Fluconazole-Resistant Candida glabrata Bloodstream Isolates, South Korea, 2008–2018

Nearly all isolates harbored Pdr1p mutations and were associated with a high mortality rate.

C andida glabrata is a commensal yeast in the human gut, genitourinary tract, or oral cavity; however, it can cause serious bloodstream infections (BSIs) that result in substantial illness and death (1). Unlike other common Candida species, C. glabrata exhibits intrinsically low susceptibility to azole drugs, especially fluconazole, and rapidly acquires antifungal resistance in response to azole or echinocandin exposure (1)(2)(3). Although the incidence of echinocandin-and multidrug-resistant (MDR) C. glabrata BSIs is low, fluconazole resistant (FR) C. glabrata BSI isolates have been increasingly reported worldwide, typically at rates of 2.6%-10.6%, although these rates can reach 17% (4)(5)(6). Fluconazole resistance in C. glabrata is of particular concern because of the increased incidence of BSIs caused by this species in various locations worldwide (1,4,5). Acquired azole resistance in C. glabrata is most commonly mediated by overexpression of the drug-efflux transporter genes CgCDR1, CgCDR2, and CgSNQ2 through a gain-of-function (GOF) mutation in the transcription factor pleiotropic drug-resistance (PDR1) (2,7,8), although other mechanisms might contribute (9)(10)(11).
PDR1 mutations in C. glabrata associated with azole resistance have been shown to cause hypervirulence in a mouse model of systemic candidiasis, suggesting the need for careful monitoring of FR C. glabrata BSI isolates and their PDR1 mutations (7,12). To date, little substantial research has been conducted on PDR1 mutation incidence among FR C. glabrata BSI isolates from multicenter surveillance cultures or on mortality rates of patients infected with these PDR1 mutants. This deficit might be attributable to Pdr1p amino acid substitutions (AAS) found in FR and fluconazole-susceptible dose-dependent (F-SDD) isolates (7,13,14), which can impede determination of whether specific Pdr1p AAS result in fluconazole resistance. Therefore, the aim of this study was to investigate the clinical outcomes, molecular mechanisms, and genotypes associated with antifungal-resistant BSI isolates of C. glabrata collected during multicenter studies in South Korea during an 11-year period (2008-2018). We focused on the mortality rates of patients infected with FR C. glabrata BSI isolates harboring the Pdr1p mutation.

Microorganisms and Antifungal Susceptibility Testing
A total of 1,158 BSI isolates of C. glabrata were collected from 19 university hospitals in South Korea during January 2008-December 2018 (Appendix Table  1, https://wwwnc.cdc.gov/EID/article/27/3/20-3482-App1.pdf). All isolates were collected from routine blood cultures by using methods that varied among laboratories; only the first isolate from each patient was included. The hospitals participating in this laboratory-based nationwide multicenter surveillance system differed each year. All C. glabrata isolates were submitted to Chonnam National University Hospital (Gwangju, South Korea) for testing. Species identification was based on matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Biotyper; Bruker Daltonics, https://www. bruker.com) with library version 4.0, or sequencing of the D1/D2 domains of the 26S rRNA gene, to differentiate C. glabrata from cryptic species (C. nivariensis and C. bracarensis) within the C. glabrata complex (15). In vitro testing of susceptibility to fluconazole, micafungin, caspofungin, voriconazole, and amphotericin B was performed for all isolates according to the Clinical and Laboratory Standards Institute broth microdilution method (16 Multidrug resistance was defined as resistance to both fluconazole and echinocandins (2).

Clinical Characteristics
Candidemia was defined as the isolation of Candida from >1 blood culture (20), and cases with invasive candidiasis without candidemia or colonization were excluded. All demographic characteristics and clinical conditions potentially related to candidemia mortality rates at the time of candidemia onset were investigated (21)(22)(23). Previous use of antifungal agents was defined as administration within 3 months before the onset of candidemia. A lack of antifungal therapy was defined as no antifungal therapy or treatment with antifungals for <3 days; appropriate antifungal therapy was defined as the administration of >1 in vitroactive antifungal (according to the susceptibility pattern of the isolate) for >72 hours (23,24). Therapeutic failure was defined as either persistence of Candida in the bloodstream despite >72 hours of antifungal therapy or development of breakthrough fungemia during treatment with the indicated antifungal agents for >72 hours (23,24). All-cause mortality rates were assessed at 30 and 90 days after the first positive blood culture result. Mortality rates also were analyzed for patients with candidemia who were infected with 297 SDD isolates of C. glabrata as controls. This study was approved by the Institutional Review Board of Chonnam National University Hospital (approval no. CNUH-2020-117).

