Prevalence Rates, Species Distribution and Antifungal Susceptibility of Rare Candida and Related Saccharomycotina Yeasts Causing Bloodstream Infections in 28 Medical Centres

ABSTRACT

Background

There is a lack of contemporary studies addressing the prevalence rates and antifungal susceptibility of bloodstream infections (BSI) caused by rare yeasts (RY) from the Saccharomycotina subphylum.

Objectives

This 16‐year multicentre study (2007–2023) aimed to assess the prevalence and antifungal susceptibility of rare yeasts isolated from BSIs in 28 Brazilian tertiary care hospitals.

Methods

Yeasts from the Saccharomycotina subphylum, excluding common Candida species and Basidiomycota, were selected from BSI episodes. Species were identified using proteomics and molecular methods. Antifungal susceptibility testing was performed by EUCAST broth microdilution.

Results

Among 2655 BSI episodes, 100 (3.76%) involved rare yeasts. Prevalence showed a slight, non‐significant increase from 3.16% (2007–2015) to 4.17% (2016–2023; p = 0.183). A total of 11 genera and 20 species were identified, with Clavispora lusitaniae (20%), Wickerhamomyces anomalus (16%), Candida haemulonii (14%) and Candida duobushaemulonii (12%) being the most frequent species. Eight species, including Pichia kluyveri and Kodamaea ohmeri, appeared only in the latter period. Most rare yeasts showed low MICs to amphotericin B and anidulafungin, except for Clavispora lusitaniae, Candida haemulonii and Candida duobushaemulonii, which were less susceptible to amphotericin B. Wickerhamomyces anomalus and Candida duobushaemulonii also showed reduced susceptibility to fluconazole.

Conclusions

This study provides valuable insights into the prevalence rates, species distribution and antifungal susceptibility of rare yeasts causing BSIs in tertiary care hospitals. Our findings highlight the need for continuous surveillance, incorporation of new diagnostic and tailored therapeutic strategies to mitigate morbidity and mortality due to invasive infections caused by emerging fungal pathogens.

1. Introduction

Incidence rates of Candida bloodstream infections (BSIs) exhibit high geographic variability ranging from 1.9 to 21.8 episodes per 100,000 inhabitants worldwide [1]. In Latin America, incidence rates range from 0.21 to 2.8 cases per 1000 hospital admissions, with Brazil reporting rates of 1.3 to 2.5 cases per 1000 admissions annually [2, 3, 4]. Candida BSIs substantially impact length of hospitalisation, mortality rates and costs of health care assistance [5, 6, 7].

Recently, the genus Candida has undergone significant taxonomic reclassification, with many species now being reclassified into different genera within the Saccharomycotina yeasts [5]. In this context, species such as Nakaseomyces glabratus (syn. Candida glabrata), Meyerozyma guilliermondii (syn. Candida guilliermondii ), Clavispora lusitaniae (syn. Candida lusitaniae) and Candida haemulonii (syn. Candidozyma haemuli) have been reclassified due to their evolutionary and phylogenetic relationships elucidated by DNA sequencing methods that are progressively using whole genomes [8, 9].

Although less common than infections caused by Candida albicans and other prevalent Candida species, BSIs caused by rare Saccharomycotina yeasts have been increasingly reported, particularly in immunocompromised and critically ill patients [9, 10, 11, 12, 13, 14, 15, 16]. These infections pose unique challenges due to difficulties in their accurate identification, limited susceptibility to antifungal drugs, and lack of clinical data to support best clinical practices in their clinical management [17, 18, 19, 20, 21, 22, 23, 24, 25].

The aims of the present study are to evaluate historical trends in prevalence rates, species distribution and antifungal susceptibility of rare Candida and related genera within Saccharomycotina yeast causing BSIs in 28 medical centres from Brazil over a 16‐year period.

