Uromycobiome in infants and toddlers with and without urinary tract infections

Abstract

Background:

The bacterial components of the urobiome have been described in children, both with and without urinary tract infections (UTI). However, less is known about the pediatric uromycobiome: the community of fungi in the urine. The objectives of this study were to describe the uromycobiome in children and determine whether the uromycobiome differs between children with and without UTI.

Methods:

This was a cross-sectional study of febrile children less than three years of age who presented to the Emergency Department and had a catheterized urine sample sent as part of clinical care. We obtained residual urine for use in this study and identified components of the uromyobiome through amplification and sequencing of the fungal ITS2 region. We then compared the uromycobiome between those with and without UTI.

Results:

We included 374 children in this study (UTI=50, no UTI =324). Fungi were isolated from urine samples of 310 (83%) children. Fungi were identified in a higher proportion of children with UTI compared to those without UTI (96% vs 81%, p=0.01). Shannon diversity index was higher in children with a UTI compared to those without (p=0.04). Although there were differences in the most abundant taxa between children with and without UTI, there was no significant difference in beta diversity between groups.

Conclusions:

Fungi was detected in the majority of catheterized urine samples from children. While a higher proportion of children with UTI had fungi in their urine compared to children without UTI, there was no difference in the composition of these groups.

Introduction

Urine, which, until recently, was considered to be sterile, is now known to contain a large variety of resident bacteria.[1–5] However, there is limited literature on the presence of fungi in the urine. Outside of the urinary tract, the mycobiome, the communities of fungi within a specific ecological niche, has been shown to play a role in a variety of conditions in the gastrointestinal tract, including obesity,[6] Crohn’s disease,[7] and pancreatic cancer.[8] As interactions between different types of microbes, such as bacteria and fungi, likely modulate homeostasis,[9,10] sole focus on the bacterial components of the urobiome may lead to an incomplete view of the larger microbial ecosystem. Indeed, there is evidence of interactions between components of the mycobiome and resident bacteria in both the GI tract as well as breast milk,[7,11] supporting the need to consider the fungal components of the urobiome (uromycobiome).

While the uromycobiome has been described in adult women,[12,13] there are no reports of the uromycobiome in children. The objectives of this study were to identify the uromycobiome in a relatively large cohort of children, and to determine if it differs between children with and without urinary tract infections (UTI).

Methods

Enrollment for this study took place between June 2019 and May 2020 in the Emergency Department of a free-standing, urban children’s hospital. Children were eligible for enrollment if they met the following criteria: age between 1 month and 3 years, febrile (either documented temperature of ≥38° C in the emergency department or by parental report) within 24 hours of presentation, and had a urine sample collected by a catheter. All patients had a urinalysis and urine culture. Pyuria was defined as any of the following: ≧ 5 urine white blood cells per high powered field, ≧ 10 urine white blood cells/mm³, or ≧1+ leukocyte esterase. Exclusion criteria included: receipt of systemic antibiotics or corticosteroids within 3 days prior to enrollment, presence of other concurrent systemic bacterial infections, immunodeficiency, and presence of a neurogenic bladder or other significant genitourinary anomalies. Institutional review board approval was obtained for this study with waiver of informed consent given the use of residual urine samples in this work. Data on bacterial abundance will be reported separately (Shaikh, under review).

Urine sample processing

Residual catheterized urine samples were placed in in a cryovial and frozen at −80° C without preservatives. Before shipping, we added 70 microliters of Urine Conditioning Buffer (Zymo Research, cat. D3061-1-8) to every 1 mL of the frozen urine sample.

