Candida auris Pan-Drug-Resistant to Four Classes of Antifungal Agents

ABSTRACT

Candida auris is an urgent antimicrobial resistance threat due to its global emergence, high mortality, and persistent transmissions. Nearly half of C. auris clinical and surveillance cases in the United States are from the New York and New Jersey Metropolitan area. We performed genome, and drug-resistance analysis of C. auris isolates from a patient who underwent multi-visceral transplantation. Whole-genome comparisons of 19 isolates, collected over 72 days, revealed closed similarity (Average Nucleotide Identity > 0.9996; Aligned Percentage > 0.9764) and a distinct subcluster of NY C. auris South Asia Clade I. All isolates had azole-linked resistance in ERG11(K143R) and CDR1(V704L). Echinocandin resistance first appeared with FKS1(S639Y) mutation and then a unique FKS1(F635C) mutation. Flucytosine-resistant isolates had mutations in FCY1, FUR1, and ADE17. Two pan-drug-resistant C. auris isolates had uracil phosphoribosyltransferase deletion (FUR1[1Δ33]) and the elimination of FUR1 expression, confirmed by a qPCR test developed in this study. Besides ERG11 mutations, four amphotericin B-resistant isolates showed no distinct nonsynonymous variants suggesting unknown genetic elements driving the resistance. Pan-drug-resistant C. auris isolates were not susceptible to two-drug antifungal combinations tested by checkerboard, Etest, and time-kill methods. The fungal population pattern, discerned from SNP phylogenetic analysis, was consistent with in-hospital or inpatient evolution of C. auris isolates circulating locally and not indicative of a recent introduction from elsewhere. The emergence of pan-drug-resistance to four major classes of antifungals in C. auris is alarming. Patients at high risk for drug-resistant C. auris might require novel therapeutic strategies and targeted pre-and/or posttransplant surveillance.

INTRODUCTION

Candida auris is among five urgent threats in the CDC’s 2019 Antibiotic Resistance Threats Report due to its global emergence, multidrug resistance, high mortality, and persistent transmissions in health care settings (1–5). The available evidence suggests either C. auris emerged in East Asia or simultaneously in parts of Asia, Africa, Europe, and South and North America (6, 7). The first US cases of C. auris were recognized in 2013, and the disease was identified in New York in 2016 (8, 9). Since 2018, C. auris has been a nationally notifiable pathogen, documented in 24 US states, and nearly 55% of the clinical and surveillance cases are from the New York and New Jersey Metropolitan area (9–11).

Antifungal resistance is widespread in C. auris South Asia Clade I isolates. These isolates are overwhelmingly resistant to fluconazole, are variably resistant to amphotericin B, and acquire resistance to echinocandins (11–14). Candida auris South America Clade IV includes isolates with variable resistance to amphotericin B (15, 16), while South Africa Clade III isolates are commonly resistant to azoles (17). Candida auris isolates resistant to three major classes of antifungal agents have emerged in the US and other countries (5, 18, 19). A recent report highlighted health care transmission of C. auris resistant to three major classes of antifungal drugs (20).

We performed genome, and drug-resistance analysis of C. auris isolates from a multi-visceral transplant recipient with refractory fungal peritonitis. Two C. auris isolates tested resistant to four major classes of antifungal agents (azoles, echinocandins, polyene, and flucytosine) while several were resistant to three major drug classes. Among drug targets, gene mutations in ERG11, CDR1, FKS1, and FCY1 were prominent. Two pan-drug-resistant C. auris isolates carried a novel deletion in uracil phosphoribosyltransferase (FUR1[1Δ33]) and subsequent elimination of FUR1 expression associated with high flucytosine resistance.

RESULTS

C. auris antifungal resistance.

Nineteen C. auris clinical isolates originated most often from intra-abdominal samples at infrequent intervals (Table 1, Fig. 1). The antifungal susceptibility-resistance pattern was confirmed by testing with two different microbroth antifungal panels and Etest (Table 1). The antifungal susceptibility test results were interpreted as per CDC C. auris tentative MIC breakpoints (10). Second-generation azole results were not-interpretable as tentative breakpoints are not available. The first isolate (20–34, hospital day [HD] 4) and a single blood isolate (20–22, HD33) were susceptible to all antifungals tested except for fluconazole-resistance. Five C. auris isolates (HD4-33) were susceptible to echinocandins, while the remaining 14 isolates (HD33-72) were variably resistant. Amphotericin B MIC values were high (0.75 to 1.0 mg/L) in 15 C. auris isolates, while four were amphotericin B resistant (2.0 mg/L). There is no CDC advisory for flucytosine breakpoints, and therefore, we considered MIC ≥ 8.0 mg/L in seven C. auris isolates as indicative of resistance akin to other Candida species (21–24). Candida auris 20–31 and 20–32 (HD72) were pan-drug-resistant to azoles, amphotericin B, echinocandins, and flucytosine.

TABLE 1.

