DNA sequencing technologies have helped to identify the urinary microbiome.1 By using new technologies such as 16S ribosomal RNA (rRNA) gene sequencing, prior undetected urine bacterial taxa have been revealed. Dysbiosis of an individual’s microbiome may be associated with pathologic states.2–6 It has been hypothesized that the urinary microbiome plays a role in many disease states, such as urinary tract infections (UTIs), urinary incontinence and bladder cancer.7–13 The majority of urinary microbiome studies have focused on adult subjects and adult pathology, including urgency, urinary incontinence, stress urinary incontinence, bladder cancer and prostate cancer.7–13 The studies that have focused on pediatric patients have involved small numbers, older children, and the collection process has been via swabs of the perineal area not via catheterization.14–16
In the non-toilet trained pediatric population, the gold standard of practice is to perform a sterile urethral catheterization to obtain a urine culture (UC) in a patient where a UTI is suspected, as a bagged specimen can be contaminated.17 The most common clinical indicator is fever with no obvious source in those less than 24 months of age.17,18 By utilizing catheterized specimens, the goal was to collect a sample with the least chance of contamination from the urethra, foreskin or perineal area and without deviating from standard care. To understand the role of the urinary microbiome in diseased states, it is important to know how the urinary microbiome develops in early childhood and thus what may constitute a healthy microbiome. The purpose of this project was to create a better understanding of a “normal” urinary microbiome in the pediatric population and how it shifted in a diseased state.
In children less than 48 months of age, urine samples were collected via sterile technique by transurethral catheterization during their Emergency Department (ED) visit. All subjects were evaluated by a pediatric emergency physician, who, following standard of care, would clinically determine if a urine catheterization was needed for the evaluation of a UTI.17 Once the urine sample was obtained, the amount necessary was extracted for the clinical tests, urinalysis and UC, while the remainder was used for study purposes. A minimum 1 mL was reserved for 16S rRNA gene sequencing and was stored at −80°C. Informed consent was obtained by at least one parent or guardian (Institutional Review Board approval no. 16-2257, Inova Health System).
The results of urinalysis and UC were recorded. A UTI was defined by growth of >50,000 colony forming units (CFU) per mL by culture with pyuria.17,18 UCs with growth >1000 (CFU) were also noted. Hematuria was defined as greater than 5 red blood cells per high power field.19
Inclusion criteria included pediatric patients who required a catheterized urine sample in the pediatric ED and were less than 48 months of age. Exclusion criteria included patients with severe global developmental delay or multi-system problems, neurogenic bladders or other disease process preventing them from normal voiding, sickle-cell trait or disease, nephrotic or nephritic syndrome, immunocompromised patients, including patients taking high-dose steroids and patients with renal transplant.
Detailed demographic and clinical information were collected for each subject, including age, gender, race, maternal ethnicity and country of origin, reason for ED visit, antibiotic use within the last 3 months, child probiotic use, circumcision status and delivery mode (cesarean section versus vaginal delivery). A χ2 or Fisher exact test was used to examine differences in clinical and demographic data between genders as appropriate.
Collected urine samples were stored at −80°C for up to 1 year before DNA extraction and analysis. Prior to DNA extraction, urine samples were thawed and centrifuged. The supernatants were discarded, and the pellets were re-suspended in 400 μL of Buffer ATL (Qiagen, CA) and treated with 25 μL of lysozyme solution (50 mg/mL) (Sigma Aldridge, MO). Samples were heated for 5 minutes at 70°C and cooled before loading on the EZ1 Advanced (Qiagen) for DNA extraction by using the EZ1 DSP Virus kit (Qiagen). Samples were then cleaned and concentrated using the DNeasy PowerClean Cleanup Kit (Qiagen). Sequencing was prepared using a Nextera XT kit according to the Illumina 16S Metagenomic Sequencing Library Preparation protocol for analysis of hypervariable regions V3-V4. The locus-specific sequences (Illumina, CA) using standard IUPAC nucleotide nomenclature were 16S Amplicon polymerase chain reaction (PCR) forward primer = 5′- GACTACHVGGGTATCTAATCC-3′, 16S amplicon PCR reverse primer = 5′- CCTACGGGNGGCWGCAG-3′. For normalization, each library was quantified with the KAPA Library Quantification Kit (Kapa Biosystems, MA). The libraries were sequenced on the Illumina MiSeq with paired-end 300 bp reads.
