Clinical and Microbiological Characteristics of Candida guilliermondii and Candida fermentati

ABSTRACT

A total of 46 clinical isolates of Candida guilliermondii and Candida famata were reidentified genetically, resulting in 27 C. guilliermondii and 12 Candida fermentati strains. The majority of C. guilliermondii strains, but not C. fermentati strains, were isolated from blood cultures. C. fermentati was more sensitive to antifungals, hydrogen peroxide, and killing by murine macrophages than was C. guilliermondii. The C. guilliermondii isolates were echinocandin susceptible in vitro but resistant to micafungin in a murine model of invasive candidiasis.

TEXT

The Candida guilliermondii complex is a genetically heterogeneous complex of several phenotypically indistinguishable species, including C. guilliermondii, Candida fermentati, Candida carpophila, and Candida xestobii (1). The incidence of candidemia due to the C. guilliermondii complex ranges from 1 to 3%, depending on the geographic region (2, 3). However, there have been limited studies reporting epidemiological and clinical information for C. guilliermondii complex infections (4–6). The C. guilliermondii complex and Candida famata share similarities in biochemical characteristics, and it has been reported that these species are sometimes misidentified in clinical laboratories (7, 8).

During a 12-year period from April 2005 to March 2017, 46 strains were isolated from 46 patients across 7 medical institutions in Japan. All isolates were originally identified as C. guilliermondii or C. famata, using the following standard laboratory identification systems: Vitek 2 (bioMérieux, Marcy l'Etoile, France), API ID 32C or API 20C AUX (bioMérieux), and matrix-assisted laser desorption ionization (MALDI) Biotyper (Bruker Daltonics, Bremen, Germany). Each institution was either a university hospital or a teaching hospital and included Nagasaki University Hospital, Aichi Medical University Hospital, Hyogo College of Medicine Hospital, Toyama University Hospital, Toranomon Hospital, Sasebo City General Hospital, and Sasebo Kyosai Hospital. For all isolates, nucleotides of internal transcribed spacer (ITS) regions were sequenced as described previously (9) and analyzed using the ClustalW algorithm in MacVector software (version 14.0.3, MacVector, Inc., NC). PCR/restriction fragment length polymorphism (RFLP) of the intergenic spacer (IGS) was performed to distinguish C. guilliermondii, C. fermentati, C. carpophila, and C. xestobii, as previously described (10). Identification results are shown in Table 1. Among 39 C. guilliermondii clinical isolates, 38 isolates were confirmed as C. guilliermondii, and one isolate was confirmed as Kodamaea ohmeri. Conversely, among seven clinical isolates originally identified as C. famata by the Vitek 2 system, six isolates were reidentified as Candida parapsilosis, and one isolate was reidentified as C. guilliermondii complex; no true C. famata strain was identified. The results of our study, in conjunction with those of previous reports (7, 8, 11, 12), suggest that C. famata is far less common as a cause of invasive candidiasis than other species. The MALDI Biotyper misidentified the K. ohmeri isolate, as this species was not included in the database when the isolate was identified in 2013 (13). All three laboratory identification systems correctly identified isolates as belonging to the C. guilliermondii complex clade, however, these systems are not able to accurately identify isolates to a species level within the C. guilliermondii complex. The ITS sequences of the clinical C. guilliermondii and C. fermentati isolates matched those of the reference strains CBS566 and CBS2022, respectively, with 99.6 to 100% similarities. Clinical C. fermentati isolates differed from C. guilliermondii isolates by 3 to 5 bp in the ITS region sequences. The homology of nucleotide sequences of the ITS regions between these two species was 99.0 to 99.3%. C. fermentati (CBS2022) differed from C. carpophila (CBS5256) and C. xestobii (CBS5975) by two nucleotides. The PCR/RFLP analyses of the IGS regions confirmed that 27 isolates were C. guilliermondii and 12 isolates were C. fermentati. There were no C. carpophila or C. xestobii isolates.

TABLE 1.

Identification results by laboratory identification systems and ITS sequencing

Identification system	Species identification (no. of isolates)
	Original	Confirmed by ITSa sequencing
Vitek 2 system	C. guilliermondii (6)	C. guilliermondii complex (6)
	C. famata (7)	C. guilliermondii complex (1)
		C. parapsilosis (6)
API series	C. guilliermondii (18)	C. guilliermondii complex (18)
MALDI Biotyper system	C. guilliermondii (15)	C. guilliermondii complex (14)
		Kodamaea ohmeri (1)

^aITS, internal transcribed spacer.