In vitro antifungal susceptibility of Candida auris from patient^a,b

C. auris	Source	Days hospitalization	CDC custom microbroth panel for C. auris								E-test		Trek diagnostic Y09 panel									Custom panel
			FLC	VRC	ITC	ISA	POS	AFG	CAS	MFG	AMB	5FC	FLC	VRC	ITC	POS	AFG	CAS	MFG	AMB	5FC	IBX
20–34	Intra-abdominal fluid	4	>256	2	1	2	0.50	0.25	0.25	0.12	1	0.094	>256	2	0.5	0.25	0.25	0.25	0.12	2	0.12	0.25
20–20	Aorta tissue	24	>256	2	1	1	0.5	4	2	4	1	0.12	>256	1	0.25	0.12	2	2	8	2	0.06	0.25
20–21	Aorta tissue	24	>256	2	1	2	0.5	0.5	0.25	0.25	2	0.12	>256	1	0.25	0.12	0.12	0.12	0.12	4	0.06	0.25
20–15	Urine	30	>256	2	0.5	0.5	0.12	4	2	4	0.75	0.06	>256	1	0.25	0.12	2	2	8	2	<0.06	1
20–16	Intra-abdominal abscess	30	>256	1	0.5	0.25	0.12	4	2	4	0.75	0.06	>256	0.5	0.12	0.06	2	2	8	4	<0.06	2
20–17	Intra-abdominal abscess	30	>256	2	0.5	0.5	0.12	2	0.25	0.25	1	0.12	>256	1	0.25	0.12	1	0.5	0.5	4	0.06	0.25
20–18	Intra-abdominal abscess	30	>256	2	0.5	0.5	0.12	4	2	4	0.75	0.12	>256	1	0.25	0.06	2	2	8	2	<0.06	2
20–19	Intra-abdominal abscess	30	>256	2	1	1	0.5	4	2	4	1	0.12	>256	0.5	0.12	0.06	2	2	8	2	<0.06	2
20–22	Blood	33	>256	2	1	2	0.5	0.5	0.25	0.25	1	0.12	>256	1	0.25	0.12	0.12	0.12	0.12	2	0.06	0.5
20–23	Intra-abdominal tissue exudate	33	>256	2	1	1	0.25	0.5	0.25	0.12	2	0.12	>256	1	0.25	0.12	0.12	0.12	0.12	4	<0.06	0.5
20–24	Intra-abdominal tissue exudate	33	>256	2	1	1	0.5	4	2	4	1	0.12	>256	1	0.25	0.12	4	2	8	4	0.06	0.5
20–25	Intra-abdominal tissue exudate	33	>256	2	1	1	0.5	4	2	4	1	0.12	>256	0.5	0.12	0.06	2	2	8	2	<0.06	0.25
20–27	Peritoneal fluid	54	>256	2	2	2	1	4	2	4	1	8	>256	1	0.25	0.12	2	2	8	4	>64	2
20–26	Intra-abdominal tissue	57	>256	2	1	2	1	4	2	4	0.75	8	>256	1	0.12	0.06	2	2	8	4	>64	2
20–28	Intra-abdominal tissue	62	>256	8	1	2	0.5	4	>16	8	1	8	>256	1	0.12	0.06	2	2	8	2	>64	2
20–29	Intra-abdominal tissue	62	>256	2	1	2	0.5	4	2	4	1	8	>256	1	0.25	0.12	2	2	8	4	>64	2
20–30	Intra-abdominal fluid	72	>256	8	1	4	0.5	4	16	8	1	8	>256	8	0.5	0.25	4	4	>8	4	>64	2
20–31	Intra-abdominal fluid	72	>256	16	1	4	0.5	4	16	8	2	8	>256	8	0.5	0.25	4	8	>8	4	>64	2
20–32	Intra-abdominal tissue	72	>256	8	1	4	1	4	16	8	2	8	>256	8	0.5	0.25	4	4	>8	4	>64	1
		median	>256	2	1	2	0.5	4	2	4	1	0.12	>256	1	0.25	0.12	2	2	8	4	0.06	1.5
		range	>256	1–16	0.5–2	0.25–4	0.12–1	0.25–4	0.25–>16	0.12–8	0.75–2	0.06–8	>256	0.5–8	0.12−0.5	0.06–0.25	0.12–4	0.12–8	0.12–>8	2–4	<0.06–>64	0.25–2

^aValues shown are MIC (mg/L).

^bFLC, fluconazole; VRC, voriconazole; ITC, itraconazole; ISA, isavuconazole; POS, posaconazole; AFG, anidulafungin; CAS, caspofungin; MFG, micafungin; AMB, amphotericin B; 5FC, flucytosine; IBX, ibrexafungerp.

FIG 1.

A timeline of the clinical course of the patient. Candida auris isolations at different intervals, including pan-drug-resistant isolates recovered on hospital day 72 are shown. Also highlighted are antimicrobial drugs and duration, major complications, and other pathogens encountered in the patient.