A negative control sample was included on each plate; ATL buffer was used from the DNA extraction through the entire process until qPCR. The qPCR analysis of the negative controls showed that there was no contamination from common potential known contaminant steps: cross contamination from other samples, contamination from experimenter and contamination from the DNA extraction kit. Two positive controls of Staphylococcus aureus (Strain NCTC 8532, ATCC, VA) and Escherichia coli (Strain NCTC 9001, ATCC, VA) were included and sequenced.
Sequence reads from the urine samples and the 2 positive control samples were trimmed, filtered and processed in the 16S rDNA analysis pipeline implemented in the QIIME 1.9 tool suite.20 First, reads were quality checked by the program FASTQC (version: 0.11.7).21 They were then trimmed and filtered using the program trimmomatic (version: 0.36).22 In the following QIIME pipeline, the open-reference operational taxonomic unit (OTU) picking strategy with the UCLUST algorithm as the search engine was employed. Reads were searched and clustered against the Greengenes v13.8 reference database into OTUs, and those reads which had any hit with at least 97% similarity were subsequently clustered into de novo OTUs. Taxonomy was assigned based on a collective of the representative sequences picked from each OTU using the RDP classifier, and the representative sequences were further aligned by PyNAST and built phylogeny tree using FastTree.
Both of the OTU table and the phylogenetic tree files were then imported into R packages phyloseq23 (version 1.19.1) and vegan (version 2.4-3) for the subsequent analysis. OTUs that were not seen more than three times in at least 20% of the samples were removed from downstream analyses.24
The raw count data was rarefied (using the function rarefy_even_depth from the R phyloseq package) and then to calculate alpha diversity measures (ie, observed species, Shannon, Simpson, Fisher). The raw count data were also normalized as relative abundance CPM (count per million) by the function transform_sample_counts from the R phyloseq package, and then to calculate beta diversity measures (unweighted UniFrac Distance, weighted UniFrac Distance, Bray-Curtis and Jensen-Shannon divergence). Both alpha and beta diversities were evaluated with the collected clinical variables, including gender, ages, delivery mode, maternal ethnicity, antibiotic use and probiotic use.
To examine relative abundance of bacteria of interest, only the OTUs with a relative abundance CPM of at least 1000 in at least one sample were investigated. The cutoff score of 1000 was set empirically due to the low read counts in some samples. The lowest count number is 1965 in this study, and an OTU with 2 reads in the sample had a relative abundance CPM score of 1017. Therefore, this cutoff can effectively exclude singletons (ie, rare OTUs).
The sequencing data from our study was submitted to the NCBI BioProject database (https://www.ncbi.nlm.nih.gov/bioproject) under accession number PRJNA433896.
Demographic and clinical information of the subjects were summarized using the R package, whereas χ2 test was applied for categorical variables by default (with continuity correction). The function 1-way test from the R package stats was used for remaining tests in the summary table. In the analysis of alpha diversity, linear regression was applied to check the association between each alpha diversity and each demographic/clinical variable, and the P-values were adjusted by the method of Benjamini and Hochberg. The adonis function from the vegan package was used in the regression analysis of beta diversity, and the P-values were also adjusted by the method of Benjamini and Hochberg.
Eighty-five subjects were included in our study. Table 1 showed demographic and clinical data for subjects. There were more females enrolled than males (59 vs. 26, respectively). More subjects less than 24 months of age were enrolled compared with greater than 24 months of age (72 vs. 13, respectively), and males were significantly younger (mean age females = 436 days, mean age males = 260 days, t test P = 0.006). The most common reason for the ED visit was fever. The population was diverse with many different maternal ethnicities and countries of origin represented; 48 of 85 mothers identified themselves as Hispanic. A total of 9 patients had UTIs; all UTIs were with E. coli, 2 being multidrug resistant. The 2 males with UTIs were uncircumcised. Eight patients had a UC that had bacterial growth >1000 CFU/mL but did not meet the definition of a UTI. Four subjects had growth of 1000–9000 CFU/mL (3 with growth of normal urogenital flora, one with growth of Proteus mirabilis), 2 had growth of 10,000–30,000 CFU/mL (both with growth of E. coli) and one had growth of 30,000–50,000 CFU/mL (with growth of P. mirabilis). In addition, one subject had growth of 50,000–100,000 CFU/mL but with growth of normal urogenital flora and without pyuria so did not meet the criteria for UTI. Nine patients were currently receiving probiotics. A total of 28 patients had received antibiotics in the past 3 months before the urine sample was obtained. Of those 28 patients, 15 had received amoxicillin.
