Short Running Title: Fungal Molecular ID in a Surveillance Study

Significance of Molecular Identification and Antifungal Susceptibility of Clinically Significant Yeasts and Moulds in a Global Antifungal Surveillance Program

Short Running Title: Fungal Molecular ID in a Surveillance Study

Michael A. Pfaller1,

Leah N. Woosley1,

Shawn A. Messer1,

Ronald N. Jones1,

and

Mariana Castanheira1

1JMI Laboratories, North Liberty, Iowa52317USA

Corresponding author: Michael A. Pfaller, M.D.

JMI Laboratories

345 Beaver Kreek Centre, Suite A

North Liberty, IA52317

Phone: 319-665-3370

Fax: 319-665-3371

Email:

ABSTRACT

The increasing diversity of opportunistic fungi causing serious invasive fungal infections (IFI) has been documented. Accurate identification (ID) is important in guiding therapy, determining prognosis for IFIs and in epidemiological surveys. We assessed the utility of PCR-based methods for the ID of yeasts and moulds that were either uncommon, failed conventional ID, or represented unusual biochemical or phenotypic profiles of common species.Among 1,790 viable fungal clinical isolates received during the SENTRY Program in 2010, 322 strains from 40 study sites had ID confirmed by molecular methods. Isolates were previously identified in participant institutions. Yeasts that were not confirmed by morphology on CHROMagar, growth at 45C (C. albicans/dubliniensis) or assimilation of trehalose (C. glabrata) as well as non-Candida yeasts and all moulds were amplified and sequenced using primers amplifying one or more of the following genes: ITS, 28S, β-tubulin (Aspergillus spp.), TEF (Fusarium spp.), IGS (Trichosporon spp.).The isolates selected for molecular ID included 149 isolates of Candida species, 77 of Aspergillus species, 73 non-Candida yeasts and 23 other moulds (a total of 41 different species). Overall the ID determined by the submitting site was confirmed for 189 isolates (58.7%): Aspergillus spp. (64.1% correct); Candida spp. (60.1% correct); non-Candidayeasts (58.9% correct); non-Aspergillus moulds (30.4% correct). Species with high levels of concordance between conventional and molecular ID included A. fumigatus (95.0 %), C. lusitaniae (100%), C. dubliniensis ( 92.3%), C. kefyr (100%) and C. neoformans (90.2%). Only 50.0% of isolates of C. albicans and 59.1% of C. glabrata selected due to unusual phenotypic or biochemical features were found to be correctly identified by the submitting site.Molecular methods for the identification of fungal pathogens are an important adjunct to the conventionalidentification of many less common clinically relevant yeasts and moulds including species of Candida with unusual or erroneous phenotypic or biochemical profiles. Molecular confirmation of fungal identification is essential in epidemiological surveys such as SENTRY.

Key words: fungi, SENTRY, molecular identification methods.

INTRODUCTION

Numerous studies have documented the ever increasing diversity of opportunistic yeasts and moulds capable of causing serious invasive fungal infections (IFI) in immunocompromised hosts [1-6]. Accurate identification (ID) of these organisms is important in guiding therapy and determining prognosis in these IFI’s as well as in epidemiological surveys [3, 6-12]. Conventional methods of ID take into account biochemical and morphological characteristics and are labor intensive requiring considerable mycological expertise [2, 3, 6-8, 10, 13-16]. In addition, these methods are often inadequate for the ID of less common species and fail to differentiate closely related species [9, 10, 13, 15-21]. Molecular methods for ID of yeasts and moulds represent rapid, sensitive and cost-effective alternatives to conventional methods [8-10, 13, 14, 21-23].

In the present study, we assessed the utility of PCR-based molecular methods for the ID of yeasts and moulds, submitted during the course of the SENTRY Antifungal Surveillance Program (2010), that were either uncommon (rare species of Candida, non-candidal yeasts, Aspergillus spp. and other moulds), failed conventional ID, or represented unusual biochemical or phenotypic antifungal susceptibility profiles of common species. The in vitro susceptibility profile of those organisms identified by molecular methods to the systemically active triazole and echinocandin antifungal agents was determined using Clinical and Laboratory Standards Institute (CLSI) broth microdilution (BMD) methods.

