAS database for the identification of yeast and Aspergillus species. MALDI-ToF successfully identified 96.3% of yeast and 92.2% of Aspergillus species following a single acquisition, with identification of 98.8% of yeasts and 98.4% of Aspergillus species from the second acquisition. Difficulty was found with identification of C. lambica (not in database). Utilization of MALDI-ToF with the Andromas SAS database could successfully classify 63/64 morphologic Aspergillus species into Fumagati, Terrei, and Flavi sections. Another study in 2011 [43] utilized the Bruker MALDI-ToF/MS system and a variety of extraction protocols for the identification of mold isolates. The most effective method was to grow molds on Sabouraud–chloramphenicol–gentamicin agar and conduct a formic-acid-based extraction step following mechanical scraping. Utilizing this process, this group identified 87% (154/177) of isolates. Identification failed in 12% (21/177) of isolates due to the absence of reference spectra in the database. Appropriate reference spectra were lacking for Alternaria, Penicillium, Scedosporium, Paecilomyces, A. clavatus, A. melleus, A. oryzae, A. sydowi, A. alliaceus, Geotrichum, and Fusarium species. One Beauveria bassiana isolate was not identified, and one Rhizopus oryzae isolate was misidentified as Mucor circinelloides.

MALDI-ToF/MS is also a promising technology for the identification of dermatophytes. Theel et al. [296] utilized the MicroFlex LT MALDI-ToF/MS system (Bruker Daltonics) to identify 100 well-characterized, archived dermatophyte isolates and 71 fresh dermatophyte cultures with either the supplied MALDI Biotyper library (MBL version 3.0) or a supplemented library containing an additional 20 dermatophyte species. Dermatophyte specimens were submerged in 70% ethanol, pelleted, and dried prior to suspension in 70% formic acid and 50% acetonitrile with subsequent matrix (HCCA, α-cyano-4-hydroxycinnamic acid) application and analysis. Out of 171 isolates, the provided Biotyper library identified 37.4% of isolates to the genus level and 20.5% to the species level while the supplemented database identified 93% of isolates to the genus level and 59.6% of organisms to the species level. It was also found that lowering cutoff score values to at least 1.5 for genus and 1.7 for species from the manufacturer-recommended values of 1.7 for genus and 2.0 for species increased the number of obtained identifications without increasing misidentifications. With the manufacturer-provided database, two isolates of Trichophyton mentagrophytes were misidentified as T. tonsurans while passing the 10% criterion for species identification. With the supplemented database, seven isolates of Trichophyton rubrum were misidentified as T. soudanense while passing the 10% criterion, however all T. rubrum isolates were correctly identified following deletion of the single T. soudanense reference spectrum from the supplemented database. A 2012 study [5] utilized the Andromas SAS software to identify 381 isolates of clinical dermatophytes and Neoscytalidium following culture on Sabouraud’s dextrose agar. Dermatophyte colonies were harvested in 70% formic acid and applied directly on a plate with subsequent matrix application. A reference database was constructed from a set of 50 reference strains from 12 clinically relevant species of dermatophytes and two species of Neoscytalidium. Utilizing MALDI-ToF/MS and the constructed database, 331/360 dermatophytes (91.9%) and 18/21 Neoscytalidium isolates (85.7%) were correctly identified. Two T. mentagrophytes var. interdigitale isolates were misidentified as either T. rubrum or T. tonsurans. In addition, spectral acquisition failed in 19% of Microsporum langeronii, 28% of T. soudanense, and 14% of Neoscytalidium isolates. Utilizing the Andromas SAS software, the authors were able to obtain higher levels of dermatophyte identification than observed results from a previous study utilizing the Bruker Biotyper database [296], and no colony extraction step was needed with the Andromas SAS software. Thus, with appropriate database reference spectra, MALDI-ToF/MS is a powerful method for the identification of dermatophytes.