Overall Microbiome Statistics
A urinary microbiome, 16S rRNA gene sequencing from bacteria, was found in all samples: the read counts ranged between 1965 and 63,010, with the median 28,980. Overall, there were 25 genera having a relative abundance over 10% in at least one sample, which collectively originated from 22 families, 16 orders, 11 classes and 6 different phyla. The median of the relative abundances of the OTUs was summarized in Figure S1, Supplemental Digital Content 1, http://links.lww.com/INF/D787. The median was used instead of the mean because in this study read counts were not evenly distributed. Extreme values (outliers) did not affect the median as strongly as they did the mean.
Shannon diversity index was significantly decreased (t test, P < 0.001) in those with a UTI compared with those without a UTI, Figure 1. In addition, there was a trend of decreased Shannon diversity index with increasing CFU/mL growth in urine specimens (R2 = 0.3879, P < 0.001, linear regression); in (Figure S2, Supplemental Digital Content 2, http://links.lww.com/INF/D788). A similar trend was found with Simpson index but not Fisher index or observed species. Alpha diversity (Shannon index and Simpson index) was also decreased in those with hematuria compared with those without hematuria (t test, P < 0.001); however, 5/9 patients with a UTI also had hematuria and only 6 patients had hematuria alone. Alpha diversity (Fisher and observed) was significantly decreased (P=0.017) in patients who had used antibiotics in the past 2 weeks before urine collection compared with those who had not used antibiotics, Figure 2, (n = 7, for those taking antibiotics within 2 weeks). The Shannon index and Simpson index, however, did not significantly change with recent antibiotic use. Alpha diversity did not differ by age, gender, antibiotic use 15 days to three months before the urine sample was obtained, maternal ethnicity, country of origin, delivery mode or probiotic use.
When examining the composition of the microbiome, subjects with a UTI clustered separately from those without a UTI (Fig. 3A; Adonis test on Bray-Curtis distance, P = 0.001). There was no difference between those with some growth of bacteria in the UC that did not meet criteria for a UTI and those without a UTI, (Figure S3, Supplemental Digital Content 3, http://links.lww.com/INF/D789). Those with hematuria also significantly clustered separately from those without hematuria (Fig. 3B; Adonis test on Bray-Curtis distance, P = 0.001. A significant difference was found between antibiotic use (P = 0.012) (Figure S4, Supplemental Digital Content 4, http://links.lww.com/INF/D790) but not among age, gender, maternal ethnicity or country of origin, delivery mode or probiotic use.
Relative Abundance of Taxa
The most abundant phyla seen in urine samples were firmicutes, proteobacteria and bacteroidetes, which accounted for median abundances of 33.1%, 28.5% and 13.3%, respectively. The 5 most abundant classes were clostridia, bacteroidia, gammaproteobacteria, actinobacteria and betaproteobacteria. The 5 most abundant order were clostridiales, bacteroidales, enterobacteriales, burkholderiales and actinomycetales. The 5 most abundant family were tissierellaceae, prevotellaeae, veillonellaceae, enterobacteriaceae and comamonadaceae, and the 5 most abundant genera were Prevotella, Peptoniphilus, Escherichia, Veillonella and Finegoldia. This was represented in (Figure S1, Supplemental Digital Content 1, http://links.lww.com/INF/D787).
Figure 4 showed an abundance plot at the order level of all samples correlating with clinical and demographic features. There was an increase in abundance of the genus Escherichia in the 9 patients with UTIs (t test, false discovery rate (FDR)-adjusted P < 0.001) (Figure S5, Supplemental Digital Content 5, http://links.lww.com/INF/D791); the abundance of the order enterobacteriales was over 99% in all 9 UTI samples except one (from the subject S4; Figure S4, Supplemental Digital Content 4, http://links.lww.com/INF/D790). The relative abundance of Shigella was also increased in those with UTIs (FDR-adjusted P < 0.001). When looking at the relative abundance of the samples that had some growth of bacteria in the UC that did not meet criteria for a UTI, there was a heterogeneous pattern among samples of the most abundant taxa.