MATERIALS AND METHODS

Organisms: Among 1,790 viable fungal isolates received during the SENTRY Program in 2010, 322 strains from 40 different study sites had ID confirmed by molecular methods. This worldwide surveillance program collects up to 40 consecutive clinically significant fungal isolates from bloodcultures (all fungi) as well as from respiratory tract specimens of patients diagnosed with pneumonia caused by Aspergillus spp. or other moulds. Confirmatory identification and susceptibility tests are processed in a central laboratory (JMI Laboratories, North Liberty, Iowa, USA). Isolates were previously identified at participant institutions using methods routinely employed at the submitting laboratory. Purity of isolates is confirmed by subculturing the isolates in appropriate media followed by visual examination. All Candida species isolates that were not confirmed upon receipt at JMI Laboratories by morphology on CHROMagar (Becton Dickinson and Company, Sparks, Maryland, USA; Candida albicans, C. tropicalis and C. krusei), growth at 45ºC (C. albicans, C. dubliniensis), or assimilation of trehalose (C. glabrata) as well as non-Candida yeasts and all moulds were submitted to molecular ID. All experiments were carried out in the Mycology and Molecular Research Sections of JMI Laboratories.

Species identification:Isolates were subcultured from storage and identified to species or species complex (SC) level by molecular methods. DNA extractions were performed using QIAquick Extraction kit (Qiagen, Hilden, Germany; yeasts only) or UltraClean Microbial DNA Isolation kit (MO BIO Laboratories, Carlsbad, California, USA) and amplification and sequencing of the following genes: internal transcribed spacer (ITS) region, 28S ribosomal subunit, β-tubulin (Aspergillus only), translation elongation factor (TEF; Fusarium only), and IGS (Trichosporon spp. only) were carried out as described previously [8, 23-25]. Nucleotide sequences were analyzed using Lasergene® software (DNAStar, Madison, Wisconsin, USA) and compared to available sequences through the internet using BLAST ( TEF sequences were analyzed using theFusarium-II database ( and the Fusarium multilocus sequence typing (MLST) database ( were considered acceptable if homology was >99.5% with other entries in the databases used for comparison. Available sequences that were considerably different from the majority of entries for one species were considered outliers and discarded in the analysis. Additionally, if no match was found in the database, the identification was based on species complex, genus, family or order, according to the most current classifications systems.

Antifungal susceptibility testing: All Candida spp., Cryptococcus neoformans, and other non-Candida yeasts were tested for in vitro susceptibility to the echinocandins and triazoles using CLSI BMD methods [26]. MIC results for anidulafungin, caspofungin, micafungin, fluconazole, posaconazole and voriconazole were read after 24-h incubation. MIC results for these agents were read after 48-h when tested against C. neoformans and other non-Candida yeasts. In all instances, the MIC values were determined visually as the lowest concentration of drug that caused significant diminution (≥50% inhibition) of growth relative to that of the growth control [5, 12, 26, 27].

In vitro susceptibility testing of Aspergillus spp. and other moulds against the echinocandins and triazoles (itraconazole, posaconazole, and voriconazole) was performed by BMD as described in CLSI document M38-A2 [28]. The triazole MICs and echinocandin minimum effective concentration (MEC) for the moulds were determined as described in the CLSI reference method [28].

Quality control (QC) was performed as recommended in CLSI documents M27-A3 and M27-

S3 [26, 29]and M38-A2 [28] using C. krusei ATCC 6258,.C. parapsilosis ATCC 22019, A. fumigatus MYA-3626 and A. flavusATCC 204304.

