Community-Onset Fungemias: Epidemiology and Genomic Characterization

1. Introduction

Community-onset fungemia characterized by a positive fungal blood culture obtained within 48 hours of hospital admission, is an important clinical condition linked with significant morbidity and mortality [1,2].

According to some studies, community-onset cases represent approximately one-third of all candidemia episodes [1,3]. Patients with community-onset fungemia frequently have recent healthcare exposure, with studies reporting that up to 75% were hospitalized in the prior three months [1]. Delayed antifungal therapy (median 2.7 days) contributes to a high 30-day mortality (~26%) [1]. By contrast, classic nosocomial candidemia (arising >48 h into admission) typically involves the ICU and invasive devices [4,5]. Distinguishing community-onset cases highlights the need to recognize at-risk patients early.

Diagnostic delays remain a significant limitation. Blood cultures, currently considered the standard method, are slow and fail to detect a substantial proportion of cases, with positivity rates reported between 21% and 71% in proven invasive candidiasis [6]. Even when cultures yield growth, several additional days are usually required for species identification and antifungal susceptibility testing, further postponing timely optimization of therapy.

Candida is the leading genus responsible for fungal bloodstream infections [7,8,9] When these infections are caused by Candida spp., they are defined as candidemia [10,11].

The distribution of Candida species responsible for candidemia varies across geographic regions and hospital settings. These differences reflect not only institutional factors but also patient-related conditions, such as underlying diseases, prior antifungal exposure, and other risk factors that shape local epidemiology. According to recent surveillance data, five species (C. albicans, C. glabrata, C. tropicalis, C. parapsilosis, and C. krusei) account for the vast majority of candidemia cases, collectively representing more than 90% of isolates [12,13,14,15].

Recent taxonomic revisions, enabled by advances in molecular biology, have reclassified several clinical Candida species into new genera. For example, Candida glabrata is now Nakaseomyces glabratus, C. krusei is Pichia kudriavzevii, and C. lusitaniae is Clavispora lusitaniae. This shift in nomenclature introduces ambiguity in the term “candidemia” as not all causative species now formally belong to the Candida genus.

In terms of susceptibility, community onset isolates generally reflect the species profile, with most remaining susceptible to first-line antifungals, although non-albicans species often show reduced susceptibility to azoles. Echinocandin resistance remains uncommon in both hospital- and community-acquired isolates. Whole-genome sequencing (WGS) is reshaping the capacity to detect antifungal resistance and accurately delineate fungal species, enabling precise identification of cryptic species together with the simultaneous detection of resistance mutations. Compared with traditional methods, WGS provides faster and more scalable analysis and has contributed to uncovering both known and novel genetic determinants of resistance [16].

Our study therefore aims to describe the epidemiological, clinical, and molecular characteristics of yeast bloodstream infections diagnosed over 21 years at a tertiary hospital. WGS was applied to provide precise species identification and to characterize antifungal resistance mechanisms.

2. Materials and Methods

2.1. Clinical and Epidemiological Data

A retrospective cohort study was conducted, encompassing all patients diagnosed with fungemia (restricted to pathogenic yeasts including Candida species and reclassified genera such as Nakaseomyces, Clavispora, and Pichia, while excluding Cryptococcus spp.) Relevant clinical and epidemiological data, including patient demographics, underlying conditions, and outcomes, were collected from medical records. Cases were identified via the hospital’s microbiology and admissions databases, and data were compiled for analysis with appropriate confidentiality safeguards. The study received approval from the Ethics Committee of the hospital (IIBSP-WIS-2022-144).

Clinical variables were defined as follows:

• Diabetes mellitus: Yes/No (documented history of diabetes)
• Chronic kidney disease: Yes/No (established diagnosis of chronic renal disease)
• Recent surgery: Yes/No (surgical procedure within the previous three months; if yes, specify localization: abdominal, neurosurgical, or joint…)
• Haematological cancer: Yes/No (history of haematological malignancy)
• Neutropenia: yes/no (absolute neutrophil count below 500 cells/µL)
• Haematological transplantation: yes/no (prior haematopoietic stem-cell transplant)
• Solid tumour: Yes/No (diagnosis of solid-organ malignancy such as carcinomas, sarcomas, or other non-haematological cancers)
• Corticosteroid therapy: Yes/No (administration of corticosteroids within 30 days prior to diagnosis)
• Chemotherapy: Yes/No (receipt of chemotherapy within 30 days prior to diagnosis)
• Other immunosuppressive therapy: Yes/No (use of non-steroidal, non-chemotherapeutic immunosuppressive agents such as calcineurin inhibitors, antimetabolites, mTOR inhibitors, Janus kinase inhibitors, or monoclonal antibodies such as anti-TNF or anti-CD20 within 30 days prior to diagnosis)
• Recent antibiotic use: Yes/No (administration of antibiotics within 30 days prior to diagnosis)
• Recent antifungal use: Yes/No (administration of antifungal agents within 30 days prior to diagnosis)
• Central venous catheter: Yes/No (presence of a central venous catheter at the time of diagnosis)
• Urinary catheter: Yes/No (presence of an indwelling urinary catheter, such as a Foley or bladder catheter/vesical catheter, at the time of diagnosis)
• Haemodialysis catheter: Yes/No (presence of a vascular catheter specifically for haemodialysis)
• Healthcare-associated fungemia: Yes/No (diagnosis within 48 h of admission in patients with recent healthcare contact: hospitalization within 3 months, surgery, dialysis, or residence in a long-term care facility)
• Community-acquired fungemia: Yes/No (diagnosis within 48 h of admission in patients without recent healthcare exposure)
• Source of fungemia: (first assessed clinically; considered positive if the same microorganism is isolated at the suspected focus. Sources are classified as “unknown” when neither compatible clinical features nor microbiological confirmation at a focus are present. Possible sources include unknown, urological, abdominal, catheter-related, cardiovascular, joint, gynaecological, or cutaneous)
• Mixed Fungemia: Yes/No (presence of more than one microorganism isolated in the same blood culture; if yes, classified as either multiple yeasts or yeast plus bacteria)
• Empirical treatment: Yes/No (whether the patient received antimicrobial therapy before microbiological evidence of fungemia; if yes, specify if only antibiotic, only antifungal, or both in combination)
• Persistent fungemia: Yes/No (positive blood cultures for yeast despite appropriate antifungal therapy)
• Recent antifungal use: Yes/No Administration of antifungal agents within 30 days prior to diagnosis)
• Mortality at 7 days: Yes/No (death occurring within 7 days of the fungemia diagnosis)
• Mortality at 30 days: Yes/No (death occurring within 30 days of the fungemia diagnosis)

2.2. Identification and Antifungal Susceptibility

Viable yeast isolates obtained from blood cultures were preserved by lyophilization, with four vials stored per isolate for long-term maintenance. For initial recovery, a lyophilized vial of each isolate was rehydrated in sterile saline and streaked onto ChromAgar Candida plates (CHROMagar™, Paris, France), followed by incubation at 35–37 °C for 48 hours. This chromogenic medium supports robust yeast growth and produces distinctive colony colours for presumptive identification of different Candida species after 48 hours. If an isolate failed to grow after two attempts on ChromAgar, an alternative recovery method was employed: the lyophilized vial was rehydrated in sterile saline and then inoculated into an aerobic blood culture bottle (BacT/ALERT VIRTUO system, bioMérieux, Marcy-l’Étoile, France) and incubated until flagged as positive. Once the blood culture bottle was flagged as positive, its contents were examined by Gram staining to verify the presence of yeast cells. Subsequently, an aliquot was subcultured onto ChromAgar Candida to recover and isolate yeast colonies. All recovered yeast isolates, whether obtained via direct plating or blood culture enrichment, were maintained with serial subcultures on ChromAgar Candida to ensure purity and viability. Subcultures were performed every 48–72 hours (two sequential passages) prior to identification and susceptibility testing. Species-level identification of the yeasts was achieved by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) using MALDI Biotyper® sirius GP System (Bruker Daltonik, Germany). Antifungal susceptibility testing was carried out by broth microdilution using the Sensititre YeastOne YO10 colorimetric panel (Thermo Fisher Scientific, Waltham, MA, USA), endpoints were interpreted according to Clinical and Laboratory Standards Institute (CLSI) species-specific breakpoints (M27 guidelines [17]). For Candida orthopsilosis (a member of the C. parapsilosis species complex lacking its own breakpoints), the CLSI recommends applying the breakpoints established for C. parapsilosis.