2. Materials and Methods

2.1. Selection of Clinical Isolates

All Candida and related genera within the Saccharomycotina yeasts cultured from BSI episodes documented in tertiary Brazilian medical centres that participated in national surveillance studies between 2007 and 2023 were selected (Table S1). The isolates had been sent to the Special Mycology Laboratory, Federal University of São Paulo, Brazil for further identification and antifungal susceptibility testing. Participating centres were selected among medical centres which participated in previous Candida BSI survey studies coordinated by our centre [5, 14, 15, 26]. Only non‐replicate isolates were used in the analysis. Common Candida species involved in BSIs, such as C. albicans sensu stricto, C. tropicalis , C. parapsilosis , Pichia kudriavzevii (syn. Candida krusei ), Nakaseomyces glabratus (Syn. Candida glabrata), Meyerozyma guilliermondii (syn. C. guilliermondii ), Candida auris (syn. Candidozyma auris), as well as Basidiomycota yeasts were not included. For species belonging to the newly proposed genus Candidozyma, including those formerly classified as Candida auris and members of the Candidozyma haemulonii species complex, and in accordance with recommendations of the ISHAM nomenclature Working group [27], the classical nomenclature of these agents was retained in this study [8, 27, 28, 29].

This study was approved by the Ethics Committee of Universidade Federal de São Paulo (CAE: 82433324.4.0000.5505, approved in December 2024).

2.2. Species Identification

Yeast identification at species level was performed using Matrix‐assisted laser desorption ionisation time‐of‐flight mass spectrometry (MALDI‐TOF MS). Protein extraction was conducted off‐plate using the formic acid/ethanol method, following protocols established by Bruker Daltonics (Biotyper v.4.1, Bruker Daltonics, Bremen, Germany). Off‐plate protein extraction was performed using the formic acid/ethanol method, in accordance with the Bruker Daltonics protocols [30, 31]. Protein fingerprinting profiles were automatically acquired using FlexControl version 3.1 (Bruker Daltonics) in linear mode. Identification was performed using the commercially available BDAL database version 3.1 (Bruker Daltonics). Data analysis followed the manufacturer's recommendation, with log‐scores interpreted as follows: log‐scores ≥ 2.0 indicated species‐level identification; log‐scores 1.99–1.7 indicated genus‐level identification; and log‐scores ≤ 1.69 were considered unreliable.

All isolates were also subjected to molecular identification by sequencing of the internal transcribed spacer (ITS; ITS1 + 5.8S + ITS2) region of the rDNA, as previously described [32]. The raw sequencing data were aligned using the MUSCLE algorithm and manually refined with MEGA X software [33]. Species identification was confirmed through phylogenetic analysis using the Neighbour‐Joining method based on the Kimura two‐parameter model, with 1000 bootstrap pseudo‐replicates and consideration of gap positions [34, 35]. Sequences were analysed and edited using Sequencher 4.1.4 (Gene Codes Corporation, Ann Arbor, MI, USA) and compared against reference sequences in the GenBank (https://www.ncbi.nlm.nih.gov/genbank/) and ISHAM Barcoding Databases (https://blast.ncbi.nlm.nih.gov; http://its.mycologylab.org). Matches with E‐values ≤ 10⁻⁵, coverage ≥ 98% and identity ≥ 98% were considered reliable for species identification.

2.3. In Vitro Antifungal Susceptibility Testing

Antifungal susceptibility testing (AST) was performed according to the EUCAST Definitive Document E.Def 7.4 broth microdilution method [36]. The following antifungal compounds were tested: fluconazole, voriconazole, anidulafungin and amphotericin B (Sigma Aldrich, St. Louis, MI, USA). Minimal inhibitory concentrations (MICs) were determined by using a spectrophotometer after 24 h of incubation. The concentrations tested ranged between 0.015 and 8 μg/mL for all antifungals, except fluconazole (ranged from 0.125 to 64 μg/mL). For azoles and anidulafungin, the MIC was determined as the lowest drug concentration giving inhibition growth ≥ 50% compared to the drug‐free control, while for amphotericin B, it was defined as the minimal concentration inhibiting ≥ 90% of cell growth [37]. MIC values were interpreted by using species‐specific recommendations for EUCAST MIC interpretation for rare yeast pathogens without breakpoints [25, 37]. When a species was not listed in the EUCAST document, the corresponding MIC values were not interpreted.

2.4. Statistical Analysis

Historical trends in the distribution of rare Candida and related genera within Saccharomycotina yeasts were assessed by comparing prevalence rates across two periods: Period 1 (PI), comprising all yeast isolates collected from 2007 to 2015; and Period 2 (PII), comprising all yeast isolates collected from 2016 to 2023.