RESULTS

Fungal isolates

Among the 1,719 isolates of fungi that were identified at the participating institutions and sent to JMI Laboratories for confirmation of ID and antifungal susceptibility testing as part of the SENTRY Antifungal Susceptibility testing, 1,557 (87.0%) were Candida spp., 112 (6.3%) were non-Candida yeasts (43 of which were C. neoformans), 90 (5.0%) were Aspergillus spp., and 31 (1.7%) were non-Aspergillus moulds (Table 1). A total of 322 isolates (18.0%) were selected for molecular ID including 149 isolates of Candida spp. (20 species), 73 isolates of non-Candida yeasts (14 species), 77 isolates of Aspergillus spp. (six species) and 23 isolates of non-Aspergillus moulds (13 species). Overall, the ID determined by the submitting site was confirmed for 189 (58.7%) isolates (Table 1): Candida spp. (60.1% correct), non-Candida yeasts (58.9% correct), Aspergillus spp. (64.1% correct), and non-Aspergillus moulds (30.4% correct).

Molecular Identification of Yeasts

Species of yeasts with high (>90%) levels of concordance between the conventional (methods used at submitting site) and molecular ID included: C. krusei, C. lusitaniae, C. kefyr, C. haemulonii, C. inconspicua, C. lipolytica, C. pelliculosa, Aureobasidium pullulans, Rhodotorula mucilaginosa, Trichosporon asahi,(100.0% correct ID); C. dubliniensis (92.3% correct); and C. neoformans (90.2% correct). In contrast, only 45.5% to 60.0% of isolates of C. parapsilosis (45.5% correct), C. albicans (50.0% correct), C. glabrata (59.1% correct), C. tropicalis (60.0% correct), and C. guilliermondii (60.0% correct), selected due to unusual phenotypic, morphologic, or biochemical factors were found to be correctly identified by the submitting site (Table 1). Very few of the rare species of Candida and the non-Candida yeasts submitted were correctly identified by the participating sites.

Among the 13 species of Candida (60 isolates including 13 Candida spp. NOS) for which the participant ID was not confirmed, there were 19 different species of yeasts (17 Candida species, 1 Saccharomyces cerevisiae, and 1 C. neoformans) identified by molecular methods (Table 2). As might be expected, some of the misidentifications involved “cryptic” species (C. dubliniensis, C. fermentati, C. bracarensis, C. orthopsilosis, C. metapsilosis), although it should be noted that 12 of 13 isolates identified by participants as C. dubliniensis were confirmed by molecular ID, attesting to the improved biochemical and phenotypic methods available for this species. Whereas five of the six isolates misidentified as C. parapsilosis were found to be either C. orthopsilosis (4 isolates) or C. metapsilosis (1 isolate), it should be noted that C. parapsilosis was frequently misidentified as another species of Candida (18 of 60 misidentifications [30.0%]). There were 42 isolates of Candida submitted as Candida spp. NOS (Table 1);13 of these were selected for molecular ID of which nine were found to be relatively common species of Candida (C. albicans and C. glabrata [1 isolate each] and C. parapsilosis [7 isolates]) and one was C. neoformans (Table 2). Although only a minority of the Candida isolates submitted lacked a species ID (42 of 1,557 [2.7%]), the misidentification of C. neoformans as Candida spp. NOS is a serious error and underscores the importance of accurate identification of clinical isolates of fungi. In this latter instance, the performance of a simple urease test (positive for Cryptococcus) would indicate that a germ tube-negative yeast was not likely to be a species of Candida and that an identification to both genus and species should be pursued; however it is important to consider that this mistake could be due to mislabeling of the isolate instead of misidentification.

There were eight species of non-Candida yeasts (30 isolates, including 16 yeasts, NOS) for which the participant ID was not confirmed (Table 2). In most of these (22 of 30;73.3%), the correct ID turned out to be a species of Candida. Again, C. parapsilosis was the species of Candida most commonly misidentified as a non-Candida yeast or yeast, NOS. Notably, three different species of Candida were misidentified by participants as C. neoformans.

Molecular Identification of Moulds

Species of moulds with high (>90%) levels of concordance between the conventional (methods used at submitting site) and molecular ID included: Aspergillus nidulans, A. niger, Lichtheimia (Absidia) corymbifera, Curvularia spp. NOS (not otherwise specified), Fusarium solani, Scedosporium apiospermum, and Wangiella dermatitidis(100.0% correct);and A. fumigatus (95.0% correct). Among the 90 isolates of Aspergillus spp. obtained in the 2010 SENTRY Program, 21 (23.3%) were not identified to the species level (Table 1).Of these isolates, 19 were available for molecular ID and represented 6 different species of Aspergillus(18 isolates) and 1 isolate of Penicillium minioluteum (Table 2). Aside from this group, notable ID errors include 2 isolates of A. alabamensis misidentified as A. terreus and one isolate each of A. flavus SC, A. fumigatus, and A. sydowii, misidentified as A. versicolor.

