MALDI-TOF for Fungal Identification

Synopsis:

Many studies have shown successful performance of MALDI-TOF MS for rapid yeast and mold identification, yet very few laboratories have chosen to apply this technology into their routine clinical mycology workflow. Some of this hesitation may be attributable to the lack of standardization regarding culture methods, extraction methods, databases, and acquisition methods. This review will provide an overview of the current status of MALDI-TOF MS for fungal identification, including key findings in the literature, processing and database considerations, updates in technology, and exciting future prospects. Significant advances towards standardization have taken place over recent years; thus, accurate species level identification of yeasts and molds, particularly within species complexes and cryptic organisms, should be highly attainable, achievable, and practical in most clinical laboratories.

Introduction

The identification of yeasts and filamentous fungi (molds) has historically relied on a combination of phenotypic and morphological characteristics. Recent studies, however, have uncovered a vast array of species-complexes and cryptic species that can only be reliably identified by the use of more delineated and targeted testing platforms such a sequencing and matrix assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF MS). Around 2010, MALDI-TOF MS technology was launched into the clinical microbiology field, and was a game-changer as clinical laboratories could now identify organisms from crude protein suspensions within minutes, at very low processing costs, and minimal processing time. The accuracy of MALDI-TOF MS for organism identification was equivalent to DNA sequencing, and its ease of use led it to be a highly sought after platform in multiple fields including clinical microbiology, veterinary care, the food and beverage industry, and environmental microbiology. Currently, there are two main MALDI-TOF MS systems on the market - MALDI BioTyper (MBT; Bruker Scientific) and VITEK MS (BioMerieux). The MALDI BioTyper was cleared by the FDA in 2013 bacteria and yeast identification only; although a separate FilFungal database became available in 2012 available for mold identification for research use only. In contrast, the VITEK MS version 3.0 was cleared by the FDA in 2017 with a database that expanded bacteria, yeasts, mold, and Mycobacteria. The critical role of databases and processing methods will be addressed in this review as methods have evolved considerably over time.

Today, MALDI-TOF MS has become a mainstream platform in most clinical laboratories for bacterial identification. However, its application in clinical mycology for yeast and mold identification has been lagging, which is mostly attributable to the lack of standardized processes and poor fungal database representation. In fact, clinical mycology expertise is sorely lacking in laboratories worldwide and MALDI-TOF MS has the potential to fill this knowledge gap. In 2018, a survey of 348 tertiary care hospitals in China were reported to have insufficient clinical mycology testing capacity ¹. Similarly, 241 laboratories surveyed across seven Asian countries reported that only 53.5% of participants had designated mycology laboratories and that nearly all participants used traditional microscopy and culture methods for fungal identification; only 16.9% and 12.3% performed DNA sequencing and MALDI-TOF MS, respectively, for organism identification ². Participant surveys from the 2019 and 2020 College of American Pathologists (CAP) mycology proficiency testing program, which comprises up to ~1000 laboratories (majority from the United States), also showed that only ~50% of laboratories utilized MALDI-TOF MS for yeast identification, while <7% of laboratories utilized MALDI-TOF MS for mold identification.

Clearly, the waning availability of mycological expertise in front line laboratories and the need for accurate fungal identification (particularly for capture of cryptic species that generally have higher resistance profiles) leads to an urgent need to expand the application of MALDI-TOF MS into clinical mycology. This review will provide an overview of the current status of MALDI-TOF MS for fungal identification, including key findings in the literature, processing and database considerations, updates in technology, and future prospects.

