Patients’ medical records were reviewed, and demographic, epidemiological, and clinical data were collected including age, gender, underlying conditions, previous exposition to antimicrobial drugs, invasive medical procedures, presence of CVC, dates of CVC removal and culturing, dates and dosages of antifungals administered, and outcome of fungemia. Mycological surveillance cultures from pharyngeal and rectal swabs, and urine were also performed weekly in all patients. Microbiological samples were processed according to the good clinical laboratory practice, including Maki and sonication techniques for catheter tip culture.

Blood cultures were collected under aseptic conditions and processed by conventional automated system (BacT/ALERT® VIRTUO™, bioMérieux, Marcy l’Etoile, France). All Candida isolates from blood cultures and catheter tips were identified by phenotypic and biochemical characteristics, proteomic profile and DNA sequencing technology. Phenotypic colony features were evaluated after 48h of incubation in Sabouraud dextrose agar (Becton Dickinson, Baltimore, USA), CHROMagar Candida® (Difco™, Becton Dickinson) and BBL Mycosel agar (Becton Dickinson). Biochemical characteristics were analyzed using commercial tests including API ID20C (bioMérieux) and AuxaColor™ 2 (BioRad-Laboratories, Marnes-la-Coquette, France). Proteomic profiles were obtained and evaluated by VITEK MS (bioMérieux) according to manufacturer's instructions. Definitive identification was performed by sequencing the internal transcribed spacer (ITS) using the primers ITS3-ITS4 and ITS2-ITS5 previously described1; sequencing reaction was performed using GenomeLab™ GeXP (Beckman Coulter, Brea, USA) equipment and the sequences obtained were compared with those in Microbial Genomes BLAST (http://www.ncbi.nlm.nih.gov/guide/sequence-analysis/). A similarity of ≥97% was used as the criterion for species identification. ITS sequences (GenBank accession no. KJ126759, KC692045) of our isolates shared a 95–99% similarity with the ITS sequences of several C. auris strains. The similarity with the isolate from Japan (Satoh, K. 2009 GenBank accession no. AB375772)18 was 94–96%.

In vitro antifungal susceptibility testing against amphotericin B, fluconazole, itraconazole, posaconazole, voriconazole, flucytosine, caspofungin, anidulafungin, and micafungin was performed by the colorimetric microdilution panel Sensititre Yeast One®Y010 (TREK Diagnostic Systems, Cleveland, USA) according to the manufacturer's instructions.

The main clinical characteristics of the patients, the antifungal treatment administered and fungemia outcome are described in Table 1. All cases were observed between April and June 2016 in adult patients (26–66 years old) admitted to SICU, with an average length of stay before fungemia of 36 days (range, 18–59 days). All patients had been previously exposed to broad-spectrum antibiotics, parenteral nutrition and multiple invasive medical procedures including CVC, indwelling urinary catheter, and surgery, but none had neutropenia. Two patients received fluconazole before fungemia was diagnosed. Candida score was ≥4 in all patients at the onset of fungemia. Clinical management included catheter removal and prompt echinocandin therapy in all cases; in one case (Case 1) liposomal amphotericin B was administered concomitantly.

In all fungemia episodes yeast growth was detected in blood culture bottles after 7–32h incubation. In the quantitative CVC tips culture of the 4 patients a significant yeast growth was also obtained (>15CFU by the Maki technique and >100CFU by sonication). Subcultures on Sabouraud dextrose agar revealed smooth white to cream-colored colonies. All isolates grew well at 37°C, but not at 45°C or on Mycosel agar. In Chromagar Candida all isolates formed smooth pinkish, creamy colonies after 48h incubation. The identification of the isolates obtained by the different techniques are shown in Table 2. All isolates were initially identified as S. cerevisiae by AuxaColor 2 technique and as C. sake with the API ID20C yeast identification system. By means of VITEK MS technology, the blood isolate from patient 1 was identified as Candida lusitaniae and the CVC tip isolate from patient 2 as C. haemulonii; the other six isolates could not be identified by this technique. DNA identification, based on the ITS region sequences, classified all isolates as C. auris.

The real burden of C. auris infections could be underestimated since this species is currently misidentified in the clinical laboratories by commercially available identification tests based on biochemical characteristics. Consequently, it is necessary to confirm the identification by DNA sequencing of all high fluconazole-resistant isolates of species usually susceptible to this antifungal agent.

Nowadays, two MALDI-TOF systems are commercially available for routine bacterial and fungal identification in the clinical microbiology laboratories: Vitek MS (bioMérieux) and MALDI Biotyper CA System (Bruker Daltonics Inc., Billerica, MA, USA). Vitek MS combines two software platforms: Vitek MS IVD for routine diagnosis (that not includes C. auris in its database), and Vitek MS Research Use Only (RUO) whose database has recently been implemented (May 2016) with the C. auris spectra only for research purposes.12 MALDI Biotyper CA System has a database library that contains spectra of 3 strains of C. auris, two from Korea and one from Japan.13 Usually, genomics studies include C. auris within the C. haemulonii complex due to striking similarities in biochemical characteristics; this may be the reason why C. auris is misidentified by Vitek MS.4,11 Furthermore, Chatterjee et al. have recently reported the first draft genome of C. auris to explore its genomic basis of virulence.6 The authors found that C. auris has a highly divergent genome and shares genes with C. albicans and C. lusitaniae, pointing out a common ancestry. These findings could explain the frequent misidentification of C. auris as C. lusitaniae using the MALDI-TOF databases.

The eight C. auris isolates showed the same antifungal MIC, independently of the sample or the patient. Fluconazole and voriconazole MICs were ≥256mg/l and 2mg/l, respectively, and those of the other antifungal agents ranged from ≤0.06mg/l of flucytosine to 0.5mg/l of amphotericin B. Although there are no defined breakpoints for this species, based on those established for other Candida species, our isolates were highly resistant to fluconazole and resistant to voriconazole, but susceptible to the other antifungals tested. Comparing these results with those reported by other authors, it seems that C. auris is a species which is globally resistant to fluconazole and voriconazole and susceptible to other antifungals, although resistance to echinocandins and amphotericin B has also been described.2,3,7,13,16 In our cases, it seems that there is no relationship between previous fluconazole treatment and C. auris fungemia, as only two patients received fluconazole before candidemia onset.

C. auris is an emerging cause of invasive candidiasis included within the C. haemulonii complex (Group II) based on physiological characteristics and isoenzimatic profile.4C. auris fungemia is associated with a high mortality rate, therapeutic failure, and widespread resistance to different antifungal agents.7,8,13,15–18 Outbreaks of nosocomial bloodstream infections by C. auris have been reported by other authors.3,7,8,16,17 In our series, the coexistence of four cases of fungemia by C. auris in a short period of time, in the same hospitalization unit, and with identical susceptibility pattern suggests a common source for all cases; to confirm this fact, we are currently conducting phylogenetic studies with all C. auris strains isolated in the SICU since the first fungemia episode caused by this species.

In conclusion, C. auris is an emerging pathogen that is underreported because it is misidentified in routine diagnostic laboratories. Furthermore, its resistance to fluconazole and other antifungal agents could make the management of deep-seated infections caused by this species difficult.

None to declare.