Although the majority of the 31 non-Aspergillus moulds were not correctly identified to the species or SC level, for the most part the correct ID to the genus level was provided by participants(Table 2). Among 23 isolates of non-Aspergillus moulds available for molecular ID, seven were correctly identified to the species or SC level and an additional 12 were correctly identified to the genus level (83% correct at the genus level; Tables 1 and 2).

Susceptibility Testing of Fungal Isolates Identified by Molecular Methods

The MIC distributions for the echinocandins and triazoles for species with more than two isolates for which molecular ID was confirmed are shown in Table 3. Those species with one or two isolates identified by DNA sequencing based methods can be found in the electronic supplementary material. Given that many of these species are uncommon and data concerning their in vitro susceptibility to most antifungals is lacking, we have elected to display results as the number of isolates at each MIC value so that these results may ultimately be combined with similarly derived data to form a more robust understanding of the MIC profiles of these unusual species.

In general, the MIC values obtained for both classes of agents and the various species of Candida conform to the wild-type (WT) MIC distributions described previously [12, 27, 30-33]. Notable observations include decreased echinocandin susceptibility (MIC > WT) among C. glabrata (4 isolates) to caspofungin (MIC, >0.12 µg/ml) and micafungin (MIC, >0.03 µg/ml), among C. dubliniensis (2 isolates) to all three echinocandins (MIC, >0.12 µg/ml), among C. guilliermondii(1 isolate) to anidulafungin (MIC, >4 µg/ml) and micafungin (MIC, >2 µg/ml), and among C. kefyr (1 isolate) to caspofungin (MIC, >0.03 µg/ml) and micafungin (MIC, >0.12 µg/ml). The MIC values for the echinocandins against the very rare species of Candida were generally low (<0.5 µg/ml) with the exception of C. fermentati, C. lipolytica, and C. nivariensis.

Cross-resistance among the azoles was detected in 2 isolates of C. glabrata (fluconazole MIC, >32 µg/ml; posaconazole MIC, >2 µg/ml; voriconazole MIC, >1 µg/ml), 1 isolate of C. tropicalis (fluconazole MIC, >2 µg/ml; voriconazole MIC, >0.06 µg/ml), 3 isolates of C. lusitaniae (fluconazole MIC, >2 µg/ml; posaconazole MIC, >0.25 µg/ml; voriconazole MIC, >0.03 µg/ml), 1 isolate of C. dubliniensis (fluconazole MIC, >0.5 µg/ml; posaconazole MIC >0.12 µg/ml; voriconazole MIC, 0.03 µg/ml), and 1 isolate of C. guilliermondii (fluconazole MIC, >8 µg/ml; voriconazole MIC, >0.25 µg/ml).

As expected the echinocandins were inactive against most of the non-Candida yeasts (Table 3). Echinocandin MIC results of≤0.5 µg/ml were only seen with S. cerevisiae and Aureobasidium pullulans. In contrast the triazoles, especially voriconazole, showed good activity against most isolates of C. neoformans, Saccharomyces, and Trichosporon tested.

The echinocandins were highly active against all species of Aspergillus, only one isolate (A. niger) exhibited an MEC of >0.06 µg/ml (Table 3). Comparable activity against Aspergillus spp. was seen with all three triazoles tested with MICs clustered between 0.25 and 1µg/ml for all three agents. In the 2009 SENTRY Program, we reported that all A. fumigatus isolates (n = 40) appeared to be WT strains (no acquired resistance mechanisms) with MIC results for itraconazole and voriconazole of 1 µg/ml or less and for posaconazole of ≤0.5 µg/ml [34]. Likewise, only one isolate (2%) ofA. fumigatus tested in 2010 showed decreased (non-WT) susceptibility to itraconazole (MIC = 2 µg/ml), posaconazole (MIC= 4µg//ml) and voriconazole (MIC = 8 µg/ml).

