Candida dubliniensis: an appraisal of its clinical significance as a bloodstream pathogen.

A nine-year prospective study (2002-2010) on the prevalence of Candida dubliniensis among Candida bloodstream isolates is presented. The germ tube positive isolates were provisionally identified as C. dubliniensis by presence of fringed and rough colonies on sunflower seed agar. Subsequently, their identity was confirmed by Vitek2 Yeast identification system and/or by amplification and sequencing of the ITS region of rDNA. In all, 368 isolates were identified as C. dubliniensis; 67.1% came from respiratory specimens, 11.7% from oral swabs, 9.2% from urine, 3.8% from blood, 2.7% from vaginal swabs and 5.4% from other sources. All C. dubliniensis isolates tested by Etest were susceptible to voriconazole and amphotericin B. Resistance to fluconazole (≥8 µg/ml) was observed in 2.5% of C. dubliniensis isolates, 7 of which occurred between 2008-2010. Of note was the diagnosis of C. dubliniensis candidemia in 14 patients, 11 of them occurring between 2008-2010. None of the bloodstream isolate was resistant to fluconazole, while a solitary isolate showed increased MIC to 5-flucytosine (>32 µg/ml) and belonged to genotype 4. A review of literature since 1999 revealed 28 additional cases of C. dubliniensis candidemia, and 167 isolates identified from blood cultures since 1982. In conclusion, this study highlights a greater role of C. dubliniensis in bloodstream infections than hitherto recognized.


Introduction
Candida dubliniensis was first described in 1995 from oral cavities of human immunodeficiency virus (HIV)-infected individuals [1]. The species forms only a minor component of normal microbiota but has a worldwide distribution [2]. Despite its close relationship with C. albicans, which is the predominant pathogenic species, the etiopathologic role of C. dubliniensis has mostly been restricted to oral candidiasis. In recent years, however, C. dubliniensis has increasingly been reported from patients with candidemia [3][4][5][6][7][8][9][10][11]. Although the species is significantly less virulent and genetically more clonal than C. albicans [12][13][14], the reasons for its expanding role in invasive disease remain largely unknown. Here, we report the prevalence of C. dubliniensis in various clinical specimens over a nine-year period and discuss its role in nosocomial candidemia.

C. dubliniensis isolates and their identification
The study was carried out at Mycology Reference Laboratory (MRL) (Department of Microbiology, Faculty of Medicine, Kuwait University) and included all Candida spp. isolates obtained between 2002-2010. Candida spp. isolates either received from 15 different hospitals within Kuwait for identification and antifungal susceptibility testing or recovered from various clinical specimens at MRL were prospectively tested for germ tube formation. All germ tube positive isolates were streaked on sunflower seed agar [15] and incubated for 48 h at 30uC for formation of fringed and rough colonies and chlamydospore production. Subsequently, their identity was also confirmed by Vitek2 Yeast identification system and/or by amplification and sequencing of the ITS region of rDNA [16,17]. The study was approved by the Ethical Committee of Health Sciences Center and Ministry of Health.

Molecular identification, genotype determination and detection of 5-flucytosine resistance
The genotypes of C. dubliniensis isolates based on internal transcribed spacer (ITS) region of rDNA were determined by PCR amplification with genotype-specific primers and DNA sequencing as described previously [18,19]. PCR products (10 ml) were resolved by electrophoresis in 2% (wt/vol) agarose gels and presence of a single amplicon of expected size indicated the specific genotype. The results were extended by direct DNA sequencing of the ITS region (containing ITS-1, 5.8S rRNA and ITS-2) of rDNA. The amplicons obtained with ITS1 and ITS4 panfungal primers were purified and both strands were sequenced using BigDye terminator v3.1 cycle sequencing kit and ABI 3130xl GeneticAnalyzer (Applied Biosystems Inc.). The ITS1FS, ITS2, ITS3 and ITS4RS were used as sequencing primers [19,20]. Specific genotypes were assigned based on maximum identity in BLAST searches [21]. The detection of 5-flucytosine resistanceconferring mutations in CdFCA1 gene codon 29 was carried out by PCR amplification using FCA1F and FCA1R primers, the amplicons were purified and subjected to restriction digestion with Mbo I to generate RFLP patterns or sequenced as described previously [18,22]. Pair-wise comparisons with sequences of 5-FC-susceptible and 5-FC-resistant C. dubliniensis isolates were performed using ClustalW.

