Nationwide Surveillance of Antifungal Resistance of Candida Bloodstream Isolates in South Korean Hospitals: Two Year Report from Kor-GLASS

We incorporated nationwide Candida antifungal surveillance into the Korea Global Antimicrobial Resistance Surveillance System (Kor-GLASS) for bacterial pathogens. We prospectively collected and analyzed complete non-duplicate blood isolates and information from nine sentinel hospitals during 2020–2021, based on GLASS early implementation protocol for the inclusion of Candida species. Candida species ranked fourth among 10,758 target blood pathogens and second among 4050 hospital-origin blood pathogens. Among 766 Candida blood isolates, 87.6% were of hospital origin, and 41.3% occurred in intensive care unit patients. Adults > 60 years of age accounted for 75.7% of cases. Based on species-specific clinical breakpoints, non-susceptibility to fluconazole, voriconazole, caspofungin, micafungin, and anidulafungin was found in 21.1% (154/729), 4.0% (24/596), 0.1% (1/741), 0.0% (0/741), and 0.1% (1/741) of the isolates, respectively. Fluconazole resistance was determined in 0% (0/348), 2.2% (3/135, 1 Erg11 mutant), 5.3% (7/133, 6 Pdr1 mutants), and 5.6% (6/108, 4 Erg11 and 1 Cdr1 mutants) of C. albicans, C. tropicalis, C. glabrata, and C. parapsilosis isolates, respectively. An echinocandin-resistant C. glabrata isolate harbored an F659Y mutation in Fks2p. The inclusion of Candida species in the Kor-GLASS system generated well-curated surveillance data and may encourage global Candida surveillance efforts using a harmonized GLASS system.


Introduction
Antimicrobial resistance (AMR) is a leading cause of human death around the world [1]. In 2015, the World Health Organization (WHO) launched the Global Antimicrobial Resistance Surveillance and Use System (GLASS), the first global collaborative effort to

Collection of Candida Blood Isolates
Along with Kor-GLASS bacterial pathogens, Candida blood isolates were prospectively collected from all patients with Candida BSIs at nine sentinel hospitals from January 2020 to December 2021. All hospitals were general hospitals that even treated pediatric patients, with a total of 7551 beds (655-1092 beds per hospital) and located in nine districts throughout South Korea ( Figure S1). At each hospital, the first Candida-positive blood isolate per patient per species was collected. All isolates of Candida species in the MycoBank Database that had a Candida anamorph name or were previously called Candida were included [5]. The collected blood isolates were transferred to the analysis center (Chonnam National University Hospital) twice a month through a cold chain delivery system. For long-term storage, subsamples of each isolate were independently kept at three separate sites: one in the sentinel hospital, two in the analysis center, and two in the national reference laboratory [3].

Collection of Clinical Data from Sentinel Hospitals
Demographic data (age and sex), infection origin [hospital origin (HO) or community origin (CO)], and admission type (intensive care unit [ICU] or ward) were recorded at each sentinel hospital for all patients for whom blood cultures were performed during the study period [3,4,13,14]. An HO infection as indicated by a blood specimen taken from an inpatient hospitalized for 2 days, including the hospitalization days in another healthcare facility before transfer. A CO infection was indicated by a blood specimen taken from an out-or inpatient hospitalized for <2 days. All other considerations for Kor-GLASS bacterial pathogens were applied when setting up the national AMR surveillance system for Candida species [3,4]. The incidence of candidemia was calculated based on the number of cases of candidemia per 10,000 patient days (PD) at each hospital [15,16]. This study was approved by the Institutional Review Board (IRB) of Chonnam National University Hospital (CNUH-2020-080) and all sentinel hospitals. The IRBs of nine sentinel hospitals waived the requirement for informed consent because of the observational nature of the study and the very low risk of breaches of participant privacy.

Species Identification and Antifungal Susceptibility Testing
All collected isolates were re-identified using matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry (Bruker Biotyper, Bruker Daltonics GmbH, Bremen, Germany) at the analysis center. Isolates with discrepant identification between the collection and the analysis centers, and isolates of uncommon species, were further analyzed by sequencing the D1/D2 domains of the 26S rRNA gene [7,11]. In vitro antifungal susceptibility testing of fluconazole, voriconazole, posaconazole, itraconazole, amphotericin B, caspofungin, micafungin, anidulafungin, and 5-flucytosine (5-FC) was performed using the Sensititre Yeast One (SYO) system (Thermo Scientific, Cleveland, OH, USA). Two reference strains, Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were included in each antifungal susceptibility test as quality control isolates. The interpretative guideline in the Clinical and Laboratory Standards Institute (CLSI) document M60 ED1 was used to classify isolates according to species-specific clinical breakpoints (CBPs) [17,18]. For external quality control of the species identification and antifungal susceptibility test results, 5% of the results obtained by the analysis center were reevaluated by comparison with those conducted in an independent quality control center. The analysis center was certified every 3 months by the external quality assurance program of the quality control center [13,14].