MALDI-TOF MS for Yeast Identification

Researchers have reported relatively excellent performance of both Bruker MBT and VITEK MS platforms for yeast identification, although success rates can vary considerably depending on the organism challenge set, database version, extraction processing method, and acceptable cutoff threshold. Thus, data from the literature must be interpreted with caution. For yeast identification, high reproducibility has been reported across different testing environments, instruments, operators, reagent and target slide lots, and sample positioning patterns through multicenter studies ^{3, 4}. The advantages of MALDI-TOF MS for yeast identification compared with phenotypic platforms is overwhelmingly evident. Table 1 highlights some of the common misidentifications by traditional phenotypic methods that are resolved by the use of MALDI-TOF MS. Many discrepancies involve delineations within the species complexes such as C. glabrata complex (C. glabrata, C. nivariensis, C. bracarensis) and C. parapsilosis complex (C. parapsilosis, C. metapsilosis, C. orthopsilosis). A study of 182 yeast isolates demonstrated a 91% concordance between the MBT and the API 20C AUX biochemical panel; discordant results were attributed to rarely encountered and phenotypically difficult to identify organisms ⁵. Similarly, MBT and Microscan, an automated biochemical-based testing platform, had a 93.6% concordance when challenged with 498 yeast isolates ⁶. Overwhelmingly, the majority of yeast misidentifications that are ultimately corrected by the use of MALDI-TOF MS are attributable to cryptic or recently emerging species which tend to display higher MICs to antifungal agents. One such yeast is Saprochaete clavata (formerly Geotrichum clavatum) which is intrinsically resistant to echinocandins and fluconazole and was the cause of a cluster of fungemia in hospitalized hematology patients ⁷.

Table 1.

Common misidentifications associated with traditional identification platforms that are resolved by MALDI-TOF MS.

Accurate identification (method)	Misidentification (traditional method)	Reference
Candida albicans (MBT)	Candida famata (VITEK 2)	8, 9
Candida auris (MBT)	Candida catenulata (Microscan), C. famata (Microscan), Candida haemulonii (VITEK 2), Candida inconspicua (Microscan)	6
Candida dubliniensis (MBT)	C. albicans (Microscan, Phoenix Yeast ID)	6, 9
Candida fabianii (MBT)	Candida pelliculosa (ID 32C), Candida utilis (ID 32C)	10
C. guilliermondii (MBT)	C. famata (API 20C AUX, VITEK 2)	5, 9
Candida intermedia (MBT)	C. famata (Microscan)	6
Candida metapsilosis (MBT)	Candida parapsilosis (Microscan)	6
Candida nivariensis (MBT)	Candida glabrata (Microscan, VITEK 2, Phoenix Yeast ID)	6, 9
Candida orthopsilosis (MBT)	C. parapsilosis (VITEK 2, Microscan)	6, 8
Candida tropicalis (MBT)	Candida viswanathii (Phoenix Yeast ID)	9
Candida zeylanoides (MBT)	Saccharomyces cerevisiae (Phoenix Yeast ID, VITEK 2), Trichosporon asahii (Phoenix Yeast ID, VITEK 2)	9
Cryptococcus gattii (MBT)	Cryptococcus neoformans (Microscan)	6
Saprochaete clavata (SARAMIS)	Geotrichum capitatum (VITEK 2)	7

Rigorous database representation is a key factor for MALDI-TOF MS success. Research use only (RUO) databases are available for both Bruker MBT (RUO) and VITEK MS (SARAMIS) in addition to each manufacturer’s FDA-cleared database. In the United States, many laboratories opt to use the FDA-cleared database only due to the significant regulatory requirements to validate an RUO database in a clinical setting. The limitations of relying solely on an approved database, however, was highlighted in 2016 with the rapid emergence Candida auris as a global public health threat. The lack of C. auris representation in FDA-cleared databases at the time and the misidentification by various phenotypic platforms (Table 1) was concerning because almost all C. auris isolates were resistant to fluconazole and elevated MICs to voriconazole and amphotericin B had also been reported ¹¹. Several groups developed their own in-house supplemental databases at the time ^12–14 because FDA clearance of updated MBT and VITEK MS databases containing C. auris became available only mid—late 2018, two years after identification of what became a global outbreak. An external quality assessment of 47 Dutch laboratories in 2019 showed that only 74% of laboratories could correctly identify C. auris ¹⁵. This finding is concerning, as most of the misidentifications were attributed to use of older VITEK MS (pre Knowledgebase 3.2) versions; thus, highlighting the importance for maintaining up to date databases and/or reflexing appropriately to additional tests such as sequencing.

