The Risk Factors and Mechanisms of Azole Resistance of Candida tropicalis Blood Isolates in Thailand: A Retrospective Cohort Study

In recent decades, an epidemiological shift has been observed from Candida infections to non-albicans species and resistance to azoles. We investigated the associated factors and molecular mechanisms of azole-resistant blood isolates of C. tropicalis. Full-length sequencing of the ERG11 gene and quantitative real-time RT-PCR for the ERG11, MDR1, and CDR1 genes were performed. Male sex (odds ratio, 0.38), leukemia (odds ratio 3.15), and recent administration of azole (odds ratio 10.56) were associated with isolates resistant to azole. ERG11 mutations were found in 83% of resistant isolates, with A395T as the most common mutation (53%). There were no statistically significant differences in the expression of the ERG11, MDR1, and CDR1 genes between the groups resistant and susceptible to azole. The prevalence of azole-resistant isolates was higher than the usage of antifungal drugs, suggesting the possibility of environmental transmission in the healthcare setting. The unknown mechanism of the other 17% of the resistant isolates remains to be further investigated.


Introduction
Candida species are the most common cause of invasive fungal infections, affecting around three-quarters of a million patients per year [1]. Although C. albicans is the major species causing worldwide candidemia, the prevalence of non-albicans candidemia is increasing, especially in the Asia-Pacific region [2]. Moreover, C. tropicalis has become the major cause of Candida bloodstream infection in Thailand [3,4].
C. tropicalis bloodstream infection also raised concerns about the high rate of death from cases [4]. Furthermore, the prevalence of azole resistance is an increasing trend in Thailand, South Korea, and China, which might be associated with the use of fungicides in agriculture and cross-transmission to the hospital [5][6][7][8]. However, the azole resistance rate is lower in some other Asian countries including India and Iran [9,10]. The prevalence of fluconazole-resistant C. tropicalis isolates in Thailand increased from 20.2% in 2013-2015 to 37.8% in 2016-2017 [8,11]. Similarly, in China, fluconazole-resistant isolates increased from less than 10% in 2010-2012 to more than 30% in 2016-2017 [7]. Several studies reported the relationship between azole-resistant Candida infection and identified multiple risk factors including previous exposure to fluconazole and a history of bacterial bloodstream infection [12,13]. The description of C. tropicalis azole resistance was attributed to the four major mechanisms related to ERG11 mutations, ERG11 overexpression, alteration of membrane sterol composition, and increased expression of the efflux pump genes [14]. However, the frequency of each mechanism has been reported to vary among different demographic locations [15,16].
In Thailand, there are few data on the association between azole-resistant isolates of C. tropicalis and the underlying molecular mechanisms of azole resistance. We aimed to analyze the ERG11 mutations and the expression of genes responsible for resistance among

Definitions and Data Collection
We collected data from medical records including demographic data, comorbidities, predisposing factors, and clinical characteristics. Comorbidities were assessed using Charlson's Comorbidity Index [17]. The source of candidemia was defined using the Infectious Diseases Society of America guidelines [18]. Predisposing factors on the day of bloodstream infection diagnosis included corticosteroid and antimicrobial administrations, parenteral nutrition, transplantation within 30 days, mechanical ventilation, placement of a central venous catheter at the time of onset, neutropenia (absolute neutrophil count ≤ 500 cells/mL). The severity of clinical presentation was evaluated using the Sequential Organ Failure Assessment Score (SOFA) [19]. Isolates from patients with either non-retrievable data or no antifungal susceptibility test results were excluded ( Figure 1). The association between azole resistance and candidemia due to C. tropicalis was divided into two groups: the susceptible group (minimal inhibitory concentration {MIC} for fluconazole ≤ 2 µg/mL) and the non-susceptible group (MIC for fluconazole ≥ 4 µg/mL) [20,21].

Antifungal Susceptibility Test
All recovered C. tropicalis isolates were subjected to a retrospective antifungal susceptibility test (AFST) using Sensititre YeastOne (ThermoFisher, Waltham, MA, USA) which comprised nine drugs (fluconazole, voriconazole, itraconazole, posaconazole, 5-flucytosine, anidulafungin, micafungin, caspofungin, and amphotericin B). Interpretations of this colorimetric microdilution method followed the manufacturer's recommendations. To ensure the commercial test's reliability, all viable isolates used for the molecular studies were retested for fluconazole susceptibility by the broth microdilution method following the Clinical Laboratory Standards Institute (CLSI) documents M27A3 and M60 [21,22]. Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were used as quality controls. MICs by the commercial test were highly reproducible, with 97.7% of repeat testing results being within a single 2-fold dilution of the samples as shown in Table S2.