Molecular Mechanisms of Antifungal Resistance
The analysis center performed an advanced characterization of the isolates to assess the molecular mechanisms of antifungal resistance. The ERG11 gene was sequenced in all fluconazole-resistant or susceptible dose-dependent (fluconazole minimum inhibitory concentration [MIC] ≥ 4 mg/L) isolates of C. albicans, C. parapsilosis and C. tropicalis [7,11,19]. The sequence of the CDR1 gene was analyzed in fluconazole-resistant (fluconazole MIC ≥ 8 mg/L) C. parapsilosis isolates without mutations in ERG11 [19]. PDR1 was sequenced in all fluconazole-resistant (fluconazole MIC ≥ 64 mg/L) isolates of C. glabrata [11]. Echinocandin resistance was confirmed for all common Candida isolates with echinocandin MIC values higher than the CLSI CBPs (either intermediate [I] or resistant) through DNA sequence analysis of FKS1 (all Candida species) and FKS2 (C. glabrata) [20].

Statistical Analysis
Fisher's exact test or the chi-squared test was used to compare categorical variables, and a t-test was used to compare quantitative variables. All statistical analyses were performed using the Diagnostic Test Evaluation Calculator (MedCalc, Ostend, Belgium) and GraphPad Prism software (version 9.3.1; GraphPad Software Inc., San Diego, CA, USA). A p-value < 0.05 was considered to indicate statistical significance.

Species Distribution of Candida BSI Isolates
The species distributions of 766 non-duplicate bloodstream isolates of 15 Candida species are listed in Table 1. C. albicans was the most common of the Candida species identified,

Species Distribution of Candida BSI Isolates
The species distributions of 766 non-duplicate bloodstream isolates of 15 Candida species are listed in Table 1. C. albicans was the most common of the Candida species identified, accounting for 45.4% of all cases, followed by C. tropicalis (17.6%), C. glabrata (17.4%), and C. parapsilosis (14.1%). The order of non-albicans Candida species (NAC) differed by year, but C. tropicalis and C. glabrata were the most common NAC in 2020 and 2021, respectively. Overall, the average incidence of candidemia was 1.58 cases per 10,000 PD, and the incidence of candidemia differed among the hospitals (1.06-2.22 cases per 10,000 PD). Species identification discrepancy between the collection center and the analysis center was noticed in nine (1.2%) isolates. The discrepancy rates varied according to the collec-tion centers (0.0-4.4%). Two isolates of C. albicans were misidentified as C. glabrata and C. tropicalis, and two isolates of C. parapsilosis were misidentified as C. albicans and C. tropicalis. Four isolates (one isolate each of C. glabrata, C. guilliermondii, C. dubliniensis, and C. ciferrii) were misidentified as C. albicans. One isolate of C. utilis was misidentified as C. fabianii.