Direct head to head comparisons of the MALDI-TOF MS systems for yeast identification generally leans towards better performance of the Bruker MBT compared with VITEK MS ^{9, 16–19}; however, older versions of VITEK MS were studied and analysis focused on cryptic species including C. nivariensis, C. bracarensis, C. metapsilosis, and C. orthopsilosis. Others, however, have reported comparable performance between Bruker MBT and VITEK MS ^20–23 while one group reported better performance of VITEK MS compared with Bruker MBT ²⁴. The variability across the literature highlights the significant impact of organism challenge sets on performance assessment and emphasizes the need for updated literature that compares current database versions.

RUO databases such as SARAMIS have proven useful on multiple occasions where VITEK MS failed to identify S. clavata ⁷, C. glabrata complex ²⁵, and C. parapsilosis complex ⁹. However, numerous authors have published better success with MALDI-TOF MS after developing in-house databases to supplement that of the manufacturer including uncommon Cryptococcus species, foodborne yeasts, environmental yeasts, and rare Candida species ^{14, 19, 26–33}. Supplemental databases have also enabled clean differentiation and identification of Crypcococcus neoformans and Cryptococcus gattii for both Bruker MBT and SARAMIS ^{32, 34}.

The requirement for full protein extraction (also known as tube extraction) for yeast identification by MALDI-TOF MS is relatively controversial. Tube extraction, a process by which the organism is lysed and proteins are extracted using a solvent system containing formic acid and acetonitrile, leads to cleaner spectra and better identification scores, and is safer from a biohazard perspective. Full protein extraction was the method initially recommended by Bruker Scientific and BioMerieux upon release of MALDI-TOF MS on the clinical market. The method takes only minutes, but the need to improve workflow efficiency and increase throughput in the clinical laboratory led to the use of direct on-plate extraction whereby the organism is smeared onto the target plate followed by matrix (with or without the addition of formic acid). Recently, a large multicenter European study (10 centers, 1,511 yeast isolates) that compared different extraction methods showed that Bruker’s direct deposition method without formic acid extraction resulted in only 23.16% correct identifications compared with full tube extraction which yielded significantly better results (78.23%) ³⁵. This finding has been echoed by others in the field for C. auris and C. neoformans ^{22, 36}.

Another area of interest for Bruker MBT users has been the lowering of cutoff scores for species- and genus-level identification. Bruker MBT software utilizes a logarithmic score between 0–3 to weight the confidence of a spectral identification. According to manufacturer’s cut offs, scores ≥2.0 are confident to species level, scores 1.7–1.99 are confident for genus level, while spectra that score <1.7 cannot be accurately identified. One study showed an 87.2% success rate in species level identification of 117 mostly Candida isolates using the manufacturer’s cutoff score of ≥2.0 ³⁷, whilst others have demonstrated that a drop in threshold to as low as ≥1.7 significantly improved performance without compromising accuracy ^{14, 27, 29, 35, 38, 39}. The cutoff score chosen by any laboratory is going to be highly dependent on the quality of the extraction method, expanse of the database, and type of test organisms studied. Regardless of the method employed, it is important that any process that deviates from the manufacturer’s recommendations is thoroughly validated in the clinical setting before implementing for patient care.

Overall, MALDI-TOF MS has demonstrated successful identification for most yeast isolates. Early reports rightfully highlighted deficiencies in manufacturer’s databases for identification of cryptic and emerging yeast species. And although current databases are more expansive with organism representation, variations amongst strains of the same species may not necessary guarantee 100% successful identification ^{6, 24, 40}. In-house developed databases can improve performance significantly, and are a useful stop gap measure for rapid response to emerging pathogens if rigorously validated and interpreted with care.