Sequencing of the ERG11 Gene
Complete sequencing of the ERG11 gene was performed on 53 isolates of the fluconazoleresistant group (one isolate was lost during sample collection) and 36 randomly selected isolates of the fluconazole susceptible group. DNA was extracted following the protocol described previously [23]. The amplified PCR products were sequenced using four primer pairs (Table S1). The PCR protocol was used following the manufacturer's recommendation for the Phusion Plus DNA Polymerase (ThermoFisher, Waltham, MA, USA) which recommended denaturation at 98 • C for 10 s, annealing at 60 • C for 10 s, and extension at 72 • C for 30 s. The amplified fragment was purified and directly used for bidirectional sequencing by Macrogen Inc. Korea using the PCR primers. The nucleotide sequences were examined for sequencing errors and heterozygous. The heterozygous was visually detected as the double peaks with strong intensity on the same locus and both strands and edited by Bionumeric software (version 8.0; Applied Maths NV, Sint-Martens-Latem, Belgium). The edited nucleotide sequences were compared with the ERG11 of C. tropicalis MYA-3404 (GenBank accession number: XM_002550939).

Quantitative Real-Time RT-PCR of the ERG11, CDR1, and MDR1 Genes
Quantitative real-time RT-PCR was carried out on 80 isolates consisting of 47 isolates with fluconazole resistance (A total of 53 fluconazole-resistant isolates were excluded eight slow-grower isolates.) and 33 randomly selected isolates with fluconazole susceptibility. The C. tropicalis isolates were grown in the yeast extract peptone dextrose (YPD) broth and incubated at 37 • C in a shaker incubator at 200 rpm until the mid-log phase. Then, fluconazole at 1 4 MIC level of each isolate was added to the media and continued incubation for 2 h which all isolates were grown to exponential phase. Meanwhile, the isolates were grown in the drug-free media under the same condition that was used as controls. After that, the isolates with-and without-the drug condition were extracted for total RNA [24]. Total RNA products were extracted using RNeasy Kits (Qiagen, Hilden, Germany) and reverse transcribed using the High-Capacity cDNA reverse transcription kit (ThermoFisher, Waltham, MA, USA). The cDNA products were obtained from the ERG11, CDR1, and MDR1 genes using primers in Table S1 and measured with the LightCycler 480 instrument (Roche, Basel, Switzerland). The experiments were technical triplicated to ensure the reliability of the method. The expression of these genes was determined by the 2 −∆∆CT method using the ACT1 gene as an internal control [25]. The studied isolates were compared to the three genes' expression during exposure to the drug relative to their expression's growth in the drug-free condition.

Statistical Analysis
The minimal sample size for the prediction model of 137 patients was calculated using an azole resistance rate of 30% with 237 isolates over 5 years [8]. The categorical variables are presented as frequencies and compared using the chi-squared test or Fisher's exact test. Continuous variables are shown as medians and interquartile ranges and analyzed with Student's t-test and ANOVA. Univariate and multivariate regression analyses selected variables with a p-value < 0.2 and presented them with an odds ratio (OR) and 95% CI. The reliability and internal validation of the prediction model were using the rule of 10 events per covariate and bootstrapping, respectively [26]. All statistical analyses were performed with IBM SPSS Statistics for Windows (version 22.0; IBM, Armonk, NY, USA) and Prism software (version 9.4.1; GraphPad, San Diego, CA, USA).

Results
188 isolates of C. tropicalis bloodstream infections were included in the study. Fortynine isolates were excluded, which consisted of 20 isolates with unretrievable data and 29 isolates with a lack of AFST results. The prevalence of non-susceptible fluconazole C. tropicalis was 40.4% (76 isolates) which consists of 54 fluconazole-resistant isolates and 22 fluconazole susceptible-dose dependent isolates.

Demographic Data and Clinical Characteristics
Age, admission to intensive care, Charlson's comorbidity index, and severity at the time of onset were similar for both the fluconazole susceptible and resistant groups. Leukemia, lymphoma, recent chemotherapy administration, and exposure to carbapenem and azole were more common among patients with non-susceptible C. tropicalis blood isolates, as presented in Table 1.

Missense Mutation in the ERG11 Gene
The ERG11 missense mutations were found in 44 of 53 resistant isolates (83.0%, Table 3) and absent in all susceptible strains. The most common base substitution was A395T (28 of 53, 52.8%) and the A395W substitution was the second most common (13 of 53, 24.5%). The A428G substitution was founded in two of 53 isolates (3.8%) and the A395W/T769C substitution was founded in only one isolate (1 of 53, 1.9%).