2.3. Molecular Analysis by Whole-Genome Sequencing (WGS)

2.3.1. DNA Extraction

Genomic DNA was extracted from recovered isolates to enable molecular analyses. Extractions were performed using the DNeasy UltraClean Microbial Kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol. The concentration of each DNA prep was assessed with a Qubit 4 fluorometer (Invitrogen, Thermo Fisher Scientific).

2.3.2. Next-Generation Sequencing (NGS)

Purified DNA extracts were sent to Novogene Europe (Germany) for next-generation sequencing using Illumina NocaSeq 6000 sequencer. Paired-end reads were generated and delivered as FASTQ files for downstream analysis.

2.3.3. Bioinformatic Analysis

Raw paired-end reads were quality trimmed using fastp [18] and quality checked using FastQC (available at https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and MultiQC [19]. Trimmed reads were then assembled into contigs using SKESA [20] wrapped in shovill (available at https://github.com/tseemann/shovill).

Taxonomic identification was performed using GAMBIT [21]. To ensure correct taxonomic classification, 28S and ITS sequences were extracted from FASTA files using BLASTn (v2.16.0+) [22], and taxonomic identification was then performed against the NCBI RefSeq Targeted Loci Project rRNA/ITS databases via the online BLASTn suite.

We used ChroQueTas [23], an open-source tool for rapid screening of fungal genomes for antimicrobial resistance (AMR) mutations, with the FungAMR database.

The combined use of automated tools and manual curation ensured comprehensive identification of both known and novel resistance mutations.

2.4. Data Analysis

Clinical and laboratory data were analysed using R (v4.4.3) in RStudio. Descriptive statistics summarized patient characteristics. Categorical data were shown as counts and percentages; continuous data as medians with IQRs. Visualizations were produced with ggplot2.

3. Results

3.1. Epidemiology and Clinical

Annual incidence ranged from one to five cases without clear trends. A total of 48 cases were included in the study. The median age of patients was 74 years (interquartile range [IQR]: 63–82 years), and 56.3% were male (Table 1).

Prevalent clinical risk factors included recent antibiotic use within the previous 30 days (21 cases, 43.8%), presence of a central venous catheter (11 cases, 22.9%), history of solid tumours (15 cases, 31.2%), diabetes mellitus (15 cases, 31.2%), and recent abdominal surgery (9 cases, 18.8%). Additional factors included chemotherapy (10 cases, 20.8%), chronic kidney disease (eight cases, 16.7%), corticosteroid treatment (eight cases, 16.7%), haematological cancer (four cases, 8.3%), neutropenia (three cases, 6.3%), recent immunosuppressive therapy (three cases, 6.3%), recent hematopoietic transplant (two cases, 4.2%), persistent fungemia (four cases, 8.3%), and antifungal treatment within 30 days (two cases, 4.2%)

Regarding the setting of acquisition, 10 cases (20.8%) were strictly community-acquired and 38 (79.2%) were healthcare-associated. The most frequent sources of fungemia included unknown origin (18 cases, 37.5%), urological infections (11 cases, 22.9), abdominal infections (10 cases, 20.8%), and catheter-related bloodstream infections (seven cases, 14.6%). Additionally, one case each was attributed to a cardiovascular and a joint focus.

Empirical antifungal therapy was administered in only one case (2.1%), receiving antibiotic and fluconazole. Empirical antibacterial therapy alone was prescribed in 40 cases (83.3%), while seven patients (14.6%) received no empirical treatment. Mortality was 20.8% (10 cases) at 7 days post-diagnosis and 29.2% (14 cases) at 30 days.

Monomicrobial fungemia was present in 39 cases (81.2%), while nine cases (18.8%) were polymicrobial: seven involved concurrent bacteraemia and two included multiple Candida species. Among these, one case involved two different Candida spp. (C. albicans and C. parapsilosis) species and another three distinct yeasts (C. albicans, C. tropicalis and N. glabratus). In total, 51 isolates were recovered from the 48 patients.

3.2. Identification and Antifungal Susceptibility Test

The most frequently isolated species was C. albicans (21 isolates, 41.2%), followed by N. glabratus (14 isolates, 27.5%), C. parapsilosis (six isolates, 11.8%), C. tropicalis (six isolates, 11.8%), and P. kudriavzevii (two isolates, 3.9%). Single isolates were identified as C. lusitaniae and C. orthopsilosis (1.9% each).