Mycobiome analysis:

We followed methods for mycobiome analysis that have been previously described.[14] Briefly, we used the PrecisionBIOME^™ Next Generation Microbial Test (Pangea Laboratory, Tustin, CA, USA), a next generation DNA sequencing (NGS) test, to analyze specimens in this work. We used the ZymoBIOMICS 96 MagBead DNA Kit (Lysis Tubes) to extract DNA from all specimens (Zymo Research Corporation, Irvine, CA). We then prepared libraries using the PrecisionBIOME^™ Library Prep Kit (Zymo Research Corporation, Irvine, CA), and used the MiniSeq Sequencing System (Illumina, San Diego, CA) to sequence barcoded amplicons. We use the PrecisionBIOME^™ bioinformatics pipeline for bioinformatics analysis. This is a validated pipeline that enables species-level identification of fungal sequences using amplicon-sequence-variant (ASV). We used primer sequences that targeted the ITS2 region, as previously described. [15] We included both positive and negative controls in the NGS workflow. The negative controls included both transport medium alone and unused swabs, as well as controls for any other possible sources of contamination (i.e., sequencing buffers, equipment, extraction process, library preparation). The positive controls were cell and DNA mock communities (ZymoBIOMICS Microbial Community Standard, Zymo Research Corp.). We used the DADA2 pipeline (R package version 3.4) to remove potential sequencing errors and chimeric sequences.[16] We then trimmed reads and merged paired-end reads. We computed phylotypes as percent proportions based on the total number of sequences in each sample and used Fungal DNA Quantification Kit (Zymo Research Corporation, Irvine, CA) to determine the absolute abundance of fungi.

Statistical analyses

Results were expressed as means with standard deviations, unless otherwise stated. We calculated α- diversity and evenness using the Shannon index and the number of observed species. We used Bray-Curtis distance, a standard method within microbiome analysis to quantify how different two samples are while adjusting for the number of different organisms, at the species taxonomic level to calculate β-diversity.

We used the Kruskal Wallis test to compare non-normally distributed variables. We used linear discriminant analysis (LDA) and effect size (LEfSe) to identify taxa that were significantly enriched in either the UTI or no UTI group using the default settings, and considered a p-value <0.05 as significant(QIIME version 1.9.1,).[17] We controlled for type 1 errors on the species-level relative abundance data using analyses of variance and false discovery rate. We compared beta diversity between children with and without UTI using PERMANOVA. GraphPad Prism (version 8) software was used for data visualization.

Results

We collected 374 samples, 50 from children with UTI and 324 in children without UTI. A uromycobiome was identified in 310 (83%) children. Children with a uromycobiome present were younger than those without a uromycobiome (mean (standard deviation): 12.3 (8.5) vs 15.1 (8.7) months, p=0.01). There was no difference in highest temperature, gender, or proportion of children with pyuria between children with and without a uromycobiome (Table 1). A uromycobiome was detected in 96% (48/50) of children with UTI vs 81% (262/324) of children without UTI (p = 0.01).

Table 1:

Demographics and Urinalysis Results from Children with and without a Uromycobiome

	Fungus Present (n=310)	No Fungus Present (n=64)	P-value
Mean age (months)	12.3 (8.5)	15.1 (8.7)	0.01
Mean highest temperature (C°)	39.3 (0.8)	39.3 (0.8)	0.99
UTI	48 (15.5)	2 (3.1)	0.01
Female	227 (73.2)	51 (79.7)	0.36
LE ≧1	49 (15.8)	4 (6.3)	0.10
Pyuria	58 (18.7)	6 (9.4)	0.10

LE: Leukocyte esterase; UTI: urinary tract infection

There was no difference in age between children with and without UTI (14.0±8.2 months, 12.6±8.6 months, p=0.29) or duration of fever prior to ED visit (1.9±1.8 days for children with UTI, 1.5±2.3 days for children without UTI, p=0.52). The most common bacteria that was cultured from urine from children with UTI was Escherichia coli (92%).

A total of 270 fungal species were detected using NGS. There were 15 identified species with a relative abundance greater than 1% within the 310 samples (Figure 1, supplemental table 1). The four most abundant species in the UTI group were Stereum complicatum (15.0%), Trametes versicolor (7.6%), Candida parapsilosis (6.3%) and Cladosporium halptplerans-sphaerospermum (6.3%). The most common identified fungi in the no UTI group were Stereum complicatum (17.5%), Cladosporium sp. (5.5%), Malassezia restricta (4.9%) and Trametes versicolor (3.8%). The four most common identifiable fungal species that differed between UTI and no UTI included: Candida parapsilosis (6.3% UTI, 0.6% no UTI); Cladosporium halotolerans-sphaerospermum (6.3% UTI, 2.2% no UTI); Fungi sp. (4.5% UTI, 2.3% no UTI) and Cantharellales sp. (2.1% UTI, 0% no UTI). Children with UTI had a higher total fungal cell count compared to those without UTI (p<0.01).