Clinical information, including specimen types, comorbidity, β-d-glucan values measured by Fungitec G Test MK II “Nissui” (cutoff value; 20.0 pg/ml; Nissui Pharmaceutical Co. Ltd., Tokyo, Japan), initial antifungal agents administered, and therapeutic outcomes, was reviewed retrospectively, for all 39 patients from which the C. guilliermondii complex had been isolated. The investigators determined whether isolates caused infection or colonized based on clinical courses. The study protocol was approved by the ethical review boards in all institutions that participated in this study. The registration number of this study is 14122267 at the principal investigator's institution, Nagasaki University Hospital. Patient characteristics are shown in Table 2. All statistical analyses were carried out using Prism 6.0 (GraphPad Software, Inc.). Nominal variables were compared using Fischer's exact test, and continuous variables of patient characteristics were compared using the Mann-Whitney U test. Of the 39 patients, 28 patients (71.8%) were diagnosed with complicated malignancies; notably, 17 patients (43.6%) had underlying hematological cancer. The C. guilliermondii complex isolates were obtained from 31 patients (79.5%) with central venous catheters; 19 patients (48.7%) administered with steroids; and 21 patients (53.8%) receiving antifungal therapy with micafungin (n = 14), itraconazole (n = 1), fluconazole (n = 2), voriconazole (n = 2), and liposomal amphotericin B (n = 5). Among these patient characteristics, no significant differences were found between the C. guilliermondii and C. fermentati groups. The only clinical difference was that none of the 12 C. fermentati isolates were obtained from the bloodstream, whereas 81.5% (n = 22/27) of the C. guilliermondii isolates were obtained from the bloodstream (P < 0.0001). The other C. guilliermondii isolates were from cerebrospinal fluid (n = 1), ear discharge (n = 1), sputum (n = 2), and the urinary tract (n = 1). In contrast, the majority of C. fermentati isolates were obtained from nonsterile sites, including stool (n = 7), sputum (n = 2), bile (n = 2), and the urinary tract (n = 1). In previous studies, among the clinical cases diagnosed as C. guilliermondii infection, 77 to 95% of cases were actually caused by C. guilliermondii, while 5 to 23% cases were due to C. fermentati (4, 5, 14, 15). The results of our study were in agreement with a previous report that C. guilliermondii was more commonly isolated from the bloodstream than C. fermentati (5). We collected C. guilliermondii complex isolates from any type of specimen, regardless of whether the isolates had caused infection; in contrast, most previous studies analyzed only infectious cases, including candidemia. The elevation of serum β-d-glucan level was more frequently found in the C. guilliermondii group (n = 16; 64.0%) than in the C. fermentati group (n = 2; 16.7%) (P = 0.01). The findings in our study suggest that C. fermentati strains colonize at nonsterile sites, but rarely invade into the bloodstream, an interpretation which may also be supported by the significantly low serum β-d-glucan levels in the C. fermentati group.

TABLE 2.

Patient characteristics associated with C. guilliermondii complex isolates

Characteristic	C. guilliermondiia	C. fermentatib	P value
Age, yrs (mean ± SD)	63.3 ± 18.7	61.2 ± 19.0	0.75
Male sex (no. [%])	20 (74.1)	7 (58.3)	0.46
Comalignancies (no. [%])	18 (66.7)	10 (83.3)	0.45
Steroid use (no. [%])	11 (40.7)	8 (66.7)	0.18
Central venous catheter (no. [%])	22 (81.5)	9 (75.0)	0.68
In-hospital death (no. [%])	13 (48.1)	6 (50.0)	1.00
Antifungal therapy (no. [%])	12 (44.4)	9 (75.0)	0.10
Elevated β-d-glucan (no. [%])	16 (64.0)c	2 (16.7)	0.01
Bloodstream isolates (no. [%])	22 (81.5)	0 (0)	<0.0001
Causes of infection (no. [%])	24 (88.9)	0 (0)	<0.0001

^an = 27.

^bn = 12.

^cSerum β-d-glucan was measured for 25 patients.