C. auris population.

Average nucleotide identity (ANI) and aligned percentage (AP), derived from genome analysis, revealed closed similarity (ANI > 0.9996; AP > 0.9764). However, the first isolate 20–34 (HD4) showed the highest ANI (0.9997) and AP (0.9847) compared to the core genome of reference C. auris B8441 South Asia Clade I, which originated from Pakistan (5). Conversely, the late-stage pan-drug-resistant isolates 20–31, 20–32 (HD72) were among the least similar to 20–34, perhaps pointing to the rapid changes undergoing the population of C. auris during the infection. A detailed comparison of assembled genomes of 19 C. auris isolates is presented in Table S1.

An SNP phylogenetic tree from the assembled genomes revealed 19 C. auris isolates belonged to South Asia Clade I relative to other representative genomes for clades I-IV (Fig. 2). In concurrence with earlier studies, the New York C. auris isolates were a distinct population within South Asia Clade I (5, 18). Furthermore, C. auris patient isolates in this study grouped as a distinct subcluster among NY isolates with strong bootstrap support (>99%).

FIG 2.

Candida auris population analysis. SNPs scored from whole-genome assemblies were analyzed by the Maximum Likelihood (ML) method to show the placement of patient C. auris isolates vis-à-vis representative C. auris South Asia Clade I isolates and three other clades. All C. auris patient isolates are in a distinct subcluster among NY isolates.

Mutations in drug-resistance genes.

The mutations in major antifungal drug-resistance genes are summarized in Table 2. All C. auris isolates carried azole-linked resistance in ERG11(K143R) and CDR1(V704L). The echinocandin resistance appeared with FKS1(S639Y) mutation, while subsequent isolates had unique FKS1(F635C) mutations. The flucytosine-resistant isolates had mutations in FCY1, FUR1, and ADE17, and FUR1 deletion (1Δ33) as summarized in Table 3.

TABLE 2.

C. auris antifungal resistance gene mutations^a

C. auris	Collection day	Gene mutations
		ERG11	FKS1	FKS2	CDR1	CcrB	FUR1	FCY1	FCY2	ADE17
20–34	4	K143R	ref	ref	V704L	ref	ref	S70R	ref	ref
20–20	24	K143R	S639Y	ref	V704L	ref	ref	S70R	ref	ref
20–21	24	K143R	ref	ref	V704L	ref	ref	S70R	ref	ref
20–15	30	K143R	F635C	ref	V704L	ref	ref	S70R	ref	ref
20–16	30	K143R	F635C	ref	V704L	ref	ref	S70R	ref	ref
20–17	30	K143R	D642Y	ref	V704L	ref	ref	S70R	ref	ref
20–18	30	K143R	F635C	ref	V704L	ref	ref	S70R	ref	ref
20–19	30	K143R	F635C	ref	V704L	ref	ref	S70R	ref	ref
20–22	33	K143R	ref	ref	V704L	ref	ref	S70R	ref	ref
20–23	33	K143R	ref	ref	V704L	ref	ref	S70R	ref	ref
20–24	33	K143R	S639Y	ref	V704L	ref	ref	S70R	ref	ref
20–25	33	K143R	F635C	ref	V704L	ref	ref	S70R	ref	ref
20–27	54	K143R	F635C	ref	V704L	V119L	ref	S70R	M128fs	ref
20–26	57	K143R	F635C	ref	V704L	ref	1Δ33	S70R	ref	G45V
20–28	62	K143R	F635C	ref	V704L	ref	S84L	S70R	ref	ref
20–29	62	K143R	F635C	ref	V704L	V119L	ref	S70R	M128fs	ref
20–30	72	K143R	F635C	ref	V704L	ref	1Δ33	S70R	ref	G45V
20–31	72	K143R	F635C	ref	V704L	ref	1Δ33	S70R	ref	G45V
20–32	72	K143R	F635C	ref	V704L	ref	1Δ33	S70R	ref	G45V
19–4	NA	K143R	S639Y	ref	V704L	ref	ref	S70R	ref	ref
19–42	NA	K143R	S639P	ref	V704L	ref	ref	S70R	ref	ref
19–43	NA	K143R	S639P	ref	V704L	ref	ref	S70R	ref	ref
19–61	NA	K143R	S639Y	ref	V704L	ref	ref	S70R	ref	ref
B8441b	NA	ref	ref	ref	ref	ref	ref	ref	ref	ref
6684	NA	Y132F	ref	ref	ref	ref	ref	S70R	ref	ref

^aNA, not applicable; fs, frameshift; Ref, reference.

^bC. auris B8441 reported earlier from Pakistan (5).

TABLE 3.

Flucytosine MIC for select C. auris isolates and corresponding mutations^a

C. auris isolate	Flucytosine MIC (mg/L)	Mutations in key target genes
		ADE17	FCY1	FCY2	FUR1
20–34 (Initial isolation)	<0.5	None	None	None	None
20–26	>512	G45V	S70R	None	1Δ33
20–27	256	None	S70R	M128fs	None
20–28	>512	None	S70R	None	S84L
20–29	256	None	S70R	M128fs	None
20–30	>512	G45V	S70R	None	1Δ33
20–31	>512	G45V	S70R	None	1Δ33
20–32 (Last isolation)	>512	G45V	S70R	None	1Δ33
C. parapsilosis ATCC 22019	0.12	NA	NAb	NA	NA
C. krusei ATCC 6258	4.0	NA	NA	NA	NA

^aMIC visually read at 50% inhibition using a custom-made microbroth dilution plate with high concentrations of flucytosine.

