During June 2016–January 2017, a total of 12 cases of fungemia occurred among neonates admitted to a 24-bed NICU of a multispecialty hospital in Delhi. Cases of BSI were defined as the isolation of D. churashimaensis from >1 peripheral blood culture in patients with signs and symptoms of sepsis. The first case of fungal sepsis was observed in a preterm infant with very low birthweight (<1,500 g) who experienced asphyxia during birth. The patient’s blood culture was positive for yeast on day 6 after birth. Fluconazole treatment was initiated by administering a loading dose of 12 mg/kg bodyweight, after which the patient received 6 mg/kg bodyweight in addition to vancomycin and meropenem. The second case occurred 2 weeks later in a preterm baby with low birthweight (<2,500 g) whose blood culture yielded D. churashimaensis on day 2. During the next 6 months, 10 additional cases of D. churashimaensis BSI were identified (Table 1).

Isolates grew as yeast-like cream to pale yellow, dry, and wrinkled colonies with fringes on the margin on Sabouraud glucose agar at 35°C after 48 h of incubation; the isolates grew slowly at 37°C over 3 days. Micromorphology showed fusiform yeast cells with hyphae and polar budding with short denticles. Identification by VITEK2 (bioMérieux, https://www.biomerieux.com) yielded Cryptococcus laurentii with 88% probability. We conducted carbon assimilation on isolates and noted assimilation of D-trehalose and N-acetyl-glucosamine at 37°C in 48 h.

Because no multilocus sequence or microsatellite typing were available, we used whole-genome sequencing (WGS) and amplified fragment-length polymorphism typing to understand the genetic relationships among isolates. We conducted matrix-assisted laser desorption/ionization-time of flight mass spectrometry by using Biotyper 3.1 (Bruker Corp., https://www.bruker.com) to identify the yeasts. However, because no database of this yeast is available, we were not able to make an identification. The in-house database created yielded correct identification in the remaining 10 isolates with high score values (>2).

We used isolates ITS and D1/D2 regions to sequence isolates, as described previously (9). We searched ITS and D1/D2 region sequences in BLAST (https://blast.ncbi.nlm.nih.gov) and identified isolates from the outbreak as D. churashimaensis. The isolates had >99% identity with D. churashimaensis sequences from GenBank (accession nos. MN758637–48 and MN158668–79).

Figure 1

Figure 1. Dendrogram of amplified fragment-length polymorphism analysis of Dirkmeia churashimaensis isolated from 12 cases of fungemia in patients in a neonatal intensive care unit, Delhi, India. The dendrogram was constructed by using...

We used MEGA 7 (https://www.megasoftware.net) to perform phylogenetic analysis of ITS sequences by using the neighbor-joining method with 2,000 bootstrap values. All isolates from the outbreak clustered together. The reference D. churashimaensis isolate formed a distinct cluster but showed 99% nucleotide similarity with isolates from the outbreak. We performed amplified fragment-length polymorphism fingerprint analysis, as described previously (10), which yielded identical banding pattern among all 12 isolates, suggesting clonal origin (Figure 1).

Figure 2

Figure 2. Whole-genome single-nucleotide polymorphism–based phylogenetic tree of 6 Dirkmeia churashimaensis isolates from cases of fungemia among patients in a neonatal intensive care unit, India. Other yeast species included for comparison. Scale bar...

We performed WGS on 6 isolates by using IonPGM (IonTorrent; ThermoFisher, https://www.thermofisher.com) next-generation sequencing technology, following the manufacturer’s protocol. We deposited sequences into BioProject (accession no. PRJEB35981). We identified average nucleotide identity and SNPs by comparing 6 genomes in MUMmer (http://mummer.sourceforge.net) and compared all the genomes against other publicly available basidiomycete yeast genomes (Table 2) by using progressiveMauve (11). We constructed a whole-genome SNP-based phylogenetic tree by using SplitsTree4 (12; Figure 2). The assembled genome size of D. churashimaensis is ≈21 Mb with a G+C content of 58%. The assembly contained 397–502 contigs ranging in length from 535,979 to 946,852 bp (average contig length 741,415 bp). D. churashimaensis isolates were genotypically indistinguishable and had 99.6% similarity among the genomes (average nucleotide identity >99.6%; Table 2). The average number of SNP differences between isolates was 1,074.5 (range 402–1,621), indicating high clonality.

We conducted antifungal susceptibility testing using microbroth dilution method published by the US Clinical Laboratory Standards Institute (13). As expected with Basidiomycota, isolates in this outbreak were resistant to echinocandins. Susceptibility testing showed that all isolates were resistant to caspofungin, anidulafungin, and micafungin (MICs >8 µg/mL). However, all isolates had low MICs for azoles, including voriconazole (MIC 0.03–0.125 µg/mL; geometric mean [GM] 0.04 µg/mL), isavuconazole (MIC 0.03–0.125 µg/mL; GM 0.05 µg/mL), itraconazole (MIC 0.03–0.25 µg/mL; GM 0.057 µg/mL), posaconazole (MIC 0.03–0.25 µg/mL; GM 0.092 µg/mL), and fluconazole (MIC 1–4 µg/mL; GM 2.37 µg/mL). Amphotericin B (GM MIC 0.198 µg/mL) and 5-flucytosine (GM MIC0.157 µg/mL) had potent activity.

All patients were treated with fluconazole at a loading dose of 12 mg/kg bodyweight and then 6 mg/kg for 10–14 days; 5 patients died, a case-fatality rate of 42%. All patients had risk factors, such as preterm birth or low or very low birthweight, and 8/12 were intubated (Table 1). The most serious risk factors were central venous catheter (n = 9), thrombocytopenia (n = 8), and severe asphyxia (n = 6). The age at the onset of fungemia ranged from 2 to 6 days, and the attack rate was 0.33 during the 6-month outbreak. The mean gestational age was 29.2 weeks and the mean birthweight was 1.1 kg. Altogether, 11 patients had 1–6 days of antimicrobial drug therapy before isolation of yeast in blood culture.

After the second case of fungemia was identified, infection control measures were implemented and surveillance cultures obtained to trace the source of infection. Doctors, nursing staff, and assistants in the NICU were screened for hand carriage of the yeast, and extensive sampling of fomites including floors, equipment, disinfectants, vials, and infusion pumps was conducted. All environmental cultures were negative, and no other cases of fungemia due to D. churashimaensis were identified after continued compliance with infection control measures, including rigorous handwashing practice.

Our report highlights not only clinical importance of rare yeast species in the NICU but also emphasizes that WGS provides a highly sensitive tool for genotyping pathogens without prior knowledge of the genomes (1,14,15). Major healthcare-associated outbreaks of uncommon and novel fungal species have occurred in recent years, including Candida auris from a clonal outbreak in India (9). Mycologists should be vigilant when they isolate unusual or rare yeasts with potential antifungal resistance. Considerable challenges remain in the diagnosis of rare and unusual yeasts and the risk for misidentification of cases and outbreaks is high. Our findings reinforce the need for awareness of this new fungal risk among the public health community.