In those who had antibiotic use within the past 2 weeks there was a significant increase in the relative abundance of genus Achromobacter (P < 0.001) from the family Alcaligenaceae (P < 0.001). Of the group having received antibiotics in the past 2 weeks (N = 7), 71.4% of subjects have a relative abundance of Achromobacter above 0.1%, this is compared with the group with no antibiotic use (N = 66), where only 24.2% of subjects have a relative abundance of Achromobacter above 0.1%. The phylum Actinobacteria was more commonly found in boys (P = 0.005) compared with girls (Figure S6, Supplemental Digital Content 6, http://links.lww.com/INF/D792). The only taxa associated with age was the relative abundance of Mobiluncus. A strong correlation was detected between age and the abundance of Mobiluncus (linear regression, FDR-adjusted P < 0.001). Further analysis suggested the correlation was mainly driven by the female subgroup [females (r = 0.52, P < 0.001), males (r = −0.032, P = 0.88)], (Figure S7, Supplemental Digital Content 7, http://links.lww.com/INF/D793). The relative abundance of taxa did not differ by maternal ethnicity or country of origin, delivery mode, probiotic use or circumcision status.
Several differences were observed in our study when compared with adult studies of the urinary microbiome or microbiome studies of other systems, such as the gastrointestinal tract.25–29 The only observed variation in diversity or abundance with age occurred in females with Mobiluncus and in gender with the phylum Actinobacteria seen in males, but no changes were observed with mode of delivery or probiotic use. The 5 most abundant genera in our study differed from the most frequently reported in past studies.1 The lack of significant gender differences observed is possibly because of the use of catheterized samples, which minimized peri-urethral and perineal contamination, and may contribute to the gender differences seen in previous studies where voided samples were used. In addition, post-pubertal hormonal differences and sexual activity may contribute to gender and age differences seen later in life. Longitudinal studies would be needed to observe the change with age.
A diverse microbiome has been associated with being protective and with a healthy status. When a UTI occurred, there was decreased diversity, and changes in the composition of the microbiome. This has been observed in other systems.30,31 In addition, any growth of bacteria in UCs, even when not meeting criteria for a UTI, was associated with decreased alpha diversity. If changes in the diversity of the urinary microbiome could be identified prior to UTI development, this could lead to identification of a high-risk microbiome or an at-risk population. Future therapeutic interventions could then be targeted toward early identification and prevention of clinical UTIs.
Antibiotics have had a profound effect on various microbiomes including the gastrointestinal system.32 One-third of the subjects in our study received antibiotics in the prior 3 months; amoxicillin was the most commonly used antibiotic. In our study, there was a significant decrease in alpha diversity between those who had been given antibiotics within 2 weeks of the urine sample being obtained compared with those who were not. However, the Shannon entropy and Simpson index did not significantly change with recent antibiotic use, suggesting that antibiotics reduce the species richness but do not significantly impact species evenness. As there was no difference observed later than 2 weeks this may suggest a relatively quick recovery of the urinary microbiome after antibiotic use. These differences may reflect distinct properties of the urinary microbiome and how it is affected by antibiotics.
There were multiple limitations in this study. The total volume of urine per subject was small. However, ethically, it would have been difficult to perform multiple catheterizations per subject to increase volume. Urine samples also have a low abundance of bacteria. As with any low abundance microbiomes, there was potential for contamination.33 To minimize this potential, positive and negative controls were included, and positive controls were reported in the same way as the urine samples. Although hematuria was found to cause changes to the microbiome, as multiple subjects also had a concomitant UTI and the hematuria may be from microtrauma, this cannot be considered definitive. Total number of subjects remained a limiting factor to investigate all variables adequately. Additionally, all children in our study were being seen for urgent reasons and so were not all necessarily in a “healthy” state.
A urinary microbiome was observed in very young children, even in 3 subjects less than 30 days of age. Changes in microbiome diversity and composition were observed in subjects with a standard culture positive UTI. Further longitudinal exploration of when these microbiome changes occur in relationship to the development of a UTI is warranted. Additionally, future studies that focus on potential variables that may influence the urinary microbiome, even including a comparison of mother and child samples, could create a better understanding of what impacts the urinary microbiome. The urinary microbiome has just begun to be explored, especially in the pediatric population, and the implications it has on long-term disease processes deserves further investigation.
The authors would like to thank Martin Tafazoli and Dennis Ponce Quintanilla, LPN for their considerable help