^bNA, not applicable.

C. auris FUR1 deletion test.

The primers and probes, anchored in the conserved part of FUR1 and outside the deleted region, yielded amplicons exemplified by low Ct values in all C. auris isolates (Fig. 3). A similar strategy to amplify only the deleted region (1Δ33), showed no amplicons only in four C. auris isolates (20–26, 20–30, 20–31, and 20–32), confirming in silico findings of deletion in FUR1 (Fig. 3). Finally, no transcript was present in four C. auris Isolates with deleted FUR1.

FIG 3.

Deletion in C. auris FUR1. A set of primers/probe anchored to amplify full-length FUR1 using real-time PCR test with positive Ct value (average from two readings). Another set of primers/probe, designed to only amplify deleted portion of FUR1, did not yield positive Ct values in pan-drug-resistant isolates of C. auris consistent with in silico prediction of a truncated FUR1. Finally, no transcript was present in four C. auris Isolates with deleted FUR1. *, a select number of isolates tested with no deletion in FUR1; **, all four isolates with FUR1 deletion.

Antifungal combinations.

In checkerboard microbroth assays, all four drug combinations resulted in lower MICs relative to drugs alone for the first C. auris isolate 20–34 (Table S2). Synergy was observed with caspofungin and amphotericin B combination for the first C. auris isolate 20–34. This synergy is a result of the elimination of the paradoxical growth associated with caspofungin. Indifference was observed for all other combinations tested. All drug combinations were ineffective against both pan-drug-resistant isolates C. auris 20–31 and 20–32 (Table S2).

The Etest strip overlays revealed that the addition of flucytosine to amphotericin B led to lower MICs (additive) relative to single drugs for the first C. auris isolate 20–34, but not for pan-drug-resistant C. auris 20–31 and 20–32 (Table S3). Echinocandin-amphotericin B antifungal combinations were either additive or synergistic for the first C. auris isolate 20–34, but mostly indifferent for pan-drug-resistant C. auris 20–31 and 20–32 compared to drugs alone (Table S3).

The fixed concentrations of two-drug combinations were also studied in time-kill assays and evaluated for fungicidal or fungistatic activity at concentrations known to be achievable in serum. All single drugs except amphotericin B and one of the four two-drug combinations displayed fungicidal activity against the first C. auris isolate 20–34 (Fig. 4, Table S4). The combination of micafungin and amphotericin B, which required at most 0.5 mg/L of each drug to inhibit growth of the first C. auris isolate 20–34 in both Etest and checkerboard combinations, displayed fungicidal activity in the time-kill assay. All drugs alone and in two-drug combinations had no significant effect on the pan-drug-resistant C. auris 20–32 as colony forming units (CFUs) were similar to the growth control (Fig. 4, Table S4).

FIG 4.

Time-kill assays for two-drug combinations against C. auris. Fixed concentration of two drug combinations were prepared and inoculated with 10³ CFU of echinocandins-susceptible (20–34) and pan-drug-resistant (20–32) C. auris isolates. Plates were incubated at 35°C for 24 h followed by enumeration of CFU. Significant killing of C. auris 20–34 with micafungin combination was observed (A) while other combinations were not that significant (B–D). No significant killing of C. auris 20–32 observed against any of the drug combinations tested (E–H).

DISCUSSION

The risks for fatal fungal infections remain high in solid organ transplant recipients despite antifungal prophylaxis. Candida auris is a relatively new and severe addition to the list of health care-associated fungal pathogens (9, 25–28). Intrinsic fluconazole-resistance and high amphotericin B MICs are common in NY C. auris isolates (11, 29). Therefore, the initial choice of caspofungin prophylaxis appeared prescient as the first C. auris isolate from the patient was caspofungin-sensitive in retrospective laboratory testing. The accumulation of echinocandin and flucytosine resistance mutations in subsequent isolates, in the setting of prolonged antifungal drug exposure and refractory peritonitis due to intestinal anastomotic leaks, likely led to pan-drug-resistant C. auris. In the published literature, there are no instances of C. auris resistance to azoles, echinocandins, amphotericin B, and flucytosine. The clinical consequences of pan-drug-resistant C. auris require further investigation, considering the complications from polymicrobial sepsis and liver failure in the present case. Nevertheless, the emergence of pan-drug-resistant C. auris documented in this study foretell unprecedented risks of infection and transmission in hospital and health care facilities (3, 20, 30).