Figure 1.

Average relative abundance of components of the uromycobiome between children with and without a urinary tract infection

When the microbiome of any given sample type is analyzed, common measures are the alpha and beta diversity of the organisms identified (Figure 2 and 3). These types of measurements provide a broad overview of the microbial diversity in a given sample set as they indicate how many different types of microbes are present in a sample, how evenly distributed the species present are, and how closely related the species are to one another. Alpha diversity looks at these parameters within a sample, while beta diversity looks at the degree of similarity in community composition between two different samples. Among the alpha diversity parameters measured, Shannon diversity index was higher in children with UTI compared to those without (p=0.04). (Figure 2). There was no difference in the beta diversity between children with and without UTI (p=0.30). (Figure 3).

Figure 2.

Alpha diversity analysis of the sample set as measured as (A) Richness, (B) Chao1, (C) Evenness, and (D) Shannon Diversity Index of the uromycobiome of children with and without a UTI. Panel A shows richness, which is an unadjusted (or “raw”) count of the number of taxa within a sample. Panel B shows Chao1, a parameter that is often used in sample with low biomass such as urine, is an adjusted measure that estimates abundance of taxa. Panel C shows evenness, which reflects whether or not each taxa is equally represented in a sample. Panel D shows the Shannon Diversity Index, a measure that is used to quantify how many different taxa are within a sample.

Figure 3.

Beta diversity of the uromycobiome of children with and without UTI

Discussion

To our knowledge, this is both the first report describing the uromycobiome in children as well as the first report comparing the uromycobiome of individuals with and without UTI. We found that the majority of samples in our cohort did have detectable fungi, and that younger children and those with a UTI were more likely to have detectable fungi. While we also found a difference in alpha diversity and in the types of fungi in the urine of children with and without UTI, we did not find a difference in beta-diversity between those with and without a UTI.

It is notable that we detected a uromycobiome in 83% of all samples. This is similar to reports of the pediatric bacterial urobiome, in which 90.5% of included children have a urobiome.[4] There were some differences in demographics between children with and without a uromycobiome: children with a detectable uromycobiome were younger than those without a uromycobiome, suggesting that age may be an important consideration for future studies focused on the uromycobiome. Indeed, the bacterial components of the pediatric urobiome do change with age,[4] supporting our hypothesis that age does affect the uromycobiome. However, we did not find a difference in sex between children with and without a uromycobiome, indicating that, in pre-pubescent children, the presence of a uromycobiome is likely not associated with sex

The number of ASVs and relative abundance of some fungal species differed significantly between children with and without UTI. The reasons for this are unclear. As this is a cross-sectional study, we are unable to determine if the changes in the uromycobiome occurred acutely in the setting of a UTI, or if the differential uromycobiome could potentially be associated with increased risk of UTI. It is possible that there are relevant fungal-bacterial interactions that may be associated with UTI pathogenesis. Indeed, there are fungal-bacterial interactions that are medically important. For example, presence of Candida sp. in the respiratory tract is associated with an increased rate of ventilator-associated pneumonia due to Pseudomonas aeruginosa.[18] There are also data suggesting interactions between Candida species and E. coli.[19–21] Within the genitourinary tract, the presence of Lactobacillus sp. in the vagina demonstrate fungicidal activity against Candida spp.[22] Therefore, while understanding the bacterial-fungal interactions within the setting of UTI remains to be elucidated, there is certainly precedence for such an interaction.