Susceptibility testing by E-test
Antifungal susceptibility by E-test was performed on RPMI 1640 agar medium supplemented with 2% glucose with pH adjusted to 7.0 with 0.165 M MOPS buffer as described previously [17]. Etest strips for fluconazole, amphotericin B, and 5-fluorocytosine were obtained from AB BIODISK (Solna, Sweden). The test was performed according to manufacturer's instructions. Briefly, 140 mm diameter petri plates were poured with 60 ml RPMI medium containing 1.5% agar and allowed to solidify. The agar surface was uniformly inoculated by nontoxic cotton swab dipped in yeast cell suspension of the test isolates after adjusting their turbidity to 0.5 McFarland standard. The plates for minimum inhibitory concentration (MIC) were read after 24 h of incubation at 35uC. The MICs were determined at the lowest drug concentrations at which the border of the elliptical inhibition zone intersected the strip scale. Reference strains of C. krusei (ATCC 6258), C. parapsilosis (ATCC22019) and C. albicans (ATCC90028) were used for quality control. The resistance to fluconazole ($8 mg/ml) was determined by applying revised CLSI/EUCAST breakpoints [21].

Statistical analysis
Mann-Whitney test was applied to determine significance of differences observed in mean MIC values of fluconazole during the three sub-periods of the study (2002 to 2004, 2005 to 2007, and 2008 to 2010). SPSS version 17.0 was used for statistical analysis and a P value of ,0.05 was considered as significant. Of these, 247 (67.1%) came from respiratory (sputum and endotracheal secretions) specimens, 43 (11.7%) from oral swabs, 34 (9.2%) from urine, 14 (3.8%) from blood, 10 (2.7%) from vaginal swabs and 20 (5.4%) from other specimens. Of the 121 isolates (including all bloodstream isolates) genotyped, 82 belonged to genotype 1 and 38 isolates belonged to genotype 4. The remaining isolates belonged to genotypes 5-9 [18]. All 38 isolates belonging to genotype 4 were resistant to 5-flucytosine and contained S29L mutation at codon 29 of CdFCA1 gene [18,22]. Of note was the enhanced frequency of isolation of C. dubliniensis from 14 blood specimens obtained from 14 patients. Of the total of 1154 Candida spp. blood culture isolates received, the distribution of C. dubliniensis during the three sub-periods of the study was as follows: one of 244 (0.4%) between 2002-2004, two of 356 (0.6%) between 2005-2007, and 11 of 554 (2%) between 2008-2010 ( Table 1). All C. dubliniensis candidemia patients were diagnosed at different time points during the indicated periods.

Results
The particulars of 11 patients yielding C. dubliniensis in blood cultures whose records were available for review are presented in Table 2. All patients were immunocompromised and had one or more risk factors. Three of them occurred in males. Their age ranged from 4-85 years. Four of the patients were treated with fluconazole, two each with voriconazole and amphotericin B (lipid formulation) and one with caspofungin. All blood culture isolates were susceptible to fluconazole, amphotericin B and caspofungin. A solitary blood culture isolate was resistant to 5-flucytosine and belonged to genotype 4. Four of the patients (Cases 1, 7, 10 and 11) died, two of them before the blood cultures became positive, hence no antifungal therapy was administered. C. dubliniensis was the only pathogen isolated from blood culture of two of these four patients.
The data on MIC50, MIC90, MIC range and geometric mean of MICs of C. dubliniensis isolates are presented in Table 3. All C. dubliniensis isolates that were available for testing were susceptible to amphotericin B and voriconazole. Eight (2.5%) isolates were resistant to fluconazole with MICs ranging between $8 mg/ml to 32 mg/ml [23].