Ongoing surveillance from NY-NJ metropolitan area supports a sustained C. auris epidemic, one of the largest recorded to date (9, 10, 31). The New York C. auris isolates are a distinct population group in South Asia Clade I, with high drug-resistant isolates constituting a subcluster (2, 18). In the current case, the pan-drug-resistant C. auris isolates are also closely related within the NY population and form a subcluster distinct from other drug-resistant isolates reported previously (18). The fungal population pattern discerned in our patient is consistent with in-hospital or inpatient evolution of C. auris isolates circulating locally and not indicative of a recent introduction from elsewhere (2, 3, 7).

Candida auris flucytosine resistance in the setting of multidrug resistance, is a novel finding in our patient. The high MIC values observed in C. auris isolates were similar to those reported from flucytosine-resistant Candida species in prior publications (21, 23, 32). While FUR11Δ33 was a novel finding, mutations in FCY1, FUR1, and ADE17 are well-known.

We found nonsynonymous mutations in the predicted sequences of purine cytosine-deaminases FCY1 and FCY2 (36). All C. auris isolates from the patient had FCY1 S70R mutation, which appeared inconsequential for flucytosine resistance. A M128 frameshift mutation in FCY2 was associated with resistance in two isolates collected after the initiation of flucytosine therapy. The uracil phosphoribosyltransferase FUR1 with a deletion in the open reading frame (1Δ33) and loss of the corresponding transcript possibly led to high flucytosine resistance in four C. auris, including pan-drug-resistant isolates. FUR1(1Δ33) deletion paralleled G45V mutation in ADE17, encoding 5-aminoimidazole-4-carboxamide ribonucleotide transformylase in the purine biosynthesis pathway (33). Our mutation-phenotype correlations for flucytosine are indirect and corroborative with other flucytosine-resistant Candida species (21–23). Flucytosine resistance was associated with following mutations: ADE17:G45V + FUR1:1Δ33, and CrcB:V119L + FCY2:M128fs. The mutations in ADE17, CrcB and FCY2 and deletion in FUR1 found in our isolates have not been reported in earlier publications, but involved genes considered critical for flucytosine resistance (21). It is possible, however, that other nonsynonymous mutations in C. auris isolates from the patient are contributing, or potentially even driving, resistance to flucytosine. Future studies with directed mutagenesis experiments are required to confirm our observations (32, 34).

Candida auris echinocandin-resistance is linked to FKS1 hot spot mutations, with S639Y being a prominent feature of Clade I South Asian isolates (12, 35, 36). In an earlier report, we found multidrug-resistant NY C. auris Clade I isolates carried FKS1 S635P and S635Y mutations instead (18). FKS1 hot spot mutations in the current case were heterogeneous, with both S639Y and F635C appearing by HD30. There was a single intermittent instance of FKS1 D642Y mutation in an echinocandin-sensitive isolate. Eventually, F635C became dominant post-HD33, including in pan-drug-resistant isolates with very high MICs for caspofungin and anidulafungin. FKS2, however, remained unchanged in all patient isolates, and likely, played no role in echinocandin resistance during the course of infection. The relative significance of three different FKS1 hot spot mutations remains unknown and await directed mutagenesis and modeling studies with C. auris Clade I. Resistance to a wide range of echinocandins was also observed for the majority of the isolates. A Fisher Exact test was used to identify potential mutations contributing to resistance, using sensitive isolates as controls and resistant isolates as case samples. Initially, a total of 16 variants were identified as potentially contributing to echinocandin resistance. Six were associated with homopolymer regions, and one with repeat regions. Only a single, nonsynonymous mutation in FKS1 (F635C) was statistically significant (P = 0.00361) and enriched in 12 of the 15 echinocandin resistant isolates. Further investigation of this gene, previously found to be associated with echinocandin resistance, revealed two additional mutations: S639Y (2 isolates) and D642Y (1 isolate) (17, 37).

Fluconazole-resistance and ERG11 K143R mutation were common in all C. auris isolates, consistent with earlier reports (12, 18). We speculate this mutation also caused cross-resistance to other azoles, especially voriconazole with very high MICs in pan-drug-resistant isolates. The 19 isolates reliably shared 1,437 common variants compared to B8441 reference genome, 387 of which were within or immediately adjacent to homopolymer regions, 18 represented insertions or deletions in repeat regions, and 493 represented mutations resulting in nonsynonymous changes to 452 protein coding genes (38). In addition to these shared variants, the entire collection of 19 isolates collectively had an additional 1,116 variants shared among one or more of them. Given the extensive number of differences from the reference genome for Clade I (B8441), there are potentially many mutations that could lead the resistance phenotype observed for fluconazole, and the various levels of resistance to the other azoles tested. That being said, all 19 C. auris isolates had the following mutations in genes previously identified as potentially related to azole resistance: CDR1V704L, CDR2P551A, CIS1K74Q, ERG4M192I, ERG11K143R, FCY1S70R, TAC1bA640V, URB2L641F, URK1S472V, and YEN1A294V. A follow-up genome analysis, including a larger collection of sensitive Clade I C. auris isolates as controls, would enable the identification of mutations enriched for azole resistance.