13.3.4.3 Vitek MS

A 2012 study [280] compared the predecessor to the Vitek MS (bioMérieux, SA, Marcyl’Etoile, France), the AXIMA Assurance MALDI-ToF/MS system (Shimadzu Biotech) with Saramis database (AnagnosTec) to the API ID 32C system (bioMérieux) in identifying yeasts from dermatological patients. Sequencing of the internal transcribed spacer (ITS) regions of the ribosomal DNA was used for confirmation, and no extra protein extraction step was utilized. From a total of 83 isolates, MALDI-ToF results concurred with ITS sequencing in 94% of cases (78 isolates), while API ID 32C results concurred in 84.3% of cases. MALDI-ToF misidentified 1/35 isolates of C. albicans as C. glabrata and misidentified 1/10 C. glabrata isolates as C. tropicalis. One C. colliculosa isolate (1/1) was not identified by MALDI-ToF as there was no reference spectrum in the Saramis database. One Rhodotorula mucilaginosa isolate (1/1) and one Trichosporon mucoides isolate (1/1) were insufficiently identified to genus only. The API ID 32C was especially susceptible to the repeated misidentification of C. krusei isolates (5/7 isolates misidentified). In this study, MALDI-ToF/MS compared favorably to the API ID 32C system for the identification of dermatological yeasts. Iriart et al. [144] utilized the Vitek MS coupled with the “advanced spectra classifier” algorithm and version 1.0 of the MS-ID database for the identification of yeast-like and Aspergillus isolates compared to either the Vitek 2 (bioMérieux, St. Louis, MO) or routine laboratory identification techniques. Yeast-like isolates were applied directly from a Sabouraud agar plate to the Vitek MS disposable target with no protein extraction step. Following lysis with 25% formic acid and drying, α-cyano-4-hydroxycinnamic acid (CHCA) matrix was applied to the spot prior to analysis. For Aspergillus isolates, CHCA matrix was applied directly to the fungal spot under laminar flow with no protein extraction or formic acid lysis step. Out of 192 yeasts and yeast-like isolates, Vitek MS correctly identified 184 isolates (95.8%) to species compared to 183 isolates (95.3%) by routine laboratory tests and 169 (88.0%) by Vitek 2. For yeast organisms present in the database, Vitek MS failed to identify 1/60 (1.7%) of C. albicans isolates, 1/45 (2.2%) of C. parapsilosis isolates, and 2/2 (100%) of Geotrichum candidum isolates for an overall identification rate of 98.2%. In addition, Vitek MS failed to identify C. norvegensis and C. orthopsilosis isolates, as these reference spectra were not in the database. A single misidentification (major error) was made for an isolate of C. palmioleophila (not in database) being identified as C. haemulonii. Out of 44 Aspergillus isolates, Vitek MS correctly identified 36 (81.8%) and 39 (88.6%) were correctly identified by routine morphological identification. For Aspergillus species in the database, all A. fumigatus (33/33), A. flavus (2/2), and A. niger (1/1) isolates were properly identified. Isolates of A. sydowi (0/2), A. terreus (0/1), A. tubingensis (0/1), A. calidoustus (0/1), A. nidulans (0/2), and A. puniceus (0/1) could not be identified as these organisms were not in the Vitek MS database. The Vitek MS system allows for rapid high-quality identification of yeasts and Aspergillus species with only a single sample deposit and no protein extraction step. At the time of this writing, the Vitek MS system has received FDA approval for the identification of 193 clinical bacterial and yeast isolates. Continued supplementation of the provided MS-ID database and validation for additional organisms will be required to improve performance in the identification of clinically relevant fungal isolates.

Overall, MALDI-ToF/MS represents a new and promising technology for the identification of clinically important fungal organisms. Results can be obtained within approximately 1 h of culture, and the assay requires minimal consumable resources and technical manipulation. It is important to obtain more comprehensive databases to fully utilize the potential of this technology. MALDI-ToF/MS will likely replace many of the current frontline biochemical techniques, however, conventional methods will be required for antibiotic susceptibility profiling for the foreseeable future. The MALDI-ToF/MS technology exhibits difficulty in mixed specimens thereby requiring subculturing of specimens containing multiple organisms.

13.3.4.4 Rapid yeast identification panel

The Microscan Rapid Yeast Identification panel (Siemens Healthcare, West Sacramento, CA) is composed of chromogenic and modified conventional tests for identification of yeasts and yeast-like organisms, such as Prototheca spp., from clinical isolates. The 96-well plate format contains 27 dehydrated chromogenic substrates, which are designed to identify yeasts without the need for ancillary tests. A heavy suspension of a fresh yeast isolate, prepared in water and standardized to a known turbidity standard, is subsequently used to hydrate the plates. The plates are incubated for 4 h at 35–37°C before the reactions can be interpreted. The chromogenic readings can be made manually, read by eye, or with the Autoscan instrument. Interpretation of the biochemical profile occurs as the results are transformed into a nine-digit biotype number, which is compared with the Microscan cumulative database via the Autoscan instrument or manually, using the Microscan Rapid Yeast Biotype Codebook. The database identifies at least 42 yeasts and yeast-like organisms. In each case, profile identifications are assigned a cumulative relative probability of identification. When the probability of identification is not sufficient to confirm identification, supplemental testing of morphological and phenotypic characteristics is recommended.

