Antifungal susceptibility, genotyping, resistance mechanism, and clinical profile of Candida tropicalis blood isolates

Abstract

Candida tropicalis is one of the major candidaemia agents, associated with the highest mortality rates among Candida species, and developing resistance to azoles. Little is known about the molecular mechanisms of azole resistance, genotypic diversity, and the clinical background of C. tropicalis infections. Consequently, this study was designed to address those questions. Sixty-four C. tropicalis bloodstream isolates from 62 patients from three cities in Iran (2014–2019) were analyzed. Strain identification, antifungal susceptibility testing, and genotypic diversity analysis were performed by MALDI-TOF MS, CLSI-M27 A3/S4 protocol, and amplified fragment length polymorphism (AFLP) fingerprinting, respectively. Genes related to drug resistance (ERG11, MRR1, TAC1, UPC2, and FKS1 hotspot9s) were sequenced. The overall mortality rate was 59.6% (37/62). Strains were resistant to micafungin [minimum inhibitory concentration (MIC) ≥1 μg/ml, 2/64], itraconazole (MIC > 0.5 μg/ml, 2/64), fluconazole (FLZ; MIC ≥ 8 μg/ml, 4/64), and voriconazole (MIC ≥ 1 μg/ml, 7/64). Pan-azole and FLZ + VRZ resistance were observed in one and two isolates, respectively, while none of the patients were exposed to azoles. MRR1 (T255P, 647S), TAC1 (N164I, R47Q), and UPC2 (T241A, Q340H, T381S) mutations were exclusively identified in FLZ-resistant isolates. AFLP fingerprinting revealed five major and seven minor genotypes; genotype G4 was predominant in all centers. The increasing number of FLZ-R C. tropicalis blood isolates and acquiring FLZ-R in FLZ-naive patients limit the efficiency of FLZ, especially in developing countries. The high mortality rate warrants reaching a consensus regarding the nosocomial mode of C. tropicalis transmission.

Introduction

Candida tropicalis is the first or second common cause of candidaemia in developing countries such as India¹ and Brazil,² where the vast majority of cases are treated with fluconazole (FLZ) because of the high cost of echinocandins.^1,3 However, an increasing number of candidaemia studies have shown a significant increase in azole resistant C. tropicalis blood isolates^4–6 and some reported pan-azole^7,8 and pan-azole and amphotericin B (AMB) resistant isolates.⁹ A comprehensive candidaemia study conducted in India revealed that the multidrug resistance (MDR) trait was equally seen for C. tropicalis and Candida auris isolates.¹ The isolation of azole-resistant C. tropicalis in azole-naive patients^4,8 will further limit the available treatment options and jeopardize the lives of patients, especially in developing countries. Furthermore, patients infected with C. tropicalis experience longer hospitalization and higher mortality compared to those infected with Candida albicans.¹⁰ Surprisingly, over the course of 7 years surveillance of a C. tropicalis candidaemia study conducted in Taiwan, the authors noticed replacement of fluconazole susceptible dose-dependent isolates by those that are resistant to all azole drugs tested, including FLZ, voriconazole (VRZ), itraconazole (ITZ), and posaconazole (PSZ).⁸ Collectively, these evidences show that C. tropicalis is not an innocuous azole-susceptible species and should be targeted by surveillance studies.

The major azole-resistant determinants in C. tropicalis are genes encoding for lanosterol 14-α-demethylase (ERG11), efflux pumps (CDR1 and MDR1),^11,12 and the ERG11 expression regulator (UPC2).¹³ In C. albicans, specific gain-of-function mutations in MRR1 and TAC1, that is, transcription regulators of MDR1 and CDR1, are linked to the overexpression of the corresponding efflux pump genes and, therefore, azole-resistance.¹⁴ However, no data on the occurrence of mutations in MRR1 and TAC1 in C. tropicalis azole-susceptible, azole-susceptible dose-dependent, and azole-resistant strains are available. In terms of echinocandin resistance, specific mutations at hotspots (HS) HS1 and HS2 of the FKS1 gene encoding a 1,3-β-glucan synthase component are directly linked to the resistance in C. tropicalis.¹⁵