Expression Level of the ERG11, MDR1, and CDR1 Genes
The mean expression level in the ERG11, MDR1, and CDR1 genes was 1.10, 1.05, and 1.16, respectively (all data shown in Table S2). There were no statistically significant differences in the expression of the ERG11, MDR1, and CDR1 genes between the azole-resistant and azole-susceptible groups (Figure 2). Interestingly, we found the azole-resistant isolates without ERG11 mutations (R-WOM) had a significantly higher level CDR1 expression (2.07) in the CDR1 gene than the resistant isolates with ERG11 mutations (R-WM, 1.01) and susceptible group (S, 1.13).

Expression Level of the ERG11, MDR1, and CDR1 Genes
The mean expression level in the ERG11, MDR1, and CDR1 genes was 1.10, 1.05, and 1.16, respectively (all data shown in Table S2). There were no statistically significant differences in the expression of the ERG11, MDR1, and CDR1 genes between the azole-resistant and azole-susceptible groups ( Figure 2). Interestingly, we found the azole-resistant isolates without ERG11 mutations (R-WOM) had a significantly higher level CDR1 expression (2.07) in the CDR1 gene than the resistant isolates with ERG11 mutations (R-WM, 1.01) and susceptible group (S, 1.13).

Discussion
This retrospective study revealed the prevalence of strains with dose-dependent fluconazole susceptibility and resistance was slightly higher than a previous report in 2013-2015, which found that the prevalence of dose-dependent susceptible and resistant strains was 9.5% to 11.7% and 20.2% to 28.7%, respectively [11].
Interestingly, the rate of recent exposure to azole was less (17%) than the rate of resistance to azole (28.7%). The earlier study reported that the azole-resistant group had a history of azole use in only one-third of the isolates [27]. Environmental spreading of resistant C. tropicalis strains was possible, according to previous reports on C. tropicalis and C. parapsilosis [6,28]. However, we have no direct evidence of patient-to-patient transmission, although the history of azole exposure was still a strong independent risk factor as in previous studies [12,13].
We report for the first time that leukemia and female sex are factors associated with C. tropicalis isolates resistant to azole. Leukemia increases exposure to antifungal prophylaxis to prevent invasive fungal disease after chemotherapy [29]. Men have lower rates of exposure to antifungal agents than women who may have chronic vulvovaginal candidia-sis [30]. Therefore, we recommend the strict use of antifungal treatment according to the indications of the guidelines to reduce antifungal pressure [31,32].
The ERG11 mutations found in our study consisted of four sites that were reported in previous studies [15,16,33,34]. Furthermore, the A395T mutation was the major missense mutation site found in more than 80% of previous studies and resulted in a high fluconazole MIC, especially as a homozygous mutation [15]. The A395T mutation was shown to alter the binding of heme to azole without altering enzyme activity which may be the reason for the isolates in our study that this mutation does not need the ERG11 gene overexpression when counteracted with azoles [35]. However, further studies of the UPC2 genes which are involved in the ERG11 regulation are warranted [36].
Upregulation of the MDR1 and CDR1 genes may play a role in resistance to azoles, which was described in previous studies [15,16]. Similar to the previous studies, our results did not report a statistically significant difference in MDR1 and CDR1 gene expression between the azole-resistant and azole-susceptible isolates [15,37]. However, we found a higher level of expression of the CDR1 gene in the R-WOM isolates. A recent study also revealed the role of the Tac1, which regulates the overexpression of the CDR1 gene [38]. However, further investigations are needed to study the precise function of the other efflux pump genes such as CDR2 and CDR3 genes in azole-resistant C. tropicalis from our institute [37,38].
In conclusion, our results revealed two factors: female sex and leukemia that had not previously been reported in association with blood isolates of C. tropicalis resistant to azole. While ERG11 missense mutations played an important role in azole resistance, neither higher expression of the ERG11 nor drug transporter genes contributed to this phenotype. Other resistance mechanisms and additional genotyping to evaluate the potential the environmental transmission of the azole-resistant isolates remain to be further studied.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jof8100983/s1, Table S1. Primers used in this study; Table S2. Antifungal susceptibility test results, missense mutation of ERG11 gene and gene expression levels of Candida tropicalis blood isolates Refs. [33,39]  Informed Consent Statement: Patient consent was waived due to the retrospective nature of this study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the Thailand Personal Data Protection Act.