Multilocus Sequence Typing and Molecular Mechanisms
Multilocus sequence typing (MLST) and PDR1 sequencing were performed for all antifungal-resistant isolates of C. glabrata and for 212 F-SDD control isolates by using methods described previously (14,21,25). PDR1 sequences of each isolate were compared and analyzed on the basis of the reference PDR1 sequence of C. glabrata (GenBank accession no. FJ550269) (14). The FKS1 and FKS2 sequences of 79 isolates that exhibited full or intermediate resistance to micafungin (MIC >0.12 mg/L) or caspofungin (MIC >0.25 mg/L) were compared with those of C. glabrata (GenBank reference sequence nos. FKS1 XM_446406 and FKS2 XM_448401) (14). The expression levels of CgCDR1, CgCDR2, and CgSNQ2 were evaluated for 30 FR isolates of C. glabrata harboring FR-specific Pdr AAS and for 65 F-SDD control isolates without FR-specific Pdr AAS, as described previously (26,27). The cycle threshold (C t ) of each gene was normalized to that of URA3 to determine the ΔC t value. For all isolates, relative gene expression (ΔΔC T ) was reported as fold change calculated as the mean normalized expression level relative to that of C. glabrata ATCC 90030 (fluconazole MIC 8 mg/L, set as 1.0).

Statistical Analysis
Quantitative variables are expressed as means with standard deviations, whereas categorical variables are expressed as counts and percentages. Categorical variables were compared by using the χ 2 test or Fisher exact test, Student t-test or the Mann-Whitney U test to compare quantitative variables, as appropriate. Cox proportional hazards models were used to evaluate potential risk factors for 30-and 90-day mortality rates by calculating the hazard ratio (HR). The Kaplan-Meier and log-rank (Mantel-Cox) tests were used to calculate the 30-and 90-day survival probabilities in subgroup analyses. All data were analyzed by using SPSS Statistics 26.0 (IBM, https://www.ibm.com). Statistical significance was determined at a level of p<0.05.

Incidence of Antifungal Resistance
The annual proportion of C. glabrata BSI isolates among all Candida BSI isolates increased from 11.7% to 23.9% (mean 18.6%) during the study period ( Table  1). The rate of fluconazole resistance (MIC >64 mg/L) increased from 0% (0/68 isolates) to 8.3% (14/168 isolates) during the study period. Among the 1,158 BSI isolates of C. glabrata, 66 (5.7%) were resistant to fluconazole, 16 (1.4%) were resistant to echinocandin, and 6 (0.5%) were resistant to multiple drugs. Of the 16 echinocandin-resistant isolates, 6 (37.5%) were also resistant to fluconazole; thus, these isolates were MDR. Isolates of echinocandin-resistant and MDR C. glabrata were initially found in 2013 and then annually from 2016 to 2018. Resistance to amphotericin B (MIC >2 mg/L) was not detected in any isolate, but 79 (6.8%) isolates had voriconazole MICs that exceeded the ECV (0.25 mg/L). All 64 FR isolates were associated with a voriconazole MIC >0.5 mg/L.

Mortality Rate of FR Candida glabrata BSIs
The mortality rate for 64 patients with FR C. glabrata BSI isolates was 60.9% at 30 days (Appendix Table  2). Univariate Cox regression analyses revealed that a high Charlson comorbidity index (p = 0.051), liver disease (p = 0.015), intensive-care unit admission (p = 0.071), severe sepsis (p = 0.039), lack of antifungal therapy (p<0.001), azole monotherapy (p = 0.005), any combination antifungal therapy (p = 0.014), and appropriate antifungal therapy  Table 3). Kaplan-Meier survival analysis showed that the mortality dynamics of the FR group (64 patients) decreased during the study period, whereas the F-SDD group (297 patients) exhibited a plateau period of decreasing cumulative survival from 30 to 90 days, which was similar in each of the 4 years of the study period ( Figure). The median survival of patients with FR C. glabrata BSI was significantly shorter than that of patients with F-SDD C. glabrata (17 days for FR vs. 90 days for F-SDD; p<0.001 by log-rank test).