Although outbreaks^16,17 and clonal expansion of C. tropicalis in some clinical settings have been documented¹⁸ and this species was found as a gut commensal in 46% of healthy individuals studied,¹⁹ the other biological niches of the species yet remain to be discovered. Typing techniques permit identification of the source of infection, which may be followed by implementing appropriate preventive strategies, for example, initiation of antifungal prophylaxis or infection control, and may also facilitate the identification of genotypes that are associated with high mortality³ and virulence.²⁰ While the typing resolution of multi-locus sequence typing is almost the same as that of microsatellite typing of six loci of C. tropicalis isolates,²¹ the resolution of amplified fragment length polymorphisms (AFLP) genotyping is even better than the MLST when applied on clinical C. albicans isolates.²² Moreover, despite the universality of this technique that obviates the need for previous knowledge about the genome of a target species,²³ AFLP has never been used for typing of C. tropicalis isolates.

Here we undertook a systematic multicenter study and retrospectively analyzed 64 C. tropicalis blood isolates recovered from candidaemia patients in Iran during 2014–2019. The isolates were characterized by MALDI-TOF MS, antifungal susceptibility testing (AFST), and sequencing of drug-resistance genes. AFLP analysis was used to assess their genotypic diversity. Since neutropenic patients and those suffering from leukemia have a high propensity for developing C. tropicalis candidaemia,¹⁰ we also systematically analyzed the clinical data of patients included in the study.

Methods

Study design, isolates, and growth conditions

Sixty-four C. tropicalis blood isolates recovered from September 2014 to February 2019 from candidaemia patients admitted to 10 hospitals in three major cities of Iran (Mashhad, Shiraz, and Tehran) were included in the study. There was no restriction of age, sex, underlying conditions, and ward. The blood bottles were incubated in Bactec devices (Becton Dickinson, Franklin Lake, NJ, USA); 100 μl of positive blood cultures were inoculated onto Sabouraud dextrose agar and chromogenic media (Candiselect, Bio-Rad, Hercules, CA, USA) to ascertain the homogeneity of species involved, and incubated at 37°C for 24–48 hours. The candidaemia studies undertaken at each center had been approved by the ethical committee of the affiliated university, with the appropriate ethical approvals granted (approval numbers IR.SUMS.REC.1397.365, IR.MUMS.REC.1397.268, and IR.TUMS.SPH.REC.1396.4195). Written consent was obtained from patients, and patient identity was blinded to the personnel performing data analysis. Antifungal naive patients were noted if a given patient did not receive any systemic antifungal 90 days prior to manifestation of candidaemia.

Isolate identification, DNA extraction, PCR, and sequencing

Strain identification was confirmed by MALDI-TOF MS (MALDI Biotyper; Bruker Daltonik, Bremen, Germany) using the full extraction method.²⁴ DNA was extracted using a CTAB-based extraction method.²⁵ Primers to amplify the full open reading frame of MRR1, TAC1, UPC2, and ERG11, and HS1 and HS2 of FKS1 were designed (Table S1) using the genome of C. tropicalis MYA-3404 (AAFN00000000.2)²⁶ as a reference (wild-type sequences are listed at the end of Supplementary files). Amplification of each gene was performed using the program and conditions specified in Table S2. Amplicons were subjected to Sanger sequencing and the obtained sequences were analyzed by SeqMan Pro (DNASTAR, Madison, WI, USA). The analyzed sequences were aligned using MEGA v7.0,²⁷ the mutations were mapped to reference genes, and the corresponding mutations peaks were rechecked by using SeqMan Pro to assure the accuracy. Heterozygosity is defined when a double, clean, and decent peak representing two different nucleotides was observed at the same position.