Discussion
To the best of our knowledge, this is the first study to incorporate a national surveillance system for Candida BSIs into the GLASS bacterial surveillance system. The results showed that a Kor-GLASS-like surveillance system including isolate collection and centralized analysis components can provide reliable data for both Candida and bacterial blood pathogens. During the two-year study period, 10,758 BSI cases were caused by 10 Kor-GLASS target pathogens, and 4048 (37.6%) BSIs were classified as HO. Candida species ranked fourth among all target blood pathogens, and second among all HO blood pathogens, thus highlighting the important casual role of Candida species in nosocomial BSIs in South Korea. Surveillance at nine sentinel hospitals showed that 87.6% of candidemia cases were of HO, similar to the findings of an Australian study [21]. The average incidence of candidemia was 1.58 cases per 10,000 PD, but the incidence differed among hospitals (range: 1.06-2.22), as also reported in a previous Asian study [22].
In earlier multicenter studies conducted in South Korea, C. parapsilosis was the most common NAC isolated from patients with candidemia [16,23,24]. However, we found that, for the period from 2020-2021, C. glabrata and C. tropicalis were the most common NACs causing candidemia, consistent with recent reports on the changes in species distributions of candidemia in South Korea [8,25]. In studies from the USA, northern Europe, and Australia, C. glabrata, which is less susceptible to antifungal drugs, was the second most common cause of candidemia [10,21,26], while in tropical and Asian countries, such as The Philippines and Thailand, C. tropicalis was the most prevalent NAC [22,27]. In the present study, most candidemia cases occurred in patients aged >60 years with C. glabrata BSIs being more frequent in the elderly (>70 years) and C. parapsilosis candidemia being more frequent in males, in line with our recent study [8]. The reasons for the changing epidemiology of NAC BSIs in South Korea are unclear, but the use of antifungal agents, infection control practices, and types of at-risk hospitalized patients enrolled may be factors [5,8,10,26]. Given that each Candida species is unique in terms of virulence potential, antifungal susceptibility, and clinical characteristics, understanding the changing epidemiology is important for the proper management of candidemia [8,28]. In South Korea, fluconazole or amphotericin B were mainly used for the treatment of candidemia until the approval of echinocandins as primary treatment for severe candidiasis by the National Health Insurance Service (NHIS) in 2014. Following the approval of echinocandin use by the NHIS, the use of echinocandins for candidemia treatment has increased after 2014, and attention is required because of echinocandin resistance emergence [29,30].
Sensititre Yeast One was used for surveillance in this study, as it has been widely adopted by clinical microbiology laboratories for antifungal susceptibility testing and shows good concordance with the CLSI reference method for Candida susceptibility testing [31]. In addition, we examined several molecular mechanisms of antifungal resistance to better understand AMR epidemiology. While echinocandin resistance remains rare in Korea for six common Candida species (<0.5%), we observed non-susceptibility to fluconazole in about 20% of these species, in addition to frequent azole resistance among BSI isolates of three common NACs. The singular mechanism of acquired azole resistance identified in clinical isolates of C. glabrata is mutation of the transcription factor pleiotropic drug-resistance (PDR1), which leads to overexpression of the drug-efflux transporter genes CgCDR1, CgCDR2, and CgSNQ2 [11,31]. Our recent Korean study showed that 98.5% of fluconazole-resistant BSI isolates of C. glabrata and 0.9% of fluconazole susceptible dose-dependent BSI isolates of C. glabrata harbored an additional one or two Pdr1p AAS after exclusion of five genotype-specific AAS. In addition, the results highlight the high mortality rate of patients infected with fluconazole-resistant C. glabrata BSI isolates harboring Pdr1p mutations [11]. By contrast, multiple mechanisms of azole resistance, such as ERG11 mutations and overexpression of efflux pumps, have been reported for C. albicans, C. parapsilosis and C. tropicalis [31]; however, Y132F in ERG11 and N1132D in CDR1 were the major mechanisms of fluconazole resistance in C. parapsilosis isolates from Korean hospitals [7,19]. Therefore, the PDR1 gene was sequenced for all fluconazole-resistant isolates of C. glabrata while ERG11 and CDR1 genes were sequenced for all fluconazole-non-susceptible isolates of C. parapsilosis, and ERG11 gene was sequenced for all fluconazole-non-susceptible isolates of C. tropicalis, although other mechanisms might contribute.
Six of the seven fluconazole-resistant isolates of C. glabrata harbored diverse Pdr1p amino acid substitutions, which suggests that PDR1 mutation is the main cause of azole resistance in C. glabrata, in line with previous studies [11,32]. An azole-resistant C. parapsilosis isolate harbored a N1132D substitution in CDR1, as in another recent report [19]. Notably, all four fluconazole-non-susceptible C. parapsilosis isolates harboring a Y132F substitution in Erg11p were collected at the same hospital, indicating possible clonal transmission of these isolates [7]. One C. glabrata isolate was classified as caspofungin-resistant (MIC, 0.5 mg/L), with intermediate resistance to anidulafungin (MIC, 0.25 mg/L) but susceptibility to micafungin (MIC, 0.06 mg/L). In this isolate, echinocandin resistance was determined by DNA sequence analysis of the FKS genes. The isolate harbored an F659Y mutation in Fks2p, which confers resistance to all echinocandins [17].