ERG11 mutations are reported to be associated with cross-resistance to amphotericin B in other Candida species (39). We used two methods to test amphotericin B resistance and found values were within one dilution of the suggested MIC breakpoints. We did not find ERG6 mutation reported earlier as crucial for C. auris amphotericin B-resistance (40). No other statistically significant nonsynonymous variants were identified as being shared among amphotericin-susceptible and -resistant isolates. This suggests that potentially multiple mutations could lead to a common phenotype of amphotericin B resistance, noncoding mutations are contributing to resistance, or as yet unknown mechanism is driving amphotericin B resistance. Using the same methodologies described above, no shared variants were identified as being enriched among the four amphotericin B-resistant isolates. Thus, it is possible, that amphotericin B resistance may have developed independently. Further genome and directed mutagenesis analyses would be required to confirm the genetic causes responsible for the observed amphotericin B-resistance in C. auris.

We and others have highlighted the potential of antifungal combination therapy in patients with multi-drug and pan-drug resistant C. auris (18, 20, 41, 42). Our limited testing in the current study provides a new evaluation of the role of antifungal combinations against C. auris, especially the incredibly ineffective flucytosine-amphotericin B combination. The failure of amphotericin-echinocandin combinations against pan-drug-resistant C. auris also portends a poor outcome with two-drug combinations. We recognize the need for additional studies, including animal models and PK/PD simulations, to further ascertain if two-drug combinations will be effective in patients infected with pan-drug-resistant C. auris.

Ultimately, we pursued treatment with the investigational agent ibrexafungerp, the first compound in the novel class of triterpenoid antifungals and a promising therapeutic for C. auris infections. Similar to the echinocandins, ibrexafungerp disrupts fungal cell wall synthesis through inhibition of (1→3)-β-d-glucan synthase with fungicidal activity against Candida spp. However, ibrexafungerp is orally bioavailable and structurally distinct as a semisynthetic derivative of the naturally occurring enfumafungin, thus interacting differently with the target enzyme (43). Notably, ibrexafungerp retains activity against echinocandin-resistant strains of Candida with fks mutations and recently demonstrated in vitro activity (MIC range 0.12 to 1 mg/mL) against C. auris isolates with resistance to echinocandins, azoles, and amphotericin B (43–45). A phase 3 open-label clinical trial of ibrexafungerp for treatment of invasive candidiasis due to C. auris is ongoing (CARES; NCT03363841).

In addition to the elucidation of resistance mechanisms and therapeutic strategies, the present case provides an opportunity to consider C. auris screening strategies in organ transplantation. Given the timing of C. auris detection immediately after transplant, our patient’s extensive history of hospitalizations in the NY and NJ area, and prior antifungal drug exposure, we think it most likely that she was colonized with C. auris at admission for transplant. To date, published reports of C. auris infection in transplant patients are rare and include a lung transplant recipient with donor-derived infection and an outbreak involving four liver transplant recipients in a surgical ICU (46, 47). To our knowledge, no studies have evaluated risk factors for C. auris colonization or infection among transplant recipients. In the nontransplant population, factors associated with C. auris colonization or infection include ICU admission, prior antifungal therapy, underlying respiratory illness, mechanical ventilation, central venous catheters, and gastrostomy tubes (9, 48). We suggest that a better understanding of the epidemiology of C. auris in solid organ transplantation is needed to guide a potential pre- and/or posttransplant surveillance strategy. However, for those patients with the above-mentioned risk factors and significant health care exposure in geographical areas with high rates of C. auris, clinicians should maintain a high index of clinical suspicion for this ominous pathogen.

Overall, this study offers several lessons for C. auris surveillance and antifungal therapy in high-risk patients. We believe retrospective susceptibility testing and genome surveillance have limited value in managing complicated C. auris cases (49, 50). The conventional phenotypic methods are too slow for real-time monitoring of resistance in C. auris. The laboratory framework for susceptibility testing with new drugs, including ones eligible for compassionate use, remains nonexistent. There is a need to institute rapid genomic screens for drug-resistant markers as a corollary to susceptibility testing. Finally, new drugs and combinations directed against C. auris and other drug-resistant pathogens must remain a high priority.

MATERIALS AND METHODS

The case.

A 29-year-old woman from New Jersey with extreme short bowel syndrome due to prothrombin gene mutation and superior mesenteric arterial thrombosis was admitted from home to a New York hospital for multi-visceral transplantation of the liver, pancreas, small bowel, and colon in August 2020 with antibacterial and caspofungin prophylaxis. Neither donor nor recipient had a known history of C. auris infection before transplantation.

The transplant surgery was complicated by hemodynamic lability and graft swelling such that gastrointestinal continuity was deferred until HD4. Intra-operatively, there was no gross evidence of infection in the abdominal cavity, but a swab of the intra-abdominal fluid was sent for routine culture and grew Candida auris (Fig. 1). This first and all subsequent C. auris isolates were sent to the Mycology Laboratory, Wadsworth Center, New York State Department of Health for confirmation and antifungal susceptibility testing.