MALDI-TOF MS for Mold Identification

Over the last decade, MALDI-TOF MS for accurate mold identification has become the inevitable quest in clinical laboratories given the growing number of recognized fungal pathogens, the emergence of cryptic species with atypical resistance patterns, and the dwindling numbers of laboratorians with skilled mycological expertise. And, even though numerous groups have demonstrated successful implementation of MALDI-TOF MS for mold identification into the routine laboratory workflow ^41–43, very few laboratories (at least in the United States), have chosen to utilize this technology for molds due to a series of non-standardized processes including culture methods, extraction methods, databases, and acquisition methods. The following section will summarize these areas of concern, and guide readers on important considerations when interpreting the literature and when implementing MALDI-TOF MS for routine mold identification in the clinical setting.

Accurate species level identification of molds has become an important diagnostic tool for patient management due to the emergence of cryptic azole resistant species such as Aspergillus calidoustus and Aspergillus lentulus, which are morphologically similar to Aspergillus ustus and Aspergillus fumigatus, respectively. When compared with traditional morphological identification, Gautier et al reported that implementation of MALDI-TOF MS led to a significant improvement in species level mold identification from 78.2% to 98.1%, and a marked reduction in misidentification rates from 9.8% to 1.2% ⁴⁴. Similar findings have been reported by others ^45–47 including accurate differentiation within the Aspergillus niger clade whereby Aspergillus tubingensis exhibited decreased azole susceptibility ⁴⁸ and accurate identification of Rasamsonia argillaceae that is often misidentified as Paecilomyces based on similar morphological characteristics⁴⁹.

The history of MALDI-TOF MS for mold identification is complex and consists of a culmination of various standardization issues. Early studies focused on the Bruker MBT as it was the first instrument launched into the clinical market with an RUO database. Given the lack of robust mold representation, Cassagne et al were the first in 2011 to design an in-house mold database consisting of 143 strains ⁴². In this landmark study, prospective analysis of 197 clinical isolates extracted from solid media resulted in an 87% correct species level identification; and, identification failure was attributed to a lack of organism representation in the database ⁴². In that study, each extract was spotted in quadruplicate and the authors applied an algorithm whereby at least three of the four deposits corresponded to the same species with at least one replicate scoring ≥1.9. This algorithm has since been widely adopted in other European centers ⁴⁴.

In the United States, a different mold MALDI-TOF MS database developed by Lau et al had published in 2013 consisting of 294 strains encompassing 76 genera and 152 species ⁴¹. In this study, blinded analysis of 421 clinical isolates from solid media demonstrated 88.9% species level identification whilst maintaining the manufacturer’s cutoff score of ≥2.0 ⁴¹. Testing was conducted in duplicate spots. Spectral analysis against Bruker’s RUO and FilFungal databases at the time had 0.7% and 16.2% sensitivity, respectively, for species level identification; and 48.4% of isolates failed to identify with the FilFungal database despite having spectral representation ⁴¹. Discussion later will highlight that this poor performance against the FilFungal database was likely attributed to spectra acquired from growth on solid media (NIH method) as opposed to liquid media (Bruker MBT method).

In the years since, a considerable number of publications describing the development of in-house supplemental mold databases have emerged. These have encompassed a variety of molds and have been built on both Bruker MBT ^50–68 and VITEK MS/SARAMIS systems ^69–73. In every single study, the in-house developed database performed better than the manufacturer’s database alone. The most impressive database to be released in recent years has been the freely-accessible, web-based application Mass Spectrometry Identification (MSI) platform, developed by Normand et al ⁵⁰. Upon its release in 2017, the MSI consisted of 11,851 spectra (938 fungal species and 246 fungal genera) and has been continuously updated since that time. In addition to demonstrating outstanding performance for mold identification from solid media (87.35% MSI vs 39.76% Bruker MBT), the authors were able to provide mass access globally to laboratories that had previously limited access to in-house built databases ⁵⁰. Others have also demonstrated excellent performance of the MSI database across clinical and veterinary fields ^{47, 74–76}. At this time, the MSI database is only compatible with spectra acquired from the Bruker MBT platform.