When the Rapid Yeast Identification Panel was compared to the API 20 C AUX system, early studies reported 94.1% correlation in identifying species of Candida, Hansenula, Pichia, Rhodotorula, and Saccharomyces when results were determined after 4 h. A lower correlation (65%) was observed for slower-growing species, such as Blastoschizomyces spp., Cryptococcus spp., Trichosporon spp., and yeast-like organisms, such as Prototheca spp. The overall correlation, using the first limited database was approximately 85% [171, 286].

The same system was also tested with results obtained after 72 h. The results were again compared to those with the Microscan Rapid Yeast Identification Panel at 4 h and the API 20 C AUX system with cornmeal-Tween 80 agar at 72 h. Both visual and Autoscan readings of the plates correctly identified 78% of the 357 yeast isolates tested with no supplementary tests required. With supplementary tests, Autoscan identified 96.6% (99.5% of common strains; 92.1% of less common strains, like C. glabrata, C. fumata, C. lusitaniae, C. lambica, C. rugosa, C. albidus, C. laurentii, and C. uniguttulatus), and the API 20 C AUX system identified 98.9%. Autoscan was shown to be reliable for the identification of common yeasts when supplementary tests were performed [286].

More recently, when the results were compared to those obtained with the API 20 C AUX system and from morphology on Tween-80 agar, the Microscan Walk-Away-96 correctly identified 67–82% of the 150 yeasts tested. The most commonly misidentified yeasts were C. tropicalis, C. glabrata, and C. parapsilosis [264].

13.3.4.5 Vitek YBC/Vitek 2 YST

The Vitek YBC system (bioMérieux) is an automated system for yeast identification. The system uses the Vitek test card, which is composed of microwells containing various identification substrates. The YBC is used for the yeasts that are most commonly encountered in hospitals. Yeast suspensions are prepared in sterile water. Tubes are placed in the Vitek filler-sealer, which uses a vacuum to load the liquid into the Vitek test card through a sterile straw. The sealer module removes the straw and seals the test card. The card contains dehydrated medium for a variety of biochemical tests. The reader–incubator unit maintains a controlled environment while the cards incubate; it rotates the cards so that readings can be measured by the laser unit and photodetector. The yeast database contains at least 36 species of commonly encountered yeasts. Results are maintained so that cumulative reports can be generated and can be formatted according to the user’s needs. Add-on software is available for purchase to maintain workflow and quality control data. The maximum capacity for the Vitek Junior model is 30 cards per run. The Vitek Senior comes in models that can analyze 60, 120, and 240 cards simultaneously.

In an early study [76], the Vitek yeast database was compared to the API 20 C AUX system. The Vitek system compared favorably for the identification of common yeasts but less favorably for identifying others. For all yeasts identified, most of the identifications were made within 24 h of incubation.

In another report [70], the YBC was compared to the API 20 C AUX system with additional biochemicals for the identification of 222 germ-tube-negative yeast isolates. The YBC identifications of commonly isolated yeasts compared favorably; however, the 48-h identifications of less common yeasts were not as successful: 55% were correctly identified, and the remaining isolates were either misidentified or not identified. In a separate study [126] the system was not able to identify an isolate of S. cerevisiae or C. tropicalis but provided correct identification for 98% of the common yeasts (n = 100).

Similarly, the Vitek YBC, with 24–48-h results, was compared with the API 20 C AUX system and morphology on Tween-80 agar [263]. Of 150 yeast isolates, 85–95% were correctly identified with the Vitek YBC. The most common misidentified yeasts were C. tropicalis and C. glabrata. In another study [99], 409 germ-tube-negative yeasts and Geotrichum spp. were also tested. The results were compared with identification using the API 20 C AUX system and yeast morphology agars. Both systems identified most yeast isolates within 24 h. The YBC correctly identified 89.7% of 409 germ-tube-negative yeasts. The API 20 C AUX system correctly identified 99.3%. Of the remaining isolates, some profiles provided no identification of the organism, some profiles converted to misidentifications (the YBC incorrectly identified 7.3% of the isolates tested; the API 20 C AUX system misidentified only 0.7%). Results were comparable for C. glabrata, C. parapsilosis, and C. neoformans. The Vitek YBC had difficulty with C. tropicalis, C. krusei, Trichosporon, and some Cryptococcus spp. [97, 98, 100, 101].

A current instrument, the Vitek 2 system, allows identification of medically important yeasts and yeast-like organisms in 15 h by using a compendium of 47 biochemical reactions that are monitored by a sensitive fluorescence-based chemistry. The new database with 51 taxa was evaluated using the Vitek lD-YST card. Compared to the ID32C strip (bioMérieux) for 241 strains from 21 species, the Vitek 2 system correctly identified 92.1% of the isolates [120].