The posttransplant course was further complicated by elevated liver function tests (LFTs), bleeding from the duodenal anastomosis, and recurrent intestinal anastomotic leaks requiring multiple surgeries for intestinal revision and repair between HD18-33. Cultures of intra-abdominal fluid and tissue continued to grow C. auris, and on HD30, C. glabrata, a then unidentified mold, and Bacteroides ovatus. The abdominal cavity was irrigated twice with amphotericin B solution. On HD33, blood cultures grew C. auris that cleared when cultures were repeated 4 days later. Antifungal therapy was changed from caspofungin to high-dose micafungin (150 mg IV daily) and liposomal amphotericin B (LAMB), then to voriconazole and flucytosine on HD40 when the mold speciated to Aspergillus terreus. A serum Aspergillus galactomannan index was >8.22 (normal < 0.5). On HD45, the investigational antifungal agent ibrexafungerp ([1→3]-β-d-glucan synthase inhibitor) was initiated via compassionate use protocol and administered via nasogastric tube. Voriconazole and flucytosine were continued with therapeutic drug monitoring.

Serial intra-abdominal surgical washouts were performed at the bedside every two to 4 days. The abdomen was left open with an aseptic occlusive dressing due to ongoing sepsis and challenging wound closure. Over the subsequent 2 to 3 weeks, the patient improved clinically: fevers resolved, the peripheral WBC count normalized, LFTs down trended, and she was weaned off vasopressors. However, cultures of peritoneal fluid continued to grow C. auris and A. terreus and intermittently grew Staphylococcus epidermidis.

On HD63, the patient was diagnosed with an acute cellular rejection of the transplanted small bowel and colon. Anti-thymocyte globulin and methylprednisolone were administered. Progressive leukopenia and thrombocytopenia ensued (nadir WBC count 0.2 × 10^3/μL), and flucytosine was discontinued. On HD70-71, the patient developed massive gastrointestinal bleeding, hemorrhagic and septic shock, and acute liver failure. She passed away on HD74.

C. auris characterization, antifungal susceptibility, and combination.

C. auris surveillance and clinical samples were tested using selective culture media, fungal isolates were identified by mass spectrometry (MALDI-TOF-MS), and C. auris Clades confirmed by Sanger sequencing of ITS and D1/D2 ribosomal genes (9, 11). Antifungal susceptibility testing used a custom 96-well microbroth test panel manufactured according to CLSI M60A (11, 51). Y09 microbroth plates from Trek Diagnostics were used as needed. The breakpoints for fluconazole, echinocandins, and amphotericin B were interpreted per CDC recommendations (10). Amphotericin B and flucytosine susceptibility testing also used Etest strips (BioMereiux, St. Louis, MO, USA) (11). Ibrexafungerp was tested with a custom-panel created in our laboratory (44). Flucytosine custom 96-well microbroth plates were prepared containing 0.03 to 512 mg/L drug dilutions and tested per published method (32).

Testing of amphotericin B in combination with echinocandins or flucytosine was done using microbroth checkerboard, Etest, and time-kill methods. No combinations, including azoles were tested as C. auris clade I is intrinsically resistant to fluconazole. Checkerboard plates for drug combinations were designed (Fig. S1) as per CLSI M27-A3 document (52–54). The 100% growth inhibition was recorded after 48 h of incubation at 35°C for all drugs alone and in combination and used to calculate the fractional inhibitory concentration index (FICI). In cases where the MIC was >16 or >32 mg/L, 2× the highest concentration (32 and 64 mg/L, respectively) was used as the drug alone MIC in the calculation. Drug interactions were defined as synergistic if FICI was < 0.50, indifferent if FICI ≥0.50 and ≤ 4.00, and antagonistic if FICI was >4.0. Antifungal combination testing used Etest strips as described in earlier publications (55, 56). Notably, we selected Etest strips for combination testing as they are also used in routine for amphotericin B and flucytosine testing. After inoculation, Etest strip for drug one (amphotericin B) was incubated on the agar surface for 1 h at room temperature. The first Etest strip was removed and replaced with the second Etest strip (flucytosine or echinocandins). Plates were incubated for 24 h at 35°C. For analysis of single drugs, E-tests were read according to the manufacturer’s instructions and at 100% inhibition to compare with tests for combinations (54). Drug interactions were defined as indifferent if the change in MIC in combination was <2 dilutions, additive for a decrease of ≥2 and <3 dilutions, synergistic if there was ≥3 dilution decrease, and antagonistic if an increase in MIC of ≥3 dilutions following a previous method (55). Time kill assay for four antifungal drug combinations were performed according to a published format using a 96-well plate (18). A fixed concentration 0.5 mg/L was used for each drug in the combination. The inoculum, incubation, and visual reading were similar to microbroth method for single drugs. After 12- and 24- h, CFUs were recorded on Sabouraud dextrose agar. Fungicidal activity was defined as ≥3 log₁₀ (>99.9%) decrease in CFU counts compared to the starting inoculum (57); the fungistatic activity was defined as less than a 200% increase in the starting inoculum (58).

Whole-genome sequencing.