Antifungal susceptibility testing (AFST)

AFST followed the CLSI M27-A3/S4 protocol.^28,29 The six antifungal agents tested were fluconazole (FLZ), voriconazole (VRZ), itraconazole (ITZ), and amphotericin B (AMB) (all from Sigma-Aldrich, St. Louis, MO, USA); micafungin (MFG; Astellas, Munich, Germany); and anidulafungin (AFG; Pfizer, NY, USA). Caspofungin was not tested because of the reported inter-laboratory variation.³⁰ Plates were incubated at 37°C for 24 hours and visually assessed. Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were included as quality controls. Minimum inhibitory concentration (MIC) values of FLZ, VRZ, AFG, and MFG were interpreted based on the species-specific clinical break points, with MIC ≥ 8 μg/ml denoting FLZ-resistance (R); and MIC ≥ 1 μg/ml denoting VRZ-R, AFG-R, and MFG-R;³¹ MIC = 4 μg/ml and 0.25 ≤ MIC ≥ 0.5 μg/ml to indicate FLZ-susceptible dose-dependent (FLZ-SDD) and VRZ-intermediate phenotypes (VRZ-I), respectively. Because of the lack of clinical breakpoints, epidemiological cutoff values (ECV) were used for AMB and ITZ, with MIC values >2 μg/ml and >0.5 μg/ml considered non-wild type (NWT) for AMB and ITZ, respectively.³¹

AFLP genotyping

DNA samples were analyzed by using a previously described AFLP protocol.³ Fluorescently labeled amplicons were resolved by capillary electrophoresis (ABI 3730xL Genetic Analyzer, Applied Biosystems, Palo Alto, CA, USA), and the data were analyzed using Bionumerics v7.6 (Applied Math, Sint-Martens-Latem, Belgium). The following reference and type strains were included in the AFLP analysis for comparative purposes: C. tropicalis CBS 433, CBS 643, CBS 2313, CBS 6862; C. albicans CBS 2704 and CBS 2705; and Candida dubliniensis CBS 7988.

Data availability

All sequences generated in the current study were deposited in GenBank (https://www.ncbi.nlm.nih.gov/genbank/) under the following accession numbers MK906127–MK906190 (ERG11), MK906052–MK906076 (MRR1), MK906077–MK906101 (TAC1), MK906102–MK906126 (UPC2), MK906191–MK906254 (HS1 of FKS1), and MK906255–MK906318 (HS2 of FKS1).

Statistical analysis

All statistical analyses were performed using SPSS v24 (SPSS Inc., Chicago, IL, USA) (Supplementary Files, statistical analysis section). The associations between genotypes, and FLZ and VRZ resistance were evaluated using two-tailed χ² test. Since the hospitalization duration data were not normally distributed, the association between genotypes and duration of hospitalization was evaluated using the Kruskal-Wallis test. To assess the direct and indirect influence of genotypes on mortality, the logistic multivariate regression and path analysis was used. P values < .05 were considered statistically significant.

Results

Clinical characteristics

Sixty-four C. tropicalis isolates were recovered from 62 patients, 42% (n = 26) of whom were male and 58% (n = 36) female, with a median age of 37 years (2 months to 90 year-old) (Table S3). Most isolates were obtained at Mashhad (n = 31, 48.4%), followed by Tehran (n = 28, 43.7%), and Shiraz (n = 5, 7.8%). Sepsis was observed in 31 patients (50%) when candidaemia was manifested. Pre-exposure to antibiotics (n = 64, 100%), central venous catheter insertion (n = 53, 84.6%), mechanical ventilation (n = 37, 59.7%), surgery (abdominal [n = 17, 27.1%] and non-abdominal [n = 8, 12.9%]), parenteral nutrition (n = 20, 32.2%), administration of immunosuppressive drugs (n = 14, 22.6%), and neutropenia (n = 12, 19.4%) were the major risk factors for the development of candidaemia (Table S3). AMB was the most widely used antifungal (n = 28, 45.2%), followed by FLZ (n = 16, 25.8%), CSP (n = 11, 17.7%), and nystatin (n = 4, 6.4%), while nearly a quarter of patients (n = 15) did not receive any antifungals (Table S3). The overall mortality rate was nearly 60% (n = 37). The highest mortality rates were reported for Mashhad (n = 21, 67.7%) followed by Shiraz (n = 3, 60%) and Tehran 57.6% (n = 15) (Tables S3a and S3b).