Antifungal susceptibility tests were performed using the Sensititre YeastOne (SYO) microtiter panel (TREK Diagnostic Systems, Ltd., East Grinstead, UK) (16). The MICs of C. guilliermondii complex isolates were interpreted by species-specific clinical breakpoints (CBPs) (17) and epidemiological cutoff values (ECVs) (18, 19). The antifungal susceptibilities of the 27 C. guilliermondii and 12 C. fermentati isolates are shown in Table 3. All of the C. guilliermondii and C. fermentati isolates were susceptible to micafungin, caspofungin, and anidulafungin, except for one C. guilliermondii isolate that was categorized as intermediate to anidulafungin when interpreted using the CBPs. All of the C. guilliermondii and C. fermentati isolates were categorized by using the ECVs as wild type for echinocandins, amphotericin B, and 5-flucytosine. C. guilliermondii is known to show intrinsically higher echinocandin MIC values than other Candida species (2), and a recent study has reported that 9.1 to 27.2% of C. guilliermondii isolates were not susceptible to echinocandins (6). In contrast, most C. guilliermondii isolates in this study were susceptible to echinocandins. The reason for this difference is unclear but may be related to geographic location, prior exposure to echinocandins, or other unknown factors. Nonetheless, all of the C. fermentati isolates were susceptible to the nine antifungal agents tested in this study, in agreement with the previous reports from China and Taiwan (4, 5). Taken together, these findings suggest a lower frequency of resistant C. fermentati strains and certain variations in the frequency of echinocandin resistance in C. guilliermondii. According to previous studies, approximately 5 to 15% of C. guilliermondii isolates are azole resistant (2, 4, 5, 15). Correspondingly, 7.4 to 14.8% of C. guilliermondii isolates were categorized as non-wild type for azoles, according to the ECVs in our study. All of the C. fermentati isolates were categorized as wild type for all azoles tested in this study.

TABLE 3.

In vitro susceptibility of 27 C. guilliermondii isolates and 12 C. fermentati isolates to 9 antifungal agents^a

C. guilliermondii complex	MIC (μg/ml)			No. (%) of isolates by ECVsb		No. (%) of isolates by CBPsc
	Range	MIC50	MIC90	WT	Non-WT	S	I	R
C. guilliermondii (n = 27)
Fluconazole	2–>256	4	128	23 (85.2)	4 (14.8)
Itraconazole	0.25–>16	0.5	2	24 (88.9)	3 (11.1)
Voriconazole	0.03–8	0.12	0.5	25 (92.6)	2 (7.4)
Posaconazole	0.06–2	0.25	0.5	25 (92.6)	2 (7.4)
5-Flucytosine	≤0.06	≤0.06	≤0.06	27 (100)	0 (0)
Micafungin	0.25–1	0.5	1	27 (100)	0 (0)	27 (100)	0 (0)	0 (0)
Caspofungin	0.12–1	0.5	1	27 (100)	0 (0)	27 (100)	0 (0)	0 (0)
Anidulafungin	0.5–4	2	2	27 (100)	0 (0)	26 (96.3)	1 (3.7)	0 (0)
Amphotericin B	0.25–2	0.5	1	27 (100)	0 (0)
C. fermentati (n = 12)
Fluconazole	2–8	4	4	12 (100)	0 (0)
Itraconazole	0.25–0.5	0.25	0.5	12 (100)	0 (0)
Voriconazole	0.06–0.25	0.12	0.25	12 (100)	0 (0)
Posaconazole	0.03–0.25	0.12	0.25	12 (100)	0 (0)
5-Flucytosine	≤0.06–0.12	≤0.06	0.12	12 (100)	0 (0)
Micafungin	0.25–1	0.5	1	12 (100)	0 (0)	12 (100)	0 (0)	0 (0)
Caspofungin	0.25–1	0.5	0.5	12 (100)	0 (0)	12 (100)	0 (0)	0 (0)
Anidulafungin	0.5–2	1	2	12 (100)	0 (0)	12 (100)	0 (0)	0 (0)
Amphotericin B	0.5–1	0.5	1	12 (100)	0 (0)

^aInterpreted by breakpoints for echinocandins and epidemiological cutoff values for other agents.

^bEpidemiological cutoff values (ECVs) (19): fluconazole, 8 μg/ml; itraconazole, 1 μg/ml; voriconazole, 0.25 μg/ml; posaconazole, 0.5 μg/ml; 5-flucytosine, 1 μg/ml; micafungin, 2 μg/ml; caspofungin, 2 μg/ml; anidulafungin, 4 μg/ml; and amphotericin B, 2 μg/ml. WT, wild type.