Discussion
After C. albicans, C. glabrata is the most common Candida species isolated from BSI in North America and in countries of central and northern Europe (1,4). C. glabrata was the fourth most common BSI-causing Candida species in many countries in Asia besides South Korea (6,28,29); however, increasing rates of C. glabrata with FR have been reported in China (30), and this strain is now the second most common species in South Korea (31). In this study, the FR rate of BSI isolates of C. glabrata were found to have increased from 0% (0/68) in 2008 to 8.3% (14/168) in 2018. No C. glabrata isolate collected during 2008-2012 was resistant to echinocandins, whereas 2%-3% were resistant to echinocandins during 2015-2018. The emergence of echinocandin-resistant BSI isolates of C. glabrata in South Korea might reflect the increased use of echinocandin antifungals as the initial option for candidemia after insurance coverage for echinocandins began in 2014 (32). Of 16 echinocandin-resistant isolates, 6 (37.5%) were also resistant to fluconazole, indicating multidrug resistance. Overall, our 11-year nationwide surveillance revealed an increasing incidence of C. glabrata causing BSI and an increasing propensity for development of antifungal resistance in South Korea, consistent with surveillance data from other countries (1,2,4,5,30).
Data are scarce regarding the mortality rates for patients with candidemia who are infected with FR C. glabrata BSI isolates. The 30-day mortality rates in patients infected with C. glabrata BSI isolates are 21.3%-48.6% (16,(33)(34)(35)(36)(37) but can reach 50%-60% among patients in intensive care units (38,39). However, few FR C. glabrata isolates were included in previous studies. We found that FR BSI isolates of C. glabrata in South Korea were associated with significantly higher 30-day (60.9%) and 90-day (78.2%) mortality rates, compared to BSIs caused by F-SDD strains (30-day mortality rate 36.4%, 90-day mortality rate 43.8%). The mortality dynamics of FR isolates indicated a rapid rise in cumulative mortality from 7 to 90 days after BSI onset. This mortality dynamic was distinct from that of patients with F-SDD BSIs, who exhibited a steady curve after 60 days, consistent with previous reports of C. glabrata BSIs (34,40). The median survival of patients with FR C. glabrata BSIs (17 days) was also significantly shorter than that of patients with F-SDD C. glabrata BSIs (90 days). These findings are consistent with the results in a recent report regarding C. glabrata BSIs in South Korea, which showed that a high fluconazole MIC was associated with a poor outcome, although only 5 isolates in that study were FR (37).
In this study, MLST revealed that 56.1% of FR and 56.3% of echinocandin-resistant BSI isolates belonged to ST7, which accords with ST7 being the most common MLST genotype (47.8%) in South Korea (21). We found that 100% (6/6) of MDR isolates belonged to ST7, which harbored the V239L mutation in the mismatch repair gene (MSH2) associated with hypermutability (21,25). Given that the utility of MSH2 gene mutations as antifungal-resistance markers remains controversial (41,42), further surveillance studies are needed. To date, few studies have been conducted on MLST genotype-specific differences in Pdr1p polymorphism among C. glabrata BSI isolates. We found that all 50 isolates of ST3 harbored the same Pdr1p AAS (P76S/P143T/ D243N), all 7 isolates of ST55 harbored E259G, and all 4 isolates of ST59 harbored T745A, suggesting the presence of MLST genotype-specific Pdr1p AAS. P76S/P143T/D243N in Pdr1p was found to be common in China, Iran, and Australia (13,14,43,44), which accords with the high prevalence of ST3 in the study collections. Thus, the results of this study suggest that 5 Pdr1p AAS are MLST genotype-specific; because these AAS were found in both FR and F-SDD isolates, we confirmed that they cannot be responsible for azole resistance.
A single-point mutation in PDR1 can contribute to azole resistance in C. glabrata (7,8). Our results show that, in FR isolates, AAS are scattered throughout the entire protein without distinct hotspots, as reported previously (7,13,41,45). Therefore, determining whether a certain Pdr1p AAS is a GOF mutation is difficult without data from gene editing experiments for all variable regions. A previous study identified 57 FR-specific AAS by comparing azole-susceptible and azole-resistant matched isolates recovered from different clinical specimens (7). Furthermore, 91% (74/81) of FR isolates from BSIs or vaginal infections contained a Pdr1 mutation, compared with 5.6% 36 Emerging  (1/18) of F-SDD isolates (25). In our study, we found that 98.5% (65/66) of FR BSI isolates and 0.9% (2/212) of F-SDD BSI isolates harbored an additional 1 or 2 Pdr1p AAS after exclusion of 5 genotype-specific AAS (P76S, P143T D243N, E259G, and T745A). After exclusion of 6 additional FR isolates that