AFST

Resistance to VRZ (MIC ≥ 1 μg/ml), FLZ (MIC ≥ 8 μg/ml), and MFG (MIC ≥ 1 μg/ml) was noted in seven (10.93%, 7/64), four (6.25%, 4/64), and two (3.12%, 2/64) isolates, respectively. Moreover, some isolates denoted VRZ-I (0.25 ≤ MIC ≥ 0.5 μg/ml, n = 18; 18/64) and FLZ-SDD (MIC = 4 μg/ml, n = 7; 7/64) (Table 1 and Table S4). All isolates were susceptible to AFG and AMB, while two were NWT for ITZ (MIC > 0.5 μg/ml, n = 2; 2/64). Three isolates were resistant to ≥2 azole drugs (4.7%); one showed pan-azole resistance to all azole drugs tested (1.6%); and two were cross-resistance to FLZ and VRZ (3.2%) (Tables 1 and 2, and Table S4). Except for two isolates (262E and N186), no multi-azole resistant isolates (to two or three azoles tested) represented a single genotype.

Table 1.

Antifungal susceptibility data for Candida tropicalis isolates obtained in the current study.

	Antifungal drugs
Susceptibility data	FLZ	VRZ	ITZ	MCF	ANF	AMB
MIC values (μg/ml)
≤0.016		5		18	23
0.03		7		11	9
0.06		16	2	17	17
0.125	2	9	23	10	8	1
0.25	12	9	23	3	3	5
0.5	18	11	14	3	3	29
1	14	6	1	2		28
2	7					1
4	7	1
8	3
16			1
32
≥64	1
Range	0.125–64	0.016–4	0.06–16	0.008–1	0.008–0.5	0.125–2
GM	0.878126	0.142408	0.2634	0.050506	0.038356	0.641435
MIC 50	0.5	0.125	0.25	0.062	0.025	0.5
MIC 90	4	1	1	0.25	0.125	1

MIC values denoted in boldface are modal values.

AMB, amphotericin B; ANF, anidulafungin; FLZ, fluconazole; GM, geometric mean value; ITZ, itraconazole; MCF, micafungin; MIC, minimum inhibitory concentration VRZ, voriconazole.

Table 2.

Sequences of the target genes in all FLZ-R (n = 4) and FLZ-SDD isolates (n = 7), and randomly selected FLZ-S isolates (n = 15).