Prevalence rates documented across PI and PII were compared using chi‐squared tests. Additionally, antifungal MIC values for each species were compared using the Kruskal–Wallis test. All statistical analyses were performed with a significance threshold of p < 0.05. Statistical analysis was performed with Jamovi Software version 2.6.25 (The Jamovi project, Sydney, Australia; https://www.jamovi.org).

3. Results

3.1. Historical Trends in Species Prevalence

Our bank of microorganisms encompassed a total of 2655 unique episodes of yeast‐associated BSIs (only Candida and related genera within the Saccharomycotina yeasts) which had been collected along a 16‐year period (2007–2023) of study, from 28 medical centres distributed across 12 Brazilian states (Figure 1). BSIs caused by rare Saccharomycotina yeasts accounted for 100 isolates (3.76%), distributed in 11 genera. During PI, rare yeasts were responsible for 34 out of 1073 BSI episodes (3.16%), while in PII they accounted for 66 out of 1582 episodes (4.17%). Although there was an apparent increase in absolute numbers of rare yeasts documented along PII, the difference was not considered statistically significant (p = 0.183).

FIGURE 1.

Geographic origin of 100 episodes of rare Saccharomycotina yeast bloodstream infections documented in 12 Brazilian states between 2007 and 2023.

Notably, species identification was primarily performed by MALDI‐TOF MS generating 15 reliable species‐level identification (n = 91 isolates; log‐score ≥ 2.0) with further confirmation by rDNA sequencing. These results were all confirmed by rDNA sequencing, with 100% concordance between methods (Table S2). The following five species could not be identified by their protein profile and required supplementary molecular identification through rDNA sequencing: Meyerozyma caribbica, Meyerozyma neustonensis, [Candida] spencermartinsiae, Wickerhamomyces onychis and [Candida] thasaenensis.

As illustrated in Table 1, the most prevalent rare species causing BSIs between 2007 and 2023 were as follows: Clavispora lusitaniae (n = 20, 20%), Wickerhamomyces anomalus (n = 16, 16%), Candida haemulonii sensu stricto (n = 14, 14%), Candida duobushaemulonii (n = 12, 12%) and Sugiyamaella intermedia (n = 8, 8%).

TABLE 1.

Species distribution and antifungal susceptibility of 100 rare yeasts from the Candida genus and related genera within the Saccharomycotina yeasts causing bloodstream infections (BSIs) in 28 medical centres across Brazil stratified by period: PI (2007–2015) versus PII (2016–2023).

Species	Number of isolates			Antifungal susceptibility results (μg/mL)
	PI	PII	Total	AMB	AFG	FLC	VRC
Clavispora lusitaniae	5	15	20	0.015–0.5	0.015–0.125	0.125–0.5	0.015
Wickerhamomyces anomalus	7	9	16	0.015–0.5	0.015–1	0.25–8	0.015–0.125
Candida haemulonii	6	8	14	0.015–2	0.015–0.5	0.125–16	0.015–1
Candida duobushaemulonii	2	10	12	1–> 8	0.015–0.25	4–> 64	0.03–1
Sungouiella intermedia	4	4	8	0.015–0.25	0.015–0.06	0.125–1	0.015
Candida dubliniensis	0	5	5	0.06–0.125	0.015–0.06	0.125–0.25	0.015
Candida haemulonii var. vulnera	2	1	3	0.5–1	0.015–0.06	0.5–2	0.015–0.06
Kluyveromyces marxianus	0	3	3	0.125–0.5	0.015–0.06	0.125–0.25	0.015
Pichia norvegensis	2	0	2	0.125–0.25	0.015–0.06	4–8	0.03–0.125
Wickerhamiella pararugosa	2	0	2	0.125–0.5	0.06	8	0.03–0.125
Kodamaea ohmeri	0	2	2	0.25	0.06–0.125	4–16	0.03–0.125
Meyerozyma carpophila	1	0	1	0.25	0.125	1	0.015
Yarrowia lipolytica	1	0	1	0.125	0.015	2	0.03
Pichia kluyveri	0	1	1	0.015	0.015	2	0.03
Candida viswanathii	0	1	1	0.25	0.015	0.25	0.015
Meyerozyma caribbica	1	3	4	0.015–0.5	0.06–1	0.125–16	0.015–0.25
[Candida] spencermartinsiae	1	0	1	0.25	0.15	2	0.03
Wickerhamomyces onychis	0	1	1	0.015	0.06	1	0.015
Meyerozyma neustonensis	0	1	1	1	0.5	0.125	0.015
[Candida] thasaenensis	0	2	2	0.03	0.015	1	0.03
Total	34	66	100