^cSpecies-specific clinical breakpoints (CBPs) of echinocandins in CLSI document M27-S4 (17): S, ≤2 μg/ml; I, 4 μg/ml; R, ≥8 μg/ml. S, susceptible; I, intermediate; R, resistant.

The sequences of the FKS1 hot spot regions of four C. guilliermondii and eight C. fermentati isolates were analyzed. Hot spot regions were amplified with the following forward and reverse primer pairs: HS1-F (5′-AATGGGCTGGTGCTCAACAT-3′) and HS1-R (5′-CCTTCAATTTCAGATGGAACTTGATG-3′) for hot spot 1, and HS2-F (5′-AAGATTGGTGCTGGTATGGG-3′) and HS2-R (5′-GTGGCGAAACCTCTACCAGT-3′) for hot spot 2. The reference sequences of FKS1 were retrieved from the NCBI database and sequence analyses were performed with MacVector software. Decreased echinocandin susceptibility of C. guilliermondii was attributed to the intrinsic amino acid changes in the hot spot 1 region of Fks1 (20–22). However, to our knowledge, there is only one study that analyzed the DNA sequence of the FKS1 hot spot region for a C. fermentati clinical isolate (23). The present study revealed that all of the C. guilliermondii and C. fermentati isolates harbored two polymorphisms (L633M and T634A) in the first Fks1 hot spot, which may account for intrinsically higher echinocandin MICs of these strains. No mutation was found in the second Fks1 hot spot region.

To examine in vivo echinocandin susceptibility of C. guilliermondii, mice were infected intravenously with the C. guilliermondii clinical isolates NGSCG1 and ACHCG213 and treated with micafungin. The Candida glabrata wild-type strain CBS138 (24) was used as a control, since micafungin is known to be highly active against C. glabrata. In vitro MICs of micafungin were 1 μg/ml for NGSCG1, 0.25 μg/ml for ACHCG213, and 0.03 μg/ml for CBS138. All animal experiments were performed in full compliance with the Guide for the Care and Use of Laboratory Animals (25) and institutional regulations and guidelines for animal experimentation, after pertinent review and approval by the Institutional Animal Care and Use Committee of Nagasaki University under protocol number 1407281164. Specific-pathogen-free, 7-week-old female BALB/c mice (Japan SLC Inc., Shizuoka, Japan) were rendered neutropenic by intraperitoneal administration of cyclophosphamide (Sigma-Aldrich, St. Louis, MO) 4 days before infection (150 mg/kg), 1 day before infection (100 mg/kg), and 2 and 5 days postinfection (100 mg/kg) (26). The mice were infected intravenously through the lateral vein with 0.2 ml of the Candida cell suspension. The actual CFU in the inocula were confirmed by plating serial dilutions of cell suspension on yeast extract-peptone-dextrose (YPD) plates and were as follows: 7.2 × 10⁶ cells for NGSCG1, 8.6 × 10⁶ cells for ACHCG213, and 1.6 × 10⁶ cells for CBS138. Micafungin (Astellas Pharma Inc., Tokyo, Japan) was administered intraperitoneally at 4 mg/kg/day for 7 consecutive days in a 0.2-ml volume, commencing 2 h postinfection. Taking into consideration the previous area under the concentration-time curve (AUC) data for micafungin in mice (27, 28), 4 mg/kg/day intraperitoneal administration to mice is expected to correspond to approximately 100 mg/day administration to humans. The mice were euthanized 7 days after infection. No mice died before euthanasia in this experiment. Appropriate dilutions of organ homogenates were plated on YPD agar, and the CFU per organ were calculated. Differences of fungal burden between the treatment and control groups were examined using the Mann-Whitney U test. In mice infected with C. guilliermondii isolates NGSCG1 and ACHCG213, micafungin at 4 mg/kg was effective at reducing the liver fungal burden compared against the controls (P = 0.0002 for NGSCG1 and P = 0.0006 for ACHCG213). However, the overall CFU reduction for C. guilliermondii was clearly less than that for C. glabrata CBS138 (Fig. 1). In mouse kidneys infected with C. guilliermondii, micafungin was not effective to reduce fungal burden compared to in the nontreatment controls (P = 0.98 for NGSCG1 and P = 0.38 for ACHCG213), although it significantly reduced the CFU in the liver and kidneys of mice infected with C. glabrata CBS138 (P = 0.0002). The lower efficacy of micafungin against C. guilliermondii may be due to this strain having higher echinocandin MIC values than does C. glabrata. A 4 mg/kg dose of micafungin was effective at reducing the liver fungal burden in mice infected with C. guilliermondii isolates, yet ineffective in the kidneys. It has been reported that micafungin concentrations in kidney tissues are less than those in liver tissues (29), and that micafungin translocates rapidly in and out of the kidneys (30). The differences in micafungin concentrations in the kidney and liver tissues may be reflected in therapeutic efficacy at those organs, although we were not able to measure the actual drug concentrations in our mouse experiments. Nevertheless, our study showed that, at a clinical dose, micafungin had a poor efficacy against C. guilliermondii, while it was markedly more effective for C. glabrata.

