This retrospective study included pediatric patients (aged 0 - 17 years) admitted to a tertiary pediatric hospital with 372 beds in southern Brazil. All the patients were diagnosed with candidemia between August 2016 and August 2022. During the study period, 191 candidemia episodes were identified. After applying the exclusion criteria, 141 episodes remained for the final analysis. Supplementary Figure 1 shows the flow diagram of screened cases, exclusions with reasons, and the final study population.

PC was diagnosed when the positive blood cultures were still obtained at ≥ 5 days after starting antifungal treatment [4]. Patients who had a positive blood culture and did not receive antifungal treatment or prophylaxis were excluded as persistence could not be assessed. Patients who died on the same day as blood culture collection were included to assess early mortality potentially related to candidemia. Crude mortality was defined as death within 30 days of a positive blood culture, whereas mortality attributed to candidemia was defined as death within 7 days of a positive blood culture [8].

Clinical and demographic data, including age, sex, hospitalization unit, previous surgery, associated comorbidities, treatment, persistence of infection, and clinical outcome information, were obtained from electronic medical records. This study was approved by the Institutional Review Board (IRB #2.096.359) of the participating hospital in accordance with the guidelines established for the protection of participants.

Statistical analyses were performed using R statistical software version 4.3.2 [9]. Measures of central tendency (mean and median) and dispersion (standard deviation) were used for descriptive analysis of the sample. Pearson's chi-square and Fisher’s exact tests were used to determine the association between categorical variables. Comparisons between independent groups were performed using Student's t-test or Analysis of Variance (ANOVA). To identify independent predictors of persistent candidemia, a multivariable logistic regression analysis was performed. All variables described in Table 1 were included as candidate variables in a stepwise selection process, and only those showing statistical significance (p < 0.05) were retained in the final model. Results were expressed as odds ratios (OR) with their respective 95% confidence intervals.

All Candida spp. clinical isolates were stored in skim milk at −80°C until use. They were cultured on Sabouraud Dextrose Agar to assess viability and purity, and identified by MALDI-TOF MS (Microflex LT Biotyper 3.0, Bruker Daltonics, Bremen, Germany) according to the manufacturer’s instructions. Biofilm formation was evaluated as previously described [10].

Antifungal susceptibility testing was carried out using the broth microdilution technique for fluconazole (0.125-64 µg/mL), voriconazole (0.016-1.0 µg/mL), amphotericin B (0.063-8.0 µg/mL), and micafungin (0.0078-8.0 µg/mL) according to the European Committee for Antimicrobial Susceptibility Testing (EUCAST). To ensure test quality, two reference strains, C. parapsilosis ATCC 22019 and C. krusei ATCC 6258, were included as controls for each antifungal susceptibility test. The minimum-inhibitory-concentration (MIC) results were interpreted according to the EUCAST breakpoints [11].

Genomic DNA was extracted using a previously described in-house protocol [12]. MLST was performed for C. albicans and C. tropicalis. Of 43 C. albicans isolates, 42 were viable and typed using seven housekeeping genes (PubMLST database; accessed 10 Feb 2025). Twenty-four of 25 C. tropicalis isolates were typed using six housekeeping genes (species database; accessed 10 Feb 2025).

PCR was conducted as previously described [13,14]. Amplicons were analyzed by 1.5% agarose gel electrophoresis, purified with ExoSAP-IT, and sequenced on an ABI 3500 (Applied Biosystems, USA) using the same primers. Electropherograms, allele assignment, and diploid sequence type (DST) determination were performed as previously described [2].

The microsatellite technique was used to genotype all the isolates of C. parapsilosis stricto sensu by amplifying eight polymorphic microsatellite markers [15]. The reference C. parapsilosis ATCC 22019 was used to compare the allelic profiles of the isolates. Determination of the allelic profiles and analysis of the clonal isolates were performed as previously described [10].

A total of 141 candidemia episodes were included. The study population comprised predominantly young children with a median age of 1 year (range, 0–17 years). Specifically, 80.2% (n = 113) were aged 0–5 years, and 40.4% (n = 57) were infants younger than 13 months. Detailed baseline characteristics are shown in Table 1.

Demographic characteristics and hospitalization units were similar between the PC and non-PC groups. However, significant risk factors for PC included CVC non-removal within 5 days of treatment initiation (p = 0.005), parenteral nutrition use (p = 0.033), and hematologic or solid malignancies (p = 0.042). Neurological disease was more frequent in the non-PC group (p = 0.02) (Table 1).

In an exploratory analysis of patients whose catheters were removed within 5 days (n = 64), logistic regression showed no significant association between each additional day of catheter retention and persistent candidemia (p = 0.34). Instead, failure to remove the catheter within 5 days was the main factor associated with persistence.