Strain no.	FLZ (µg/ml)	VRZ (µg/ml)	ITZ (µg/ml)	MRR1	TAC1	UPC2	Genotype
Fluconazole-control isolates (n = 12)
N8	0.5	0.125	0.5	A87T, V133A, M1022I, T1042N, T1044N, I1130M	WT	A251T, Q289L, G392E	G5
N15	0.5	0.125	0125	WT	L278S	N98S, L158V	G2
N71	0.125	0.016	0.125	S523F, K757E	L278S, D350N, F470C, D790N	G392E	MG
N104	1	0.25	0.125	M1022I, T1042N, T1044N, D1092E, I1130M	L278S	WT	G4
N147	0.5	0.031	0.125	M1022I, T1042N, T1044N, D1092E, I1130M	L278S	WT	G4
N195	0.25	0.031	0.125	S523F, K757E	L278S, D350N, F470C, D790N	A297S*, G392E	G4
N210	0.125	0.062	0.125	WT	L278S	N98S, L158V, A251T	G2
SU-221	0.25	0.015	0.06	A87T), V133A, M1022I, T1042N, T1044N, D1092E, I1130M	L278S, F470C	WT	G5
SU-267	0.25	0.125	0.125	WT	L278S, F470C, S884G	N98S, L119F, A147T, L158V	MG
10BC	1	0.5	0.5	I408T, M1022I, T1042N, T1044N, D1092E, I1130M	L278S, F470C, D790N, D790N	A251T, Q289L, A297S*	G6
24BC	0.5	0.5	0.125	S523F, K757E	L278S, F470C, D790N	G392E, T560N	G6
115-1BC	0.5	0.125	0.125	A87T, V133A, M1022I, T1042N, T1044N, I1130M	WT	A251T, Q289L	G5
Fluconazole-susceptible isolates (n = 3)
N26	2	0.5	0.125	S523F, K757E	L278S, D350N, F470C, D790N	A297S *, G392E	G6
SU-235	2	0.125	0.125	WT	L278S	N98S, L158V, A251T	G2
85BC	2	0.25	0.125	WT	L278S	N98S, L158V	G2
Fluconazole-susceptible dose-dependent isolates (n = 7)
SU-239	4	0.06	0.125	WT	L278S	L158V, N98S, L158V, F571Y	G2
N17	4	0.5	1	S523F, K757E, I1130M	L278S, L340*, D350N, F470C, D790N	NA
75BC	4	0.125	0.25	S523F, K757E, I1130M	L278S	WT	G4
82BC	4	1	0.25	I1130M	WT	A263T	G4
107BC	4	0.062	0.125	S523F, K757E	L278S, F470C, D790N	WT	G3
113-1BC	4	1	0.125	I1130M	L278S, F470C	A251T, Q289L	G4
115-2	4	0.5	0.06	A87T, V133A, M1022I, T1042N, T1044N, I1130M	WT	A251T, Q289L, G392E	G5
Fluconazole-resistant isolates (n = 4)
99BC	8	1	0.5	V133A, A647S, M1022I, T1044N, D1092E, I1130M	L278S, F470C	WT	G5
113-2BC	8	0.062	0.25	WT	L278S	N98S, L158V, N230S, T241A	G2
262E	64	4	16	M1022I, T1042N, T1044N, D1092E, I1130M	R47Q, N164I, L278S	Q340H, T381S	G4
527E	8	1	0.5	T255P	R47Q, N164I, L278S, F470C, D790N	A147T, A251T, Q289L	MG

Underlined boldface amino acid substitutions were only identified in FLZ-R or FLZ-SDD isolates; asterisk-denoted boldface amino acids were exclusively identified in resistant isolates in previous studies; boldface italicized amino acids were exclusively found in susceptible isolates in the current and previous studies. All strains carried the ERG11 WT sequence, except for SU-239 (K90I) and SU-267 (I25A). FLZ, fluconazole; ITZ, itraconazole; VRZ, voriconazole.

^*Stop codon; NA, not amplified.

Mutation analysis of the isolates

We did not find previously known mutations in ERG11 directly causing fluconazole-resistance in our fluconazole-resistant isolates.^11,12 Since FLZ MIC values depend on the heterozygosity and homozygosity status of the MRR1, TAC1, and UPC2 genes³² and in order to identify specific mutations for each MIC category, 26 isolates were categorised as control (C, MIC < 2 μg/ml) (n = 12), S (MIC = μg/ml) (n = 3), SDD (MIC = 4 μg/ml) (n = 7), and FLZ-R (MIC ≥ 8 μg/ml) (n = 4). Subsequently, target genes of those 26 isolates were sequenced (Table 2 and Table S4). Of those, T255P and A647S in MRR1, R47Q and N164I in TAC1, and T241A, Q340H, and T381S in UPC2 were exclusively identified in FLZ-R isolates, while F571Y in UPC2 and L430* (stop codon) in TAC1 were only identified in an FLZ-SDD isolate (Table 2). The only pan-azole resistant isolate simultaneously carried FLZ-R specific mutations in both UPC2 (Q340H and T381S) and TAC1 (R47Q and N164I) genes. Although those ITZ-R isolates did not harbor any specific mutations, one of the VRZ-R isolates showed a unique mutation (A263T) in UPC2. No association between FLZ exposure and FLZ resistance was observed, as patients carrying FLZ-R strains had never been administered FLZ (Table S4).