No isolates showed resistance to anidulafungin, micafungin, or amphotericin B. Two N. glabratus isolates had reduced susceptibility to caspofungin (MIC = 0.25 μg/mL). Azole resistance was detected in two cases: one C. albicans isolate (voriconazole MIC > 16 μg/mL; fluconazole MIC > 256 μg/mL) and one N. glabratus isolate (voriconazole MIC > 16 μg/mL; fluconazole MIC = 128 μg/mL). P. kudriavzevii is intrinsically resistant to fluconazole (Figure 1).

3.3. WGS

Of the 51 isolates recovered, 45 were successfully sequenced. Among them, 19 corresponded to C. albicans (17 with mutations), 13 to N. glabratus (all with mutations), four to C. parapsilosis (one mutated), five to C. tropicalis (one mutated), two to P. kudriavzevii (without mutations), and the single isolates of C. lusitaniae and C. orthopsilosis (both without mutations). WGS could not be performed on the remaining six isolates due to insufficient DNA quantity or quality. No discrepancies were observed between MALDI-TOF identification and WGS results in the sequenced isolates.

The 13 sequenced N. glabratus isolates carried the mutations I207V in ERG2, E839D in PDR15, and V91I and L98S in PDR1. Table 3 summarizes all mutations previously described for N. glabratus identified in our study. Of the 18 previously reported mutations in N. glabratus, all were detected in susceptible strains except for the F659C substitution in FKS2, which was present only in the caspofungin-intermediate strain. The detailed summary of resistance-related genes analysed, and key mutations identified by WGS is provided in the Supplementary Material. Table 2 provides a summary of all identified mutations that had been previously reported for N. glabratus.

Of the 18 previously reported mutations in N. glabratus, all were detected in susceptible strains except F659C, which was found only in the caspofungin-intermediate strain. Two N. glabratus isolates were of clinical interest due to their resistance phenotypes. One caspofungin-intermediate isolate carried mutations in the glucan synthase genes FKS2 as well as FKS3 (R1039L, N1825S). The other, one fluconazole-resistant isolate, showed alterations in efflux pump regulators (PDR1 S76P, V91I, L98S, T143P, G857V; CDR1 H58Y; PDH1/PDR15 E839D) and in ergosterol pathway genes (ERG2 I207V, ERG8 N448S). The only mutation unique to this isolate was the substitution G857V in PDR1. Table 3 summarizes the novel mutations identified in N. glabratus.

In the case of C. albicans, 16 previously reported mutations were detected, all of which were neutral. Similarly, in C. parapsilosis, another neutral mutation was identified in ERG11 (R398I) in two fluconazole-susceptible isolates. Table 4.

The only one C. albicans, fluconazole-resistant showed multiple mutations affecting transcriptional regulators CDR1, MRR2 and the ergosterol biosynthetic enzyme ERG11. Some of these mutations appeared to be unique to this strain (Table 5).

4. Discussion

Our study provides an integrated clinical, epidemiological, and genomic analysis of community-onset candidemia cases diagnosed over a 21-year period. Only a limited number of studies have examined community-onset candidemia [1,2]. These infections are more frequently nosocomial in origin, and most cases that appear to originate in the community are in fact linked to recent healthcare exposure. Our data are consistent with those reported in the literature, with most cases being healthcare-associated [1].

The clinical profile of our cohort reflects the risk factors most frequently described in the literature for candidemia [34], including recent antibiotic exposure [35,36], central venous catheters [37,38], malignancy [39], and recent surgery [40]. However, in our series no single factor exceeded one-third of the cases, which makes clinical suspicion difficult and often hampers the timely initiation of appropriate empirical antifungal therapy.

Several clinical prediction scores have been proposed to estimate the risk of candidemia in hospitalized patients, including the Candida Score [41] and the Ostrosky-Zeichner rule [42]. These tools incorporate factors such as severe sepsis, recent surgery, central venous catheter, total parenteral nutrition, and broad-spectrum antibiotic use. They show a high negative predictive value, making them useful for ruling out infection, although their positive predictive value is low [43,44]. While they have been primarily validated in ICU populations, their components overlap with the risk factors observed in our cohort, suggesting that they could aid in the early identification of patients at higher risk of candidemia. However, no score has yet been specifically validated for community-onset cases.

All-cause mortality in our series was high, at 20.8% at 7 days and 29.2% at 30 days after diagnosis. These figures are consistent with previous reports in community-onset candidemia cohorts and reflect the severity of this condition, although it remains infrequent [45,46,47].