The culture process and extraction method of molds for MALDI-TOF MS have been areas of high variability in the literature. Most studies describing the development and validation of in-house databases, as well as the VITEK MS method, support the testing of molds grown on solid media as this follows routine clinical mycology processes. However, because of the natural heterogeneity of mold isolates and hence generated spectra, it is recommended that databases hold multiple mass spectral profiles (MSPs) for a single strain to increase the likelihood for identification ⁷⁷. The VITEK MS protocol consists of full protein extraction of conidia harvested from growth of molds on solid media. Numerous studies of the VITEK MS for mold identification, including multicenter evaluations, have demonstrated excellent performance ranging between 76.8% to 91% ^{43, 49, 71, 78}, although improvement in the representation of cryptic Aspergillus species is warranted ⁷¹.

In contrast, the Bruker MBT FilFungal database was developed using mold subcultured into liquid broth, which is an uncommon work practice in clinical mycology labs. Utilization of liquid broths was applied in an effort to achieve uniform consistency of mycelium, leading to acquisition of more standardized and cleaner spectra; thus, overcoming the natural heterogeneity of mold and spectral differences observed when testing isolates directly from solid media ⁷⁹. Performance of the FilFungal database following cultivation in liquid media resulted varies considerably (15.4% to 94.5%) for species level identification across studies ^{45, 80,81}.

Unfortunately, there are no studies that directly compare the performance of the manufacturer’s methods and databases alone for mold identification (ie. Bruker MBT liquid cultivation method against FilFungal database, and VITEK MS solid media cultivation against Knowledgebase 3.2). There are, however, two studies that compare both systems and databases from growth on solid media. These modified head to head comparisons show an overwhelmingly better performance of the VITEK MS compared with Bruker MBT for mold identification from solid media ^{76, 82}.

Given the variable performance of MALDI-TOF MS for mold identification in the literature, multicenter studies have become the key towards addressing development of a standardized process. Four groups have made headway in this area. The first study by Normand et al showed wide inter-laboratory performance for mold identification across five European centers, two different databases (MSI and Bruker MBT), and different cut off scores ⁵⁰. In Canada, Stein et al compared three different databases and also found high inter-laboratory variability between three clinical laboratories ⁴⁷. Interestingly, these authors reported that the highest and lowest scores were consistently obtained by the same laboratory suggesting instrument variability or inherent reproducibility problems ⁴⁷. One could hypothesize that the high inter-laboratory variability reported by these two studies may be attributable to variations in test isolates between laboratories for prospective clinical analysis. However, a large multicenter study published by Lau et al in 2019 demonstrated a wide range in performance (33–77%) across eight different testing centers analyzing 80 identical isolates using the routine Bruker MBT acquisition program against the NIH database ⁸³. Adjustment of key parameters in the acquisition program, along with optimized instrument maintenance, significantly improved performance for most centers ⁸³. Following this study, Bruker Scientific announced release of an updated software module (version 3.0) that contained adjusted acquisition parameters specifically to improve mold identification rates. Surprisingly, issues with inter-laboratory agreement has only been associated with Bruker MBT platforms; one multicenter study conducted on VITEK MS showed relatively good reproducibility between testing sites ⁴³. More studies are needed in this area to evaluate the extent of improving inter-laboratory agreement based on modified acquisition parameters. But, the findings thus far significantly advances the likelihood for successful implementation of MALDI-TOF MS in laboratories for routine mold identification.