AFLP genotyping of the isolates

AFLP analysis revealed five main genotypes (G2–G6) accounting for 89% of the isolates (n = 57) and seven minor genotypes, each represented by a single isolate (Fig. 1). Considering the major genotypes, G4 was the most prevalent (n = 25, 38.4%), followed by G6 (n = 11, 17.1%), G2 and G5 (n = 9 each, 14%), and G3 (n = 3, 4.6%) (Fig. 1). The isolates from Shiraz and a hospital from Tehran did not exhibit conspicuous accumulation of any specific genotype. However, 58% (n = 18) of Mashhad isolates represented G4, and 61% (n = 11) of those isolates were from different wards of a single hospital (Imam Reza, years 2015–2019). Furthermore, 37.5% (n = 6) of isolates from the Children's Medical Centre in Tehran represented G6 and all originated from the intensive unit wards (years 2015–2016). In case of three patients with duplicate isolates, except for 115-1 and 115-2BC that clustered in the same genotype, the isolates represented different genotypes (368 and 369E, and 113-1 and 113-2BC). Multivariate logistic regression, path analysis, and Kruskal-Wallis test did not indicate any association between the genotypes and patient mortality (P = .47), or genotypes and duration of hospitalization (P = .6) (Supplementary Files, Statistical analysis section). Further, as determined by using the two-tailed χ² test, the genotypes (G2–G6) and azole resistance were not significantly associated (Supplementary Files, Statistical analysis section).

Figure 1.

AFLP fingerprinting revealed five major and seven minor genotypes (shown with gray color). Each genotype (G) was defined by distinct color as follows: G2 = yellow, G3 = Orange, G4 = Green, G5 = Light blue, and G6 = Ultraviolet. G4 was the most predominant genotype found in all centers involved.

Discussion

The patients included in the current study had common risk factors for the development of candidaemia, such as central venous catheter insertion, pre-exposure to antibiotics, mechanical ventilation, and abdominal surgery.¹ Even though leukemic patients show a high propensity for developing C. tropicalis candidaemia,¹⁰ we found that, similar to a study from Italy,¹⁸ other complications were the most prevalent underlying condition. This discrepancy could be explained by differences in the target populations examined. The mortality reported in the current study was even higher than that reported for C. glabrata (60% vs. 37.5%),³ which is consistent with studies from Italy³³ and the United States,³⁴ and corroborates the highly virulent nature of C. tropicalis and its poor prognosis when compared to the other non-albicans Candida (NAC) species.^35,36

Among the azoles tested in this study, we found the highest level of resistance to VRZ (n = 7, 7/64), followed by FLZ (n = 4, 4/64) and ITZ (n = 2, 2/64). In the current study, the observed low level of resistance to major antifungal drugs (except for ITZ) was comparable with that reported for Asian¹ and Middle Eastern countries,³⁷ and Italy and Spain,³⁸ and contrasted with the high reported resistance rates to FLZ and VRZ in China⁵ and Taiwan.⁸ Although previous and prolonged exposure is the main driving factor for emerging antifungal resistant isolates,^39,40 surprisingly, we did not find any association between FLZ-R and previous exposure with FLZ, as patients infected with FLZ-R isolates did not receive FLZ 90 days prior to candidaemia manifestation. This is in agreement with a previous study conducted in Japan⁴ and Taiwan⁸ where almost 50% of patients infected with fluconazole-resistant strains were azole-naive. We speculate that either host conditions triggered alternative pathways leading to resistance⁴¹ or the azole-resistant strains were acquired from the hands of healthcare workers (HCWs),⁴² in addition to the possible link between antibiotic prophylaxis and FLZ-R.⁴³ Alternatively, a study in Taiwan noticed that a fruit-related azole resistant C. tropicalis isolate clustered with the fluconazole non-susceptible (FNS) blood isolates and this coincided with a fourfold increase in use of fungicides in agricultural applications in this country.⁸ Therefore, the authors assumed that azole-naive patients might have acquired these FNS isolates from the environment,⁸ the same as what was observed for Aspergillus fumigatus.⁴⁴