Discussion
The true prevalence of C. dubliniensis fungemia largely remains unknown because of the difficulty in readily distinguishing this species from the morphologically similar species, C. albicans. This study is noteworthy as it prospectively identified all germ tube positive Candida bloodstream isolates for the presence of C. dubliniensis. We observed that prevalence of C. dubliniensis among bloodstream isolates increased from 0.4% between 2002-2004 to 2% between 2008-2010 ( Table 1). The reasons for increased occurrence of candidemia cases due to C. dubliniensis are unclear. Bloodstream C. dubliniensis isolates formed 3.8% (14 of 368) of the total isolates of this species recovered from all clinical specimens. In a previous study from Saudi Arabia, the overall prevalence of C. dubliniensis was 3.3% among 823 yeast isolates recovered from different clinical specimens [24]. Two (16.7%) of their bloodstream isolates were re-identified as C. dubliniensis. Several retrospective and prospective studies published during 2002-2011 have reported on the prevalence of C. dubliniensis among bloodstream Candida spp. isolates (Table S1) [6,10,[24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42]. Generally, the prevalence of C. dubliniensis varied between 0.5% to 7.0% with the exception of two studies involving small number of isolates [24,31], where it was 16.7% and 10.0%, respectively. In these two studies, only germ tube positive bloodstream isolates were included. In a recent fungemia surveillance study from Denmark, Arendrup et al. [41] reported a prevalence of 1.2% to 3.1% over a six-year period and 74 (2.6%) of C. dubliniensis isolates came from blood cultures.
Since the first description of C. dubliniensis from oral cavities of HIV-positive patients from Ireland [1,43], subsequent epidemiological studies have revealed that this species is prevalent globally in association with human [2] and non-human habitats with a possibility of inter-host transmission [44,45]. The species has now been reported from other body sites/specimens, such as vagina, urine, skin, and feces/gastrointestinal tract of both HIV-positive and HIV-negative patients [2,16,17,46,47]. There are now increasing reports that C. dubliniensis has the potential to cause invasive disease in different groups of immunocompromised patients [2], possibly originating from host's own flora. Although C. dubliniensis is closely related to C. albicans, it is responsible for far fewer infections in humans. It is rare that patients colonized with this species develop candidemia [2]. The reasons for this limited ability of C. dubliniensis to cause invasive disease has been the focus of recent studies [48][49][50]. It has been shown that C. dubliniensis genome lacks important hypha-related virulence genes (e.g., ALS3 and HYR1) and it also has limited ability to undergo yeast-tohyphal transformation [49,50], which in turn may affect its potential to invade deeper tissue. A review of literature since 1999 revealed 32 cases of bloodstream infection due to C. dubliniensis [3,4,[6][7][8][9]25,42,[51][52][53][54][55][56][57][58][59]. They originated from different geographic regions (Europe-8, North America-17, Argentina-4, Australia-2, Singapore-1). All of them had underlying conditions or risk factors including six with HIV infection. Their ages ranged between 1-68 years. Of 28 patients for whom detailed particulars were available (Table S2), 15 were males. Six of these cases occurred in pediatric age group and 14 of 28 (50%) patients died. In three of them, no antifungal agent was administered. One patient died due to rhabdomyosarcoma despite receiving treatment with fluconazole. It is noteworthy that first of the three cases of C. dubliniensis fungemia were reported from Europe in non-HIV-infected patients with bone marrow transplantation and chemotherapy-induced neutropenia [8]. This was followed by a report of four additional cases from the United States including one who was infected with HIV [3]. This was believed to be the first case of C. dubliniensis candidemia in HIV-infected patient. Subsequently, 21 additional cases of C. dubliniensis fungemia were reported from many other countries (Table S2). However, the isolation of nearly 200 strains of C. dubliniensis from blood indicates that 28 described cases of candidemia represent only a fraction of total candidemia cases caused by this species. Furthermore, since blood culture positivity from candidemia patients seldom exceeds 50%, C. dubliniensis may be responsible for far greater number of candidemia cases than hitherto recognized.
All our C. dubliniensis isolates were susceptible to voriconazole and amphotericin B. However, 2.5% (8 out of 320) of the isolates were considered resistant (MICs$8 mg/ml) according to harmonized CLSI and EUCAST susceptibility breakpoints for Candida spp., which do not include C. dubliniensis [23]. It is noteworthy that none of the 11 C. dubliniensis bloodstream isolates was resistant to fluconazole. Generally, C. dubliniensis isolates are known to be susceptible to a wide range of antifungal agents [36]. Recently, Arendrup et al. [41] reported occurrence of fluconazole resistance in 3.1% (2 of 65) of C. dubliniensis bloodstream isolates using EUCAST breakpoint (MIC.4 mg/ml). It is unclear if marginal increase in fluconazole MICs (as indicated by geometric mean, Table 3) in the present study have in any manner contributed to increased occurrence of C. dubliniensis candidemia during 2008-2010. In this context, a reference may be made to a recent publication by Oxman et al. [60], who found that a significant number of candidemia episodes were caused by isolates that showed reduced susceptibility to fluconazole while still considered to be fully susceptible. Although none of our candidemia patient was on fluconazole prophylaxis, it has been shown that exposure to fluconazole may enhance adherence of C. dubliniensis to oral epithelium [61] and may also facilitate replacement of C. albicans with C. dubliniensis [62]. The impact of fluconazole therapy/ prophylaxis on the epidemiology C. dubliniensis candidemia is not known. Some investigators believe that widespread exposure to azoles may have contributed to increasing incidence of less susceptible non-albicans Candida spp. as bloodstream pathogens [63], a view that has not been shared by others [64].
In conclusion, a 9-year prospective study on the prevalence of C. dubliniensis among bloodstream Candida spp. isolates with an overall prevalence of 1.2%, is presented. Of 14 cases of C. dubliniensis candidemia, 11 were diagnosed between 2008-2010, thus highlighting an increasing role of C. dubliniensis in bloodstream infections in Kuwait in recent years. These observations are consistent with the global trend pointing towards changing epidemiology of candidemia in favor of non-albicans Candida spp.