Among the non-Aspergillus moulds, decreased susceptibility to both echinocandins and triazoles was observed for most of the genera tested (Table 3). Low MEC values for anidulafungin and micafungin were observed for all isolates of Penicillium and Paecilomyces species. The single isolate of Paecilomyces variotii tested exhibited low MECs foranidulafungin (MEC, 0.015 µg/ml) and micafungin (MEC, 0.008 µg/ml) and an elevated MEC for caspofungin (MEC, 4 µg/ml). The importance of this apparent discrepancy within the echinocandin class remains to be determined.

DISCUSSION

There are several important findings that can be derived from this study. First, we have demonstrated the importance of molecular methods in confirming the ID not only of uncommon species of Candida but also non-Candida yeasts, Aspergillus species, and non-Aspergillus moulds. This also applies to more common species of Candida that despite the results provided by standard commercial methods, present with unusual morphologic, biochemical, and/or phenotypic features suggesting that the commercial method ID may be incorrect. Among the more common species of Candida, C. parapsilosis is most likely to be misidentified as another species of Candida or as a non-specified yeast.

The use of molecular methods for fungal ID is essential in the context of a global surveillance program and may play an increasing role in the clinical laboratory as expertise in clinical mycology wanes [8, 10]. One of the main limitations for the use of molecular ID is that some of the existing databases may be lacking with regard to both the quality and the accuracy of their entries [8, 22, 35]. Having said this, the advantages of molecular methods involving gene sequencing are impressive [22]: the results are objective, are not influenced by growth conditions, are frequently more rapid and cost-effective than classical biochemical and phenotypic approaches, and are capable of discriminating between fungi that fail to produce distinctive morphological features.

Accurate ID of opportunistic fungi clearly has clinical consequences (e.g. misidentification of C. neoformans as Candida spp., NOS; Table 2) and is also important in assessing the epidemiology, pathogenicity, and antifungal susceptibility, especially of less common or “cryptic” species. We have included the antifungal susceptibility profiles of those species from the SENTRY Program that have undergone sequence-based ID in an effort to provide MIC data, not only for the relatively common species of yeasts and moulds, but also for those that may be less frequent but still pose problems in selecting optimal therapy. In doing so, we have identified decreased susceptibility to both echinocandins and triazoles in several species of Candida, non-Candida yeasts, and non-Aspergillus moulds. The emergence of azole resistance among Aspergillus spp. has been reported by others [36]; however, the present exercise has identified only 2% of A. fumigatusisolates that warrant further investigation regarding acquired resistance mechanisms.

There are also several potential limitations to this survey that must be acknowledged. First of all the SENTRY Surveillance Program is a sentinel survey designed to investigate the antifungal susceptibility profiles of important fungal pathogens on a global scale and is not designed, nor funded, to serve as a formal assessment of the accuracy of the methodologies used routinely in the participating laboratories for the identification of clinical isolates. We routinely screen all submitted isolates of Candida species using morphology on CHROMagar (C. albicans, C. tropicalis and C. krusei), growth at 45C (C. albicans and C. dubliniensis), and assimilation of trehalose (C. glabrata) as a simple means of confirming the submitted ID for these more common species. Isolates that fail these screening tests, as well as those with unusual antifungal susceptibility profiles (e.g. echinocandin resistance in isolates of C. albicans, C. glabrata, C. tropicalis and C. krusei), less common species of Candida (species other than C. albicans, C. glabrata, C. tropicalis, C. parapsilosis, and C. krusei), non-Candida yeasts and all moulds are processed further using the molecular methods described herein to arrive at a final ID. As such we examined 9.5% of all Candida isolates, 65% of all non-Candida yeasts, 86% of all Aspergillus isolates and 74% of all non-Aspergillus moulds submitted in 2010 using molecular methods to confirm the ID. Although microscopic morphology was taken into account, this was not performed in every instance. The low level of ID confirmation (58.7%) for these isolates must be considered to represent a “worst case” scenario concerning the accuracy of the participating sites as only the most difficult or unusual isolates were considered for molecular ID.