Note: Square brackets around genus indicates that the name awaits appropriate action by the research community to be transferred to another genus.

Abbreviations: Antifungals tested: AFG, anidulafungin; AMB, amphotericin B; FLC, fluconazole; VRC, voriconazole.

Although we fail to demonstrate a statistically significant increase in the proportion of BSIs caused by rare yeasts over the study period, some particular species were only reported along PII: Candida dubliniensis (n = 5), Kluyveromyces marxianus (n = 3), Kodamaea ohmeri (n = 2), [Candida] thasaenensis (n = 2), Pichia kluyveri (n = 1), Candida viswanathii (n = 1), Wickerhamomyces onychis (n = 1), Meyerozyma neustonensis (n = 1).

The emergence of some rare species in PII may be partially related to improvements in identification methods, which allowed more accurate recognition of rare yeast species. However, at‐risk populations, geography and clinical management practices may have also contributed to this phenomenon [1, 5, 9, 10, 13].

3.2. Antifungal Susceptibility Testing

Using the latest EUCAST guidelines for interpreting MIC results generated by assays with rare yeasts (when applicable) [24], most isolates exhibited high susceptibility to amphotericin B, anidulafungin, fluconazole and voriconazole. The lowest MIC results across all antifungals tested were obtained with Candida dubliniensis, Clavispora lusitaniae, Kluyveromyces marxianus, Pichia kluyveri, Sugiyamaella intermedia, Yarrowia lipolytica and Candida viswanathii (Tables 1 and 2).

TABLE 2.

Antifungal susceptibility results of 62 rare yeasts from the Candida genus and related genera within the Saccharomycotina yeasts.

Species/number of isolates tested	MIC	Antifungal susceptibility results (μg/mL)
		AMB	AFG	FLC	VRC
Clasvispora lusitaniae (n = 20)	MIC50	0.125	0.06	0.25	0.015
	MIC90	0.25	0.125	0.5	0.015
Wickerhamomyces anomalus (n = 16)	MIC50	0.125	0.015	4	0.06
	MIC90	0.5	0.06	4	0.125
Candida haemulonii (n = 14)	MIC50	1	0.06	1	0.015
	MIC90	2	0.25	2	0.06
Candida duobushaemulonii (n = 12)	MIC50	> 8	0.03	8	0.125
	MIC90	> 8	0.125	16	0.5

Note: Non‐parametric multiple comparison analysis revealed significant difference in MIC distributions among antifungal agents. For Amphotericin B and Fluconazole, the MICs increased in the following order: Candida duobushaemulonii > Candida haemulonii > Wickerhamomyces anomalus > Clavispora lusitaniae (p < 0.005). For Anidulafungin, Wickerhamomyces anomalus demonstrated significantly lower MICs than Candida haemulonii (p = 0.005) and Candida duobushaemulonii (p = 0.0136). Finally, for Voriconazole, the MICs followed the following spectrum: Candida duobushaemulonii > Candida haemulonii > Clavispora lusitaniae (p < 0.005).

Abbreviations: AFG, anidulafungin; AMB, amphotericin B; FLC, fluconazole; MIC₅₀, Minimum inhibitory concentration able to inhibit 50% of cell growth; MIC₉₀, Minimum inhibitory concentration able to inhibit 90% of cell growth; VRC, voriconazole.