Other variables such as extraction methods have also been studied; however, it is abundantly clear that full protein extraction (with or without mechanical disruption) rather than direct plate deposition produces far superior spectra and better quality identification ^{68, 82, 84}. Research into a new media formulation called ID FUNGI plate (IDFP; Conidia, Quincieux, France) is making recent headlines as it utilizes a transparent low adherence membrane across the agar surface that allows for clean mold harvest without interference from the media. Early studies suggest that IDFP performs better than conventional media for mold identification on Bruker MBT and against a variety of databases including FilFungal, NIH Database, and MSI Database ^{75, 85–87}. Some also showed successful identification of mold following direct deposition ^{86, 87}. And, similar to yeast studies, lowering the cut off score for Bruker MBT to as low as ≥1.7 has been used by several group to improve mold identification without compromising in accuracy ^{45, 52, 58, 60, 64, 66, 68, 76, 80, 88}.

Overall, significant advances have been made over the last decade to highlight the clinical utility of MALDI-TOF MS for mold identification. Key findings regarding acquisition programs and the availability of free online databases such as MSI will hopefully make mold identification by MALDI-TOF MS a real and feasible platform across many clinical laboratories.

MALDI-TOF MS for Fungal Identification Directly from Positive Blood Culture Bottles

Early and accurate detection of fungal bloodstream infections still remains challenging. Rapid identification directly from a positive blood bottle has demonstrated significant improvement in patient outcomes through the timely initiation of appropriate and targeted antimicrobial therapy. Bal et al studied 74 patients with candidemia and showed that appropriateness of therapy was significantly higher for the rapid identification (MALDI-TOF MS) group (90.9%) compared with the conventional identification group (62.2%), which also resulted in more than £10,000 cost savings within the first three days of treatment ⁸⁹. Similarly, de Almeida reported that five of seven patients with fungemia had targeted antifungal therapy guided by MALDI-TOF MS species identification directly from positive blood cultures ⁹⁰.

Sample preparation and cleanup is critical for ensuring quality spectra and good identification results. To that end, various sample processing methods for red blood cell lysis and removal of inhibitors prior to MALDI-TOF MS have been evaluated. The Sepsityper kit from Bruker Scientific utilizes a series of washes for cell lysis and cleanup, resulting in a pellet that can be then be chemically extracted following the routine formic acid and acetonitrile process. This method ^{89, 91}, and those applied by others including saponin extraction ⁹² and SDS lytic extraction ⁹⁰, have resulted in only moderate performance for the detection of yeasts directly from positive blood cultures. Short term incubation of the subculture for 4–6 h did not improve performance even when a lowered threshold of 1.7 was applied ^{91, 93}, although Florio et al reported 100% concordance in identifying yeasts from 17 positive blood cultures, including polymicrobial cultures, whilst maintaining the manufacturer’s threshold of 2.0 ⁹⁴. In addition to lowering the threshold, multiple spots (eg. quadruplicate), full chemical protein extraction, and research use only databases may need to be applied to improve the performance rate of MALDI-TOF MS for fungal identification directly from positive blood cultures. Regardless of which method is chosen, it is critical that laboratories validate the method to ensure consistent and reliable results. In January 2021, Bruker Scientific announced FDA approval of the Sepsityper kit which includes detection of bacteria and yeasts. This may assist in inter-laboratory standardization and a willingness to implement such tests in the clinical setting to benefit patient care.

MALDI-TOF MS for Antifungal Resistance Detection in Yeasts and Molds

Given the details above, it is clear that MALDI-TOF MS has had an overwhelmingly positive impact on patient care and epidemiology through the provision of rapid, accurate, and detailed fungal identification. In recent years, MALDI-TOF MS has been used successfully for the detection of antimicrobial resistance markers in bacteria, even sometimes through the use of the same spectral profile already acquired for organism identification ^{95, 96}. The clinical application of MALDI-TOF MS for predicting antifungal susceptibility profiles, however, is still in its infancy, although early research has shown some promise.

The MBT ASTRA method, a supplemental software available by Bruker Scientific and designed originally for resistance detection in bacteria, has been applied by some groups for antifungal resistance detection. The MBT ASTRA method involves the comparison of spectral profiles obtained from an isolate exposed and not exposed to an antimicrobial. Through computational analysis, shifts detected in the spectral profile of the isolate exposed to the antimicrobial may indicate resistance or susceptibility to that agent, depending on the pattern observed. The pre-incubation exposure time is generally short (~3–6 hours), resulting in a method that may be faster, cheaper, and simpler than traditional CLSI testing and/or sequencing for mutation analysis.