Candida auris DNA was extracted using a commercial kit (Qiagen QIAamp DNA minikit), and genomic DNA content measured with NanoDrop 2000/2000c Spectrophotometers (Thermo Scientific, Waltham, MA) and Qubit (Thermo Scientific, Waltham, MA). The sequencing libraries were constructed using Illumina Nextera Flex Library Preparation kit (59). The 2 × 250 bp read length sequencing was done with Illumina MiSeq sequencer to obtain approximately 100× or higher genome coverage. Draft de novo genome assemblies of C. auris isolates were constructed using CLC Genomics Workbench v20 (18). In addition, de novo assemblies were constructed for C. auris isolates if raw sequencing data were available in NCBI’s Sequence Read Archive (SRA), but there was no draft assembly already available (B12679, B12079, B12447, B12045, and NG-19339). Raw sequencing data were deposited to SRA as part of NCBI BioProject PRJNA640677. Candida auris genomes were used to score single nucleotide polymorphisms (SNPs) and maximum likelihood (ML) phylogenetic tree was constructed with the inclusion of representative genomes from various C. auris clades as described in earlier publications (5, 18).

Antifungal resistance gene targets.

NCBI protein databases were searched for genes associated with resistance to each antifungal drug as detailed earlier (18). We also compared antifungal gene targets with the earlier published reports on drug resistance (5, 12, 14, 21, 22, 35, 38, 60). Collectively, our analysis included a wide-range of previously identified or suspected resistance gene targets, including Ade17, Cdr1, Cdr2, Cis2, CrcB, Erg1, Erg10, Erg11, Erg12, Erg13, Erg2, Erg20, Erg24, Erg251, Erg26, Erg27, Erg28, Erg3, Erg4, Erg5, Erg6, Erg7, Erg8, Erg9, Fcy1, Fcy2, Fen1a, Fks1, Fks2, Flu1, Fur1, Mdr1, Mec3, Nag3, Tac1b, Tpo2, Tpo4, Urb2, Urk1, and Yen1. Clusters of related proteins from each homolog family were annotated manually as “cluster IDs.”

Real-time PCR FUR1 DNA and RNA.

Flucytosine resistance, associated with mutations in FUR1, FCY1, or FCY2 of C. albicans and C. lusitaniae, is well-described by other investigators (22, 23, 32). However, the deletion in FUR1 has not been documented in the published literature. Therefore, we confirmed in silico discovery of FUR1 (1Δ33) by designing new real-time PCR test per approaches described in detail by our group (61, 62). Briefly, a set of primers and probe were designed from the deleted region of the FUR1: V2621F forward, 5′-CACTTCACCGATCTTCATCT-3′, V2621R reverse, 5′-CCCTCTTAACTGGTTGGTTT-3′, and V2621P probe 5′-FAM (6-carboxyfluorescein)-TGTCAACAC/ZEN/CCCAGAACGTCAGCA-3′ IABkFQ (Iowa Black fluorescent quencher). Another set of primers and probe were designed from the conserved region of the FUR1 as a control, and they were as follows: V2622F forward, 5′-CAGGATCATCAGACTCTTGG-3′, V2622R reverse, 5′-CCATTTCCATTGACTCTCCA-3′, and V2622P probe 5′-Cy5-CCAGTTGCC/TAO/CGTGGAAGAAGCC-3′IAbRQSp (Iowa Black red quencher spacer). The internal ZEN and TAO quencher was incorporated between bases 9 and 10 from the 5′ end of the probes V2621 and V2622, respectively. All primers and probes were obtained from Integrated DNA Technologies (Coralville, IA).

The RT-qPCR was performed using a ABI7500 real-time PCR system per general scheme described earlier (61). Candida auris RNA was extracted with Epicenter MasterPure Yeast RNA purification kit (Lucigen, Middleton, WI). DNA contamination in the nucleic acid extracts was removed using TURBO DNA-free kit (Invitrogen, Carlsbad, CA). RNA prep was quantified, and quality checked using NanoDrop 2000/2000c Spectrophotometers (Thermo Scientific, Waltham, MA) and the Qubit RNA HS assay kit (Thermo Scientific, Waltham, MA). Approximately 3 μg of RNA was used as a template for the first-strand cDNA synthesis using the SuperScript III One-Step cDNA synthesis kit (Invitrogen, Carlsbad, CA). The qRT-PCR was performed using a ABI7500 real-time PCR system with probes (V2621P and V2622P) and primers (V2621F,V2621R reverse; V2622F forward, V2622R reverse). Each sample was tested in duplicate in 20 μL volumes using an optical 96-well reaction plate. Each reaction mixture contained 1× PerfeCTa Multiplex qPCR ToughMix (Quanta Biosciences), a 500 nM concentration of each primer (V2621 and V2622), a 100 nM concentration each probe (V2621P and V2622P). Each reaction volume was 17 μL of master mix and 3 μL cDNA as a template. The real-time qPCR conditions were 95°C for 20 s, 45 cycles (95°C for 3 s, 60°C for 30 s). Each reaction was run in duplicate, and the average threshold cycle (Ct) calculated for each sample (62).