In contrast, Candida duobushaemulonii exhibited the lowest susceptibility to antifungals, once all isolates tested showed MICs ≥ 1 μg/mL for amphotericin B, MICs ≥ 4 μg/mL for fluconazole, and 4 out of 12 isolates tested exhibited anidulafungin MICs ≥ 0.125 μg/mL (Tables 1 and 2). Despite being highly susceptible to amphotericin B, 3 out of 4 clinical isolates of Meyerozyma caribbica exhibited MICs ≥ 0.25 μg/mL and MICs ≥ 4 μg/mL against anidulafungin and fluconazole, respectively.

It was possible to determine antifungal MIC₅₀ and MIC₉₀ values for all species represented by ≥ 10 isolates tested (Table 2). In this regard, it should be noticed that Candida duobushaemulonii exhibited fluconazole MIC₅₀ and MIC₉₀ of 8 μg/mL and 16 μg/mL, respectively, as well as high amphotericin B (MIC₅₀ and MIC₉₀ ≥ 8). Candida haemulonii isolates exhibited MIC₅₀ and MIC₉₀ values of 1 and ≥ 2 μg/mL for amphotericin B, respectively, whereas anidulafungin MIC₉₀ value was 0.25 μg/mL.

Finally, the interpretative breakpoints for rare yeasts provided by EUCAST guidance for rare yeast with no breakpoints [25, 37] are available for only 9 of the 20 species identified in this study (n = 53 isolates). Interestingly, after applying these interpretive breakpoints in our collection, only two isolates were considered resistant to antifungals: (i) 1 isolate of Wickerhamomyces anomalus, which exhibited an MIC value of 1 μg/mL for anidulafungin and (ii) 1 isolate of Meyerozyma caribbica which presented an MIC value of 0.25 μg/mL for voriconazole.

4. Discussion

Rare Candida species and related genera within Saccharomycotina yeasts are able to cause life‐threatening invasive infections and have been increasingly reported worldwide [9]. Many of these species were recently reassigned to new genera and require advanced diagnostic tools for accurate identification that are often unavailable in routine clinical laboratories in low‐ and middle‐income countries [38, 39, 40]. As a result, data on their true prevalence, risk factors, clinical presentation, antifungal resistance patterns and associated mortality remain limited [9, 10, 11, 12, 13].

In our multicentre study evaluating a 16‐year period, rare yeasts were responsible for 3.76% (100 isolates) of 2665 BSI episodes across 28 Brazilian hospitals. This prevalence rate aligns with findings from two other multicentre studies recently published which reported 3.5% and 3.9% of rare yeast infections among 399 and 2586 episodes of candidemia documented in European medical centres and Australia, respectively [16, 41].

We observed a slight increase in the prevalence of rare yeast infections over time: from 3.16% in PI (2007–2015) to 4.17% in PII (2016–2023). A similar trend has been reported in Denmark, where the proportion of BSIs caused by rare yeasts increased from 2.7% to 3.6% along two national surveillance studies conducted in two periods: 2004 and 2009 versus 2012 and 2015 [42, 43]. Importantly, in our study several species such as Pichia kluyveri, Kodamaea ohmeri, Candida dubliniensis, Kluyveromyces marxianus and Candida viswanathii were exclusively detected during PII, suggesting a shift in the species distribution of rare yeast causing BSIs during the study period.

Notably, some authors classified Meyerozyma guilliermondii as a rare etiologic agent of fungemia [16, 41]. However, in the Brazilian context, this pathogen has been commonly reported in BSI, as documented by previous national surveys of candidemia performed in our region [14, 44, 45]. Consequently, Meyerozyma guilliermondii sensu stricto was excluded from our list of rare pathogens causing fungemia.

Four species accounted for 62% (n = 62 isolates) of rare yeast BSIs in our cohort: Clavispora lusitaniae, Wickerhamomyces anomalus, Candida haemulonii sensu stricto, Candida duobushaemulonii. They all can be reliably identified using commercial MALDI TOF‐MS systems equipped with updated commercial databases, supporting the utility of this technology for species‐level identification with facilitating their accurate fast identification [46, 47, 48].

In our casuistic, MALDI‐TOF MS accurately identified with high‐confidence scores (log‐score ≥ 2.0) all isolates from 15 rare yeast species.