Mechanistically, we did not identify any accountable mutations in the ERG11 gene, but several suggestive mutations in MRR1 (T255P, 647S), TAC1 (N164I, R47Q), and UPC2 (Q340H, T381S) were exclusively identified in FLZ-resistant isolates. Furthermore, unlike a previous report of A297S amino acid substitution found only in FLZ-R isolates,¹³ we here identified this mutation exclusively in FLZ-S isolates. Although susceptible isolates were included in that study, the authors did not explore the occurrence of mutations in MRR1 and TAC1; therefore, they might have been biased and other accountable mutations in those genes might have been overlooked. In our study, one VRZ-R isolate carried a unique mutation in UPC2 (A263T); while this mutation was previously found in VRZ-S isolates,¹³ hence it may not drive resistance to VRZ.

AFLP revealed that isolates from all the analyzed centers represented the predominant genotype G4, which might be an indication for intra-hospital and/or clonal transmission of C. tropicalis. Considering that 80% of yeasts isolated from the hands of HCWs are C. tropicalis,⁴² a specific genotype was found to be enriched in Taiwan¹⁷ and Italy,¹⁸ and the same clone of C. tropicalis blood isolates was identified in a unit environment and on hands of HCWs,¹⁸ thus likely suggesting indeed transmission may have occurred via the hands of HCWs. Interestingly, implementation of routine infection control strategies led to termination of an ongoing C. tropicalis outbreak,¹⁸ which in view of the high mortality rate posed by this species further highlights the importance of application of typing techniques to assess the genotypic diversity of C. tropicalis in healthcare settings. The notable difference in typing protocols, study design, and patient size and isolates numbers hinder drawing a clear conclusion regarding the mode of transmission of C. tropicalis in the hospital settings and the current knowledge in this regard remained speculative. Therefore, application of standardized and resolutive typing techniques, such as whole genome sequencing, might address this question.

Although other studies reported a link between genotype and mortality,³ we did not find such a link in the current study. Similarly, we did not find links between the genotype and duration of hospitalization, and genotype and azole susceptibility. Interestingly, two duplicate isolates from two patient belonged to different genotypes than the original isolate, which could be explained by either host and/or antifungal-triggered stress followed by minimal to gross chromosomal changes⁴⁵ or introduction of a new isolate into the bloodstream.

The current study has some limitations. For example, we did not analyze the expression of efflux pump genes, such as CDR1 and MDR1, as an alternative azole resistance mechanism. Further, mutations identified in FLZ-R isolates are purely suggestive and heterologous expression in a susceptible C. tropicalis isolate is required to confirm involvement in FLZ-resistance.

The high mortality rate noted in the current study might be alleviated if resolutive typing techniques become part of a routine clinical procedure, considering the speculation that this species might be horizontally transferred. Furthermore, the presented data suggested that a full picture should be considered (MRR1, TAC1, and UPC2 sequencing) to understand the underlying molecular azole-resistance mechanisms. Finally, the increasing risk of non-azole resistant C. tropicalis from blood isolates and FLZ-R isolates without previous exposure to this drug highlight the importance of species-specific candidaemia studies to extensively explore and highlight the clinical and microbiological differences between various Candida species, leading to better patient management strategies.

Declaration of interest

M.K. is an employee of Bruker Daltonik GmbH, Bremen, Germany, the manufacturer of the MALDI-TOF MS system used for Candida identification in the current study. There are no other conflicts of interest to declare. The authors alone are responsible for the content and the writing of this paper.

Funding

This work was supported by the European Union's Horizon 2020 research and innovation program under the Marie Sklodowska-Curie call [grant no. 642095]; National Health Department of China [grant no. 2018ZX10101003]; National Natural Science Foundation of China [grant no. 31770161]; Second Military Medical University [grant no. 2017JZ47]; Shanghai Science and Technology Committee [grants no. 14DZ2272900, 14495800500]; and Shiraz University of Medical Sciences [grant no. 98-01-43-21203].