harbored 2 Pdr1p AAS (because determining which of the 2 AAS was critical for fluconazole resistance was difficult to determine), we found 49 Pdr1p AAS that were present alone in 59 FR isolates, strongly suggesting that these AAS were FR-specific. Of the 49 Pdr1p AAS, 16 have been described for FR isolates (7,13,14,25,(43)(44)(45)(46). In this study, FR isolates exhibited higher mean CgCDR1, CgCDR2, or CgSNQ2 expression levels, compared with F-SDD isolates; all FR C. glabrata isolates were also resistant to voriconazole (MIC >0.5 mg/L), implying that fluconazole and voriconazole resistance are governed by the same mechanism (i.e., a GOF mutation in the transcription factor for Pdr1p) (7,8). Overall, our findings demonstrate that most FR BSI isolates of C. glabrata in South Korea harbor FRspecific Pdr1p AAS.
The cause of the high mortality rate associated with FR C. glabrata BSIs remains unclear. In this study, we focused on FR-specific Pdr1p AAS. PDR1 mutations are associated with increased virulence of C. glabrata, expression of adhesins, and adherence to host epithelial cells (7,12,47,48). The fungal loads in the kidney, spleen, and liver were higher in mice infected with the FR Pdr1 mutant of C. glabrata than in mice infected with F-SDD isolates (12). C. glabrata might persist in the body by replicating inside phagocytes, eventually leading to cell lysis, rather than by active escape (the method used by C. albicans) (47,49). This process might partly explain the elevated cumulative mortality rate for patients with Pdr1 mutants. Appropriate antifungal therapy was the only independently associated protective factor, with respect to 30-and 90-day mortality rates, in patients infected with FR C. glabrata isolates. In patients who received inadequate antifungal therapy and azole monotherapy, the 30-day mortality rates were 90% (antifungal therapy) and 88.9% (azole monotherapy), which were significantly higher than those of the patients receiving combination therapy (36.4%) or appropriate antifungal therapy (47.7%). Previous antifungal exposure was not an independent risk factor for death among patients with FR isolates, although it was identified in 62.5% (40/64) of patients. Given that previous antifungal exposure is a risk factor for antifungalresistant Candida BSI (50), further studies including F-SDD C. glabrata BSIs might elucidate the relationship between previous antifungal exposure and death. Taken together, these findings suggest that the high mortality rate associated with FR C. glabrata BSIs can be explained by the combination of FR and the virulence of Pdr1 mutants.
The first limitation of our study is that a Candida species might develop resistance within a patient during antifungal therapy; such resistance can be iden-  (2) (1), K776E (1), G788W (2), tified through serial isolates, but we tested only the first isolate from each patient during 2016-2018. Second, our results did not show that FR, PDR1 mutants, or previous antifungal exposure were independent risk factors for death in patients with C. glabrata BSIs. A total of 1,158 nonduplicate BSI isolates of C. glabrata from 19 university hospitals in South Korea were obtained during the 11-year study period, and the hospitals participating differed each year; therefore, we could not select an appropriate control group of F-SDD isolates. These limitations were partly overcome in a recent study involving 197 adult patients with C. glabrata BSI during January 2010-February 2016 at 7 university hospitals in South Korea. In that study, FR was shown to be associated with the 30-day mortality rate in a multivariate analysis (37). Third, only limited numbers of patients infected with F-SDD BSI isolates of C. glabrata were included in our mortality analysis. Nevertheless, we included a total of 297 patients infected with F-SDD BSI isolates of C. glabrata, which included all patients with C. glabrata from the participating hospitals in 2010, 2014, 2017, and 2018. The 30-day mortality rates of patients infected with F-SDD C. glabrata isolates were similar among those 4 years (32.0%-39.2%), despite differences in participating hospitals and collection periods; the 30day mortality rate was similar to those reported in previous studies (21,(33)(34)(35)(36)(37).
In conclusion, we demonstrated that nearly all FR BSI isolates of C. glabrata in South Korea harbored FR-specific Pdr1p mutations by excluding MLST genotype-specific Pdr1p AASs and that the isolates were associated with higher 30-day (60.9%) and 90day (78.2%) mortality rates. These results suggest that Pdr1 mutants are associated with a risk for death in such patients. In addition, appropriate antifungal therapy was the only independent protective factor against death in patients with FR isolates. Because of the increasing prevalence of FR BSI isolates of C. glabrata worldwide, improved detection and appropriate antifungal treatments are critical.