FIG 1.

Treatment efficacy of micafungin against C. guilliermondii clinical isolates and a C. glabrata wild-type strain in a mouse model of disseminated candidiasis. Immunocompromised mice were inoculated with C. guilliermondii NGSCG1, C. guilliermondii ACHCG213, and C. glabrata CBS138. Micafungin was administered intraperitoneally at 4 mg/kg/day for seven consecutive days. The number of cells recovered from the liver and both kidneys is indicated for individual mice in the plots, and error bars represent standard deviations (SDs). NT indicates no treatment. Asterisks and NS indicate a statistically significant difference (***, P < 0.001) and no significance (P > 0.05), respectively.

To examine differences in phenotype between C. guilliermondii and C. fermentati, we performed various assays related to pathogenicity. Biofilm formation capacity of Candida cells was examined using the 2,3-bis(2-methoxy-4-nitro-5-sulfo-phenyl)-2H-tetrazolium-5-carboxanilide (XTT) reduction assay (31). There was no significant difference in the capacity of biofilm formation between C. guilliermondii (optical density at 492 nm [OD₄₉₂], 0.112 ± 0.027) and C. fermentati (OD₄₉₂, 0.081 ± 0.026) (Kruskal-Wallis test; P = 0.059). The biofilm formation capacity of C. guilliermondii and C. fermentati was significantly less than that of the Candida albicans wild-type strain SC5314 (OD₄₉₂, 0.774 ± 0.036), respectively (P = 0.039 for C. guilliermondii versus C. albicans and P = 0.0001 for C. fermentati versus C. albicans). We also performed spot dilution tests as described previously (32) and found that C. fermentati isolates were more sensitive to H₂O₂ than were C. guilliermondii isolates (Fig. 2A). There were no differences in sensitivity between C. guilliermondii and C. fermentati strains tested in this study to other oxidative stress inducers, including diamide and menadione, osmotic stresses induced by sorbitol and NaCl, acid and alkaline stresses (pH 2.2 to 9.4), and growth at 30°C and 37°C (data not shown). To further examine H₂O₂ sensitivity of these species, viable cell counts of nine isolates for each of C. guilliermondii and C. fermentati were evaluated after incubation with 5 mM H₂O₂ for 4 h. To avoid potential for bias, these isolates were selected by considering medical institution, specimen type, and antifungal susceptibility profiles. Logarithmic-phase cells were suspended in phosphate-buffered saline (PBS) (pH = 7.2) at the concentration of 5 × 10⁶ cells/ml and treated with H₂O₂ at 30°C with agitation (250 rpm). Serial dilutions of cell suspensions were plated on YPD agar and incubated at 30°C for 48 h to count viable cells. H₂O₂ exerted significant fungicidal effects against C. fermentati (Fig. 2B and C). The average viable cell counts before treatment were 6.8 ± 0.3 −log cells/ml for C. guilliermondii and 6.7 ± 0.3 −log cells/ml for C. fermentati. A 4-h treatment with 5 mM H₂O₂ reduced viable cell counts of C. guilliermondii and C. fermentati by 1.1 ± 0.4 and 3.4 ± 0.5 −log cells/ml, respectively. There was a significant difference between the two groups (Mann-Whitney U test, P < 0.001). These results suggest that C. fermentati was more sensitive to H₂O₂ than was C. guilliermondii.

FIG 2.