An increasing incidence of candidemia in ICUs has been reported in many parts of the world [15,[33][34][35]. In a previous study based on Korean National Healthcare-Associated Infections Surveillance System, which included patients older than 15 years who developed candidemia during a stay of >2 days in the ICU, Candida species were the most frequently identified blood pathogens from 2013 to 2017 [15]. In the present study, 41.3% of candidemia cases occurred in ICU patients, and Candida species were the second most common BSI pathogen recovered from these patients (after E. coli). The rate of fluconazole resistance was significantly higher in ICU patients (4.7%) than ward patients (1.6%), as was the fluconazole resistance of C. glabrata (9.5% vs. 1.4%). Our recent study showed that, among patients with candidemia, mortality rates were approximately twofold higher in ICU than ward patients, and that fluconazole resistance was a predictor of C.-glabrata-associated mortality [8]. Taken together, these results highlight the importance of candidemia control in ICUs, including improved prevention and treatment strategies such as optimal use of antifungal treatments and less unnecessary use of catheters.
Accurate species identification is crucial considering the increasing rate of antifungal resistance among uncommon Candida species, such as Candida auris [36]. Our surveillance system used MALDI-TOF MS, supplemented by sequence-based identification of Candida species at an analysis center. In our study, discrepant identification results between the collection and analysis centers were seen for only 1.2% of the isolates (from five common and four uncommon Candida species). Given that all nine hospitals participating in Kor-GLASS used the MALDI-TOF MS system for the identification of yeast isolates, the misidentification of five isolates of common Candida species might be worthy of note. We found that a collection center using the MALDI-TOF Biotyper, which did not use FA extraction routinely, exhibited a relatively high misidentification rate (4.4%), even in common Candida species. For the analysis using the MALDI-TOF Biotyper, in-tube formic acid/acetonitrile (FA/ACN) extraction is recommended prior to the analysis; however, the use of a simple FA extraction method is preferable in order to facilitate the routine use in clinical microbiology laboratories [37]. Therefore, it should be considered that although MALDI-TOF system allows reliable and accurate identification of the clinical isolates of yeast species, their performance can be differed according to the protein extraction protocol (i.e., use of formic acid or not) or MALDI-TOF MS system databases used at the hospitals [37,38].
For some BSI isolates of rare Candida species for which CBPs were unavailable, including C. auris, C. fabianii, and C. orthopsilosis, resistance to fluconazole was higher than for other species (fluconazole MIC > 8 mg/L). In addition, a C. haemulonii isolate was determined to be multidrug-resistant (i.e., to both azole and amphotericin B), as reported previously [39]. These results suggest that the distribution of azole-and echinocandinresistant NAC isolates should be continuously monitored at the national level, and the resistance mechanisms determined.
A total of 20 university hospitals in South Korea participated in those studies, but the participating hospitals differed among years. In addition, in some of the studies, more than half of the Candida BSI isolates were from the two largest hospitals (>2000 beds), both of which are in Seoul, which is also where most of the antifungal-resistant isolates were found [7,11]. Also, in this study, hospital H, located at Jeju-island, off the peninsula's southern tip, showed a bit of a different pattern in the epidemiology of candidemia, in which C. albicans did not exceed NAC. It can be partly explained by their geographical and ecological conditions that may have influenced the epidemiology of candidiasis. Overall, the incidence rate and AMR of Candida BSIs vary greatly among institutions, with factors such as the number of beds, patients, and PD in the ICU or hemato-oncology wards likely playing important roles. Although whether the participating hospitals are representative is always debatable in national candida AMR surveillance studies, this concern can be partly overcome by incorporating a national surveillance system for BSI into the GLASS system, which permits continuous monitoring of target blood pathogens and is free from collection bias and isolate duplication. The Kor-GLASS system was designed to collect and analyze complete, non-duplicate clinical isolates and information from sentinel hospitals located in various districts throughout South Korea (each with a capacity of 600-1100 beds) caring for both inpatients and outpatients. The surveillance data demonstrated both the changing epidemiology of Candida species causing BSIs during 2020-2021, and the continued emergence of azole and echinocandin resistance (and thus of new resistance mechanisms) among BSI isolates of NAC. Continuous monitoring is therefore warranted.

Conclusions
We incorporated AMR surveillance for invasive Candida into preexisting GLASS bacterial surveillance system, and herein, we describe results from the two year (2020-2021) of operation of Kor-GLASS Candida surveillance based on early implementation protocol. The well-curated surveillance data highlight the significance of candidemia both as the second-most common cause of BSIs of HO and the second-most common pathogen in ICU patients in South Korea. Our study showed that the inclusion of Candida species in the Kor-GLASS system allowed representative and accurate monitoring of the antifungal resistance of BSI-causing candida on a national scale. As such, it will support further global efforts aimed at candidemia AMR surveillance.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jof8100996/s1, Figure S1: Distribution of the collection centers of the Kor-GLASS surveillance system in South Korea.  Informed Consent Statement: A written informed consent was waived by the nature of this study.
Data Availability Statement: All data generated or analyzed in this study are included in this published article, and the datasets are available from the corresponding author within the limits imposed by ethical and legal dispositions.