With the emergence of echinocandin resistance in Candida species, and reliance of echinocandins for first line therapy of invasive candidiasis, Vatanshenassan et al utilized MBT ASTRA to evaluate the performance of MALDI-TOF MS to detect echinocandin resistance in Candida species. In their proof of principle study, Vatanshenassan et al ⁹⁷ demonstrated 100% categorical agreement between MBT ASTRA and CLSI broth microdilution for caspofungin susceptibility profiling of C. albicans (n=58). A lower categorical agreement was obtained for C. glabrata (n=57) resulting in a sensitivity and specificity of 94% and 80%, respectively, between MBT ASTRA and the CLSI broth microdilution. A follow up study by this same group evaluated MBT ASTRA for the rapid detection of anidulafungin-resistant C. glabrata isolates directly from positive blood cultures (n=100) ⁹⁸. Here, the MBT ASTRA had a sensitivity and specificity of 80% and 95%, respectively, compared with CLSI broth microdilution, and a positive and negative agreement of 100% and 80%, respectively when MBT ASTRA was compared with sequencing analysis of hot spots in FKS1 and FKS2.

Similar studies have also been conducted for C. albicans and fluconazole resistance ⁹⁹, C. tropicalis and fluconazole resistance ¹⁰⁰, as well as C. parapsilosis complex and echinocandin resistance ¹⁰¹. All of these groups demonstrated moderate to high success rates for MALDI-TOF MS antifungal susceptibility profiling when compared with a reference method. In contrast, Saracli et al showed that results were too variable when MALDI-TOF MS was applied to triazole resistance detection in various Candida species ¹⁰², and Vella et al demonstrated assay variability depending on antifungal exposure time during pre-incubation ¹⁰³. More studies are warranted, but the data thus far suggests that MALDI-TOF MS may be a promising tool for some drug/yeast combinations, but more work is required before such testing becomes mainstream. Conversely, MALDI-TOF MS is unlikely to be a favorable alternative for the susceptibility profiling of molds. A single study showed that voriconazole resistance detection in A. fumigatus was possible but offered no advantages to traditional CLSI testing or CYP51A sequence analysis ¹⁰⁴.

CONCLUSIONS

In conclusion, the clinical mycology field has advanced significantly in technological capabilities over the last decade. As summarized above, accurate species level identification of yeasts and molds, particularly within species complexes and cryptic organisms, is highly achievable and practical with MALDI-TOF MS. The wide availability of these platforms in diagnostic settings will likely have a significant impact on patient outcomes by enabling more rapid identification and initiation of appropriate therapy. The simplicity of testing and low cost consumables makes MALDI-TOF MS an ideal platform in low resource settings ¹⁰⁵. And, even though this review focused only on Bruker MBT and VITEK MS platforms, other systems including ASTA MicroIDSys system (ASTA Inc, South Korea) and Xiamen Microtyper (China) have shown excellent performance in preliminary studies ^{106, 107}. The current state of the diagnostic clinical mycology laboratory is exciting, and data suggests that there is a very promising era ahead towards standardization and wide implementation of MALDI-TOF MS for fungal identification.

Key points:

• MALDI-TOF MS has demonstrated excellent performance for rapid yeast identification. Updated databases provide accurate identification of cryptic species that have increased resistance to antifungal agents.

• MALDI-TOF MS for mold identification in highly reliant on the use on in-house developed databases to supplement the manufacturer’s databases.

• Successful implementation of MALDI-TOF MS for mold identification has been demonstrated; however, culture methods, extraction methods, databases, and acquisition methods lack standardization.

• Early research shows that MALDI-TOF MS may be used to detect antifungal resistance in yeasts; poor performance was demonstrated for mold.