Growth and killing assay in the presence of H₂O₂ and killing assay by RAW 264. (A) Serial dilutions of logarithmic-phase cells of C. guilliermondii and C. fermentati strains were spotted onto synthetic complete medium (SC) plates containing H₂O₂ at the indicated concentrations. C. guilliermondii strains: NGSCG3, TYMCG251, and CBS566; C. fermentati strains: NGSCF1, ACHCF243, and CBS2022. (B) Logarithmic-phase cells of C. guilliermondii and C. fermentati were treated with 5 mM H₂O₂ for 4 h. The number of cells at 0 h was C. guilliermondii was 6.8 ± 0.3 −log cells/ml and for C. fermentati was 6.7 ± 0.3 −log cells/ml. After 4 h of treatment, the number of cells for C. guilliermondii was 5.7 ± 0.6 −log cells/ml and for C. fermentati was 3.3 ± 0.5 −cells/ml. (C) The reduced viable cell counts of C. guilliermondii and C. fermentati were 1.1 ± 0.4 and 3.4 ± 0.5 −log cells/ml, respectively. There was a significant difference between the two groups (***, P < 0.001). (D) Logarithmic-phase cultures of Candida cell suspensions were cocultured with RAW 264 cells at 37°C for 4 h. Mean in vitro killing of C. guilliermondii and C. fermentati is expressed as the percent reduction of CFU recovered from cocultures compared with the CFU from control cultures (Candida cells without macrophages). Dots represent results of the killing ratio of individual clinical isolates. The percent killing ratios of C. guilliermondii and C. fermentati were −0.6 ± 14.4% and 24.4 ± 14.6%, respectively. There was a significant difference between the two groups (**, P < 0.01).

In the host environment, phagocytes are the first line of defense against fungal infections. These cells produce reactive oxygen species, such as superoxide, H₂O₂, and hydroxyl radicals, for damaging biomolecules and killing phagocytosed pathogens (33, 34). Murine macrophages are known to be capable of killing microbes, including Candida species (35, 36). We performed macrophage killing assays against C. guilliermondii and C. fermentati using murine RAW 264 macrophages. RAW 264 cells were cultured at 37°C in 5% CO₂ in Dulbecco's modified Eagle medium (DMEM; Sigma-Aldrich) supplemented with 10% fetal bovine serum (Life Technologies, Japan) and 1% penicillin and streptomycin (Sigma-Aldrich). Macrophages were scraped with trypsin-EDTA (0.25%) and phenol red (Invitrogen, Carlsbad, CA) and rinsed in DMEM. The nine clinical isolates of C. guilliermondii and C. fermentati, which were the same strains used for the H₂O₂ assay, were incubated in YPD broth at 30°C, and logarithmic-phase cells were prepared. The killing assay was performed by reference to the methods described previously (37, 38). On the basis of pilot studies, 8.0 × 10⁴ macrophages were cocultured with 5.3 × 10³ Candida cells (15:1 ratio) in 1.5-ml microtubes, with rotation at 37°C for 4 h. The cultures were sonicated, diluted, and spread on YPD agar to count the viable cells. The CFU of the cocultures were compared with the CFU of growth control tubes containing Candida cells without macrophages. The percentage of killing ratio was calculated as [1 − (CFU from coculture tubes/CFU from control tubes)]. The cell ratios before treatment were 14.7 ± 3.4: 1 (macrophage Candida) for C. guilliermondii and 13.1 ± 5.5: 1 for C. fermentati. Viable cell counts after a 4-h treatment were 2.43 ± 0.80 × 10⁴ cells/ml for C. guilliermondii and 2.56 ± 1.11 × 10⁴ cells/ml for C. fermentati without RAW 264 cells, and 2.39 ± 0.81 × 10⁴ cells/ml for C. guilliermondii and 1.86 ± 0.92 × 10⁴ cells/ml for C. fermentati with RAW 264 cells. The killing ratio of C. fermentati was significantly higher than that of C. guilliermondii, at 24.4 ± 14.6% versus −0.6 ± 14.4%, respectively (Mann-Whitney U test, P < 0.01) (Fig. 2D).

In conclusion, the present study demonstrates that C. guilliermondii and C. fermentati are closely related, but have different microbiological and clinical characteristics. Among the C. guilliermondii complex, C. guilliermondii was highly associated with bloodstream infections, but C. fermentati was not. This may be explained at least in part by the lower resistance of C. fermentati to oxidative stress and killing by macrophages. Since no C. guilliermondii isolates were accidentally identified as C. fermentati, and all of the C. fermentati isolates were more susceptible and less pathogenic, it may not be necessary to distinguish C. fermentati from C. guilliermondii in clinical practice. However, in vitro echinocandin MICs for C. guilliermondii should be regarded with some caution, because this species was less susceptible to micafungin in vivo. A limitation of this study is the limited number of isolates analyzed, and therefore our findings need to be confirmed in future studies.