Indeed, the spectrometry failed to generate reliable identification of only five species: Meyerozyma caribbica, Meyerozyma neustonensis, [Candida] spencermartinsiae, Wickerhamomyces onychis and [Candida] thasaenensis. All those isolates were further identified by DNA sequencing. These findings highlight the limitations of current spectral databases for identifying some rare yeasts causing human disease. We underscore the need for continuous updating of MALDI‐TOF MS libraries incorporating protein profiles of emergent human pathogens after their correct identification by complementary molecular assays. This is particularly relevant for species such as Meyerozyma caribbica, which accounted for 4% of the cases in our study and has also been reported in other analyses [14, 18]. It is worth noting that this is the first study to report episodes of fungemia caused by the following three rare yeast pathogens: Meyerozyma neustonensis, Wickerhamomyces onychis and [Candida] thasaenensis.

In this present study, Clav. lusitaniae accounted for 0.7% of 2665 episodes of BSIs. Although global data on Clav. lusitaniae BSIs remain limited, recent reports suggest a rise in their incidence, particularly among immunocompromised and paediatric patients [49, 50, 51, 52]. A US cancer centre reported an incidence rate of 1.45 episodes of Clav. lusitaniae BSIs per 100,000 inpatient days assisted along the period between 1998 and 2013 [49]. In Kuwait, a retrospective study reported a prevalence rate of 2.25% of Clav. lusitaniae among 990 Candida BSIs episodes, with 45% of them reported among neonates [52]. Other recent surveys of candidemia have found prevalence rates of Clav. lusitaniae BSIs ranging from 0.9% in Australian medical centres to 2% in European hospitals and 3.2% in Beijing [16, 53, 54, 55]. Despite showing low MICs to all antifungal agents tested, both in vitro and clinical data suggest that Clavispora lusitaniae should not be considered a suitable target for amphotericin B therapy [25, 50, 56, 57].

Wickerhamomyces anomalus has emerged as an opportunistic pathogen among neonates and low‐birth‐weight infants [58]. This species was responsible for 0.6% of all BSIs evaluated in our study, a finding similar to data recently reported in medical centres from Europe (0.5%) and Beijing (0.8%) [41, 55]. Regarding the antifungal susceptibility of our collection, the MIC₅₀ and MIC₉₀ values for fluconazole and voriconazole were 4/4 μg/mL and 0.06/0.125 μg/mL, respectively. Some authors have already reported that W. anomalus may have limited in vitro susceptibility to fluconazole and voriconazole [59, 60, 61, 62].

The emerging multidrug‐resistant Candida haemulonii species complex has been increasingly reported, particularly in Latin America [15, 63, 64]. In Brazil, Lima et al. [15] observed a rise in prevalence from 0.9% to 1.7% over 11 years, while rates reported in European medical centres are still low (0.3%) [41]. In the present series, we identified an increase of Candida haemulonii species complex BSIs from 0.3% in PI to 0.6% in PII. Particularly concerning are the MIC₅₀ and MIC₉₀ values obtained with Candida duobushaemulonii isolates when tested against amphotericin B (both were > 8 μg/mL) and fluconazole (8/16 μg/mL). Indeed, other authors already reported that Candida duobushaemulonii should not be considered a target for fluconazole and amphotericin B therapy [15, 63, 64, 65].

To conclude, this study provides valuable insights into the prevalence rates, species distribution and antifungal susceptibility of rare yeasts causing BSIs in tertiary care hospitals. Our findings highlight the need for continuous surveillance of fungal agents causing BSIs, the incorporation of new diagnostic and tailored therapeutic strategies to mitigate morbidity and mortality due to invasive infections caused by emerging pathogens.

Conflicts of Interest

A.L.C. has received educational grants from Knight‐United Medical, Gilead, Mundipharma, Sandoz, Accord, IMMY. The other authors declare no conflicts of interest.

Supporting information

Table S1: Distribution of 100 episodes of bloodstream infections caused by rare Saccharomycotina yeasts documented across 28 medical centres in 12 Brazilian states between 2007 and 2023.

Table S2: Acession number code from the DNA sequencing that have been deposited in the GenBank database (https://www.ncbi.nlm.nih.gov/genbank/).

Data Availability Statement

All sequence data have been deposited in the GenBank sequence database, and the corresponding accession numbers are provided in Table S2.