Antifungal therapy was initiated within 5 days of the positive blood culture in 137 of 141 patients (97.2%). Three patients (2.1%) did not receive treatment because they died on the day of blood culture collection, and one patient (0.7%) started therapy after 8 days. Among treated patients (n = 137), 119 (86.9%) received an adequate antifungal dose. Patients with persistent candidemia required a significantly longer treatment duration compared to those without persistence (p = 0.00015).

Regarding the clinical outcomes of the patients, the overall mortality at 30 days was 23.4% (n = 33), and the candidemia-related mortality was 16.3% (n = 23).

In addition to the significant factors in the univariate analysis, the results of the multivariate analysis assessing independent predictors of persistent candidemia are presented in Table 2.

C. parapsilosis stricto sensu(n = 52, 39%) was the most prevalent species in the candidemia cases studied, followed by C. albicans (n = 43, 30.5%) and C. tropicalis (n = 25, 17.7%). Other candidemia-causing species identified in this study included C. lusitaniae (n = 5), Wickerhamomyces anomalus (n = 4), Pichia kudriavzevii (n = 2), C. orthopsilosis (n = 2), C. famata (n = 2), C. guilliermondii (n = 1), Nakaseomyces glabratus (n = 1), C. haemulonii (n = 1), C. dubliniensis (n = 1), C. metapsilosis (n = 1), and C. utilis (n = 1).

Regarding the 49 isolates identified in patients with PC, the distribution of the prevalent isolates remained the same: C. parapsilosis stricto sensu (n = 20, 40.8%), C. albicans (n = 12, 24.5%), and C. tropicalis (n = 9, 18.4%). Other species included C. lusitaniae (n = 2), W. anomalus (n = 2), Pichia kudriavzevii (n = 1), C. guilliermondii (n = 1), C. metapsilosis (n = 1), and C. utilis (n = 1).

There were no significant differences between patients with PC and without PC regarding Candida species.

Among the 141 isolates, 137 were included in the biofilm assay, as four were not viable for testing. Biofilms were produced by 136 (99.3%) of 137 isolates tested. Of these, 108 (78.8%) were classified as strong biofilm producers (Supplementary Fig. 2).

In total, 124 isolates, including C. parapsilosis stricto sensu, C. albicans, C. tropicalis, P. kudriavzeveii, and N. glabratus, were tested for antifungal sensitivity using EUCAST species-specific clinical breakpoints. The remaining isolates were excluded from susceptibility analysis due to the lack of clinical breakpoints for interpretation. Except for one isolate of C. tropicalis resistant to fluconazole (MIC = 8 μg/mL), all the others were sensitive to the antifungals tested. Supplementary Table 1 shows the MIC50 and MIC90 values of the four antifungal drugs tested against the most important species enrolled in this study.

A total of 42 C. albicans isolates were viable for molecular typing. According to the PubMLST database, all C. albicans isolates presented with previously unreported DSTs, and no isolates with the same allelic profile were found. Regarding C. tropicalis, 24 isolates were tested, 18 were classified as new and unique DSTs, and another six isolates were classified as belonging to previously described DSTs: 7, 124, 232, and 238. DST124 was detected in more than one isolate, indicating the presence of clonal profiles (Supplementary Tables 2 and 3).

C. parapsilosis sensu stricto isolates were typed using microsatellites to assess genetic similarity. Of 51 isolates, 47 distinct non-clonal genotypes were identified, while seven clonal isolates formed three clusters. Figure 1 shows allelic profiles, including C. parapsilosis ATCC 22019, and the distribution of clonal isolates. Genetic relationships among the 47 genotypes, their hospital distribution, and year of isolation are depicted in a minimum spanning tree (Figure 2).

Figure 1.

Allelic profile of C. parapsilosis stricto sensu isolates grouped in dendogram by genetic similarity. The lines indicate the number (id) of isolates. The black square indicates the presence of an amplification product. The columns correspond to an allele of the respective polymorphic microsatellite marker (8 gene loci). Isolates with the same allelic profile (clones) are highlighted.

Figure 2.

Minimum spanning tree of the relationship between the 47 genotypes among the 51 isolates of C. parapsilosis sensu stricto isolated from the bloodstream. Each circle represents a unique genotype, while circle size corresponds to the number of isolates of the specific genotype and the color of the isolates represents the year of isolation (A) or hospital setting. ICU, intensive care unit; CICU, cardiac intensive care unit; NICU, neonatal intensive care unit; GICU, general intensive care unit; Non-ICU, Other clinical units; HO, hematology-oncology; SICU, surgical intensive care unit; BMT, bone marrow transplant.