While there is a paucity of data on the pediatric uromycobiome, there are data on the mycobiomes of other sites in children. One study found that the infant oral, skin, and anal mycobiomes consists predominantly of the follow taxa: Candida parapsilosis, C. tropicalis, Saccharomyces cerevisiae, C. albicans, C. orthopsilosis, Cryptococcus pseudolongus, Cladosporium velox, Debaryomyces renaii, D. hansenii, Hanseniaspora uvarum, and Issatchenkia orientalis.[23] Others have reported that the gastrointestinal (GI) mycobiome of consists predominantly consists of Candida albicans, Rhodotorula mucilaginous, Candida parapsilosis, Torulaspora delbrueckii and Pichia fermentans.[24] Interestingly, none of the top seven taxa in our study were identified in the GI mycobiome in one study of 111 healthy volunteers, which included both infants and children.[24] Given the evidence that bacteria in the GI microbiome is found within the bladder,[25] the lack of concordance between the uromycobiome and GI mycobiome is surprising. To the best of our knowledge, there are no reports of Stereum complicatum, the most abundant taxa in our UTI group, in humans. However, there is some overlap in our results and report of various human mycobiomes. Cladosporium sp., which were abundant in our samples, have been identified in the oral mycobiome of healthy individuals.[26] It is also interesting to note that Candida parapsilosis, which was the species that differed most between those with and without UTI and a reported component of the GI mycobiome,[24] is an emerging nosocomial pathogen that mostly affects the immunosuppressed and neonatal intensive care unit populations.[27]

In the few studies of the adult uromycobiome, the presence of both Agaricomycetes, the class to which Stereum complicatum belongs, as well as Dothideomycetes, which is the class to which Cladosporium belongs have been reported. Indeed, Dothideomycetes was the most common class identified in the clean-catch, mid-stream samples from adults.[13] Other work, which reports the uromycobiome in adult women with interstitial cystitis, does have some overlap in identified taxa in our work, including Cladosporium, Malassezia, and Candida.[12]

Interestingly, Candida parapsilosis, a fungal species significantly more abundant in children with UTI in our data, is increasingly recognized as a urogenital pathogen causing UTIs in animal models and humans.[28,29] This may have further implications for the patients as this fungus is recognized for its increasing resistance to antifungal agents.[29] The data presented here further highlights the need for more studies investigating the clinical significance of fungi in urine and their potential role in UTIs.

There are several limitations to this study, including that it is a single-center study from children being evaluated in an Emergency Department for suspected UTI. Therefore, the results may not be generalizable to other settings or patient populations. Although none of the children had known abnormalities of the kidney or urinary tract at the time of this study, it is possible that some children were later diagnosed with such an abnormality, which may have affected our results. While we excluded children who received antibiotics in the three days before their ED visit, we did not obtain data about earlier antibiotic exposure. It is possible that more remote antibiotic exposure may have affected our results. Further, urine is a low biomass fluid, and thus has an increased risk of contamination affecting the results.[30] Additionally, although all urine samples were obtained by catheterization, it is possible that sample contamination may affect our results. Although we also included all relevant precautions during sample processing to avoid contamination and included negative controls as quality control, contamination during sample processing is still a possibility. Another limitation is that the majority of the uromycobiome identified in the UTI group are unspecified, meaning that they could not be identified by mapping to currently available databases. Although it is notable that unspecified fungi predominate in the UTI group and not in the non-UTI group, this lack of information limits our ability to draw conclusions about the UTI group. Future studies may consider using whole genome sequencing to further identify the fungal species present beyond the ITS genome sequence. The inability to identify the specific components on the uromycobiome in UTI may also have affected the alpha diversity results, which need to be validated in future studies when more components of the uromycobiome can be identified. Finally, total cell counts may be affected by the life cycle of fungi that can significantly impact copy number.

In this pilot study, fungi were detected in the majority of catheterized urine samples from children. While a higher proportion of children with UTI had fungi in their urine compared to children without UTI, there was no difference in the beta diversity between these groups. Future work will focus on elucidating the stability of the uromycobiome, whether a specific uromycobiome can increase risk of UTI, and the interactions that occur between the bacterial and fungal components of the urobiome.