Species distribution varies depending on age, underlying disease, or geographic region [48,49,50], with C. albicans being the most frequent species in Europe. In our cohort, after C. albicans the second most frequent species was by N. glabratus and C. parapsilosis consistent with recent European multicentre study [51].

Echinocandins are the recommended first-line therapy for candidemia, given their fungicidal activity and favourable safety profile [52,53]. In our study we did not observe echinocandin resistance, aligning with contemporary surveillance showing low echinocandin resistance [54,55].

Echinocandin resistance in N. glabratus, echinocandin resistance arises from amino-acid substitutions in conserved hotspot regions of FKS1 and especially FKS2, which encode subunits of the 1,3-β-D-glucan synthase complex. One caspofungin-intermediate N. glabratus harboured the hotspot substitution FKS2 F659C, which is well established to confer reduced susceptibility to echinocandins [26,56,57]. By contrast, FKS3 is expressed at least 100-fold lower than FKS1, is non-essential for vegetative growth, and its deletion does not change echinocandin MICs, indicating no established role in clinical echinocandin resistance [58].

In N. glabratus, azole resistance is most commonly mediated by gain-of-function mutations in the transcription factor PDR1, which upregulates the expression of ATP-binding cassette (ABC) efflux pump genes such as CDR1, PDH1 (CDR2), and SNQ2 [59–61]. These mutations promote overexpression of efflux pumps, leading to reduced intracellular azole accumulation and cross-resistance.

Our fluconazole-resistant N. glabratus isolate carried previously reported substitutions in PDR1 (S76P, V91I, L98S, T143P) [56], as well as CDR1 (H58Y) [23,54,55] and PDH1/PDR15 (E839D) [23,54], all of which are considered polymorphisms not known to confer azole resistance. The only mutation unique to this isolate was PDR1 G857V, which has not been previously reported and whose functional impact remains unknown.

Although mutations in ergosterol biosynthesis genes such as ERG2 or ERG8 have occasionally been reported to affect azole susceptibility, their role in resistance is less well defined [51,57]. N. glabratus can adapt to azole stress by altering sterol composition, including uptake of exogenous sterols when ergosterol biosynthesis is impaired, which may limit the impact of such mutations. Our isolate also harboured polymorphisms in ERG2 (I207V) [23] and ERG8 (N448S) [56], which are known not to confer azole resistance.

In C. albicans, azole resistance is commonly mediated by gain-of-function mutations in the transcription factor MRR1, which activates MDR1 expression, leading to efflux-mediated resistance. Additional regulators such as MRR2 can modulate CDR1 expression and contribute to azole resistance [30,62]. Our fluconazole-resistant C. albicans isolate harboured multiple mutations in efflux regulators (MRR1: E1020Q, V27del, PQS166 insertion, L248V, E336del, V340E; MRR2: L144V, T145A, S165N, S480P, I204del, S580del), the target enzyme ERG11 (Q142del), and the efflux pump CDR1 (T365del, Q790del, T947S, E949P, N1499stop, K1500del, K1501del). However, most CDR1 changes, including C-terminal deletions, have been reported across isolates with varying azole susceptibility and are not considered resistance-associated Similarly, ERG11 Q142del lies outside recognised hotspots (e.g. Y132/K143) and alone does not predict resistance [29,63].

Taken together, the combination of MRR1/MRR2 alterations with the ERG11 indel (Q142del) provides a plausible genetic basis for the fluconazole-resistant phenotype in CAC_937, whereas the functional contribution of the individual CDR1 changes (including N1499stop/K1500–K1501del) remains uncertain and would require targeted experimental validation.

This 21-year single-centre cohort provides an integrated clinical, epidemiological, and genomic view of community-onset fungemia. Most episodes were healthcare-associated, with C. albicans as the leading species, followed by N. glabratus. Mortality remained high (29.2% at 30 days). While echinocandin resistance was absent and azole resistance uncommon, WGS confirmed species identification, revealed both known and novel resistance-associated mutations.

The study’s strengths lie in its two-decade surveillance, integration of clinical and genomic data, and identification of novel resistance markers. However, limitations include its retrospective, single-centre design, modest sample size, and lack of functional validation of mutations. Despite these constraints, the findings underscore the value of WGS in resistance surveillance and the ongoing clinical burden of community-onset fungemia, supporting the need for multicentre studies to confirm and expand these results.