A second-generation yeast identification card (YST), the currently available card – converted from fluorescence-based to colorimetric chemistry, was introduced with the Vitek 2. The card has 64 wells and an expanded database of 53 species and 14 genera. There are several comparisons of the ID-YST and the new YST cards [9, 133, 177, 273, 304]. Loiez et al. [177] tested 172 strains representing 19 Taxa by the ID-YST, YST, and the API 32C system, which was considered the reference. The recorded concordance between the Vitek 2 cards and results with different identifications that could not be resolved by simple additional phenotypic testing were submitted for sequencing of the 18S–28S rDNA. Correct identifications to the species level were 93.6% for ID-YST and 83.7% for the YST card.

A different outcome was reported in another 2006 publication [9]. Ninety-seven isolates from CHROMagar Candida were compared with the ID-YST, YST, and API 20C as the reference identification. Their results show a superiority of the YST over the ID-YST, with correct identifications of 94.8% and 83.5% respectively. A 2007 study [273] evaluated the YST card against the RapID Yeast plus (Remel) system. A total of 750 clinical isolates were tested by both systems and rRNA ITS region sequencing as the reference standard identification. The Vitek 2 YST card and the RapID correctly identified 98.2% and 95.5% isolates, respectively.

13.3.4.6 BD Phoenix Yeast ID Panel

The BD Phoenix Yeast ID Panel (Becton Dickinson, Sparks, MD) is a recently introduced panel for their Phoenix Microbiology System. It utilizes modified, conventional, fluorogenic, and chromogenic substrates for the identification within 4–18 h. Early studies demonstrate high accuracy compared to other systems. The reader should refer to recent publications for this relatively new system.

13.3.4.7 Biolog YT Microplate

In the YT Microplate test (Biolog, Hayward, CA), yeast cells are suspended in sterile water. The YT Microplate is configured with both of its carbon source oxidation tests using tetrazolium violet as the colorimetric indicator. Carbon-based assimilation tests are scored turbidimetrically and can only be read with the automated Biolog MicroStation. A few wells contain two carbon sources and a test for colorization of various carbon sources with D-xylose. Software support exists for the Biolog Yeast Database. In addition, the SF-N and SF-P Microplates can be used for metabolic testing of sporulating and filamentous microorganisms, such as the actinomycetes and fungi. Special culture medium is used to enhance sporulation, and a gel-forming colloid, which keeps the organisms more evenly suspended, is used to prepare the organism suspension. The SF Microplates are identical to the GN and GP microplates used for aerobic bacteria except they do not contain the tetrazolium redox dye, which is toxic to many of these species. The plates are incubated at 25–27°C for several days and read with the automated MicroStation.

Sand et al. [272] compared the Biolog YT to the YBC and API 20 C AUX system with 171 germ-tube-negative yeasts; the Biolog system identified 48% of the isolates. The YBC and the API 20 C AUX system correctly identified 97% of the 171 strains tested. Both systems required repeat testing less than 10% of the time and required supplemental testing with 28% of the strains. With the Biolog system, there is a need for considerable expertise and a battery of supplemental reagents for correct identification of germ-tube-negative yeasts.

13.3.4.8 Microbial Identification System

The Microbial Identification System (MIS; Microbial ID, Inc., Newark, DE) relies on fatty acid analysis to identify yeast isolates. Cellular fatty acid chromatograms are analyzed and compared to the fatty acid profiles in the system’s database. Prior to analysis, yeast isolates must be subcultured, with subsequent saponification, methylation, and extraction of fatty acids. In one study using MIS (with the yeast clinical library version 3.8), 477 isolates and 23 different species of yeast were tested. The MIS generated results with a performance of 75% correct identifications compared to a combination of identification methods, including the Vitek YBC, morphology on cornmeal agar, and the API 20 C AUX system. Predictive values for different species ranged from 47 to 100%, depending on the species [155].

In another similar study [60], the MIS was compared to the API 20 C AUX system and conventional methods. The results indicated that 374 (68%; n = 550) yeasts were correctly identified to the species level. After repeat testing, 18 more were identified, bringing the overall correct identification to 71.3%. Of all yeasts tested, 15.8% were incorrectly identified and 16.2% were given no identification. The isolates tested represented a variety of species: C. albicans (294), C. glabrata (145), C. tropicalis (58), C. parapsilosis (33), and other yeasts (20). Candida glabrata was the most commonly misidentified yeast, and it was misidentified as S. cerevisiae 32.4% of the time. Another study performed with the MIS provided evidence that the results can be dependent on the supplier of the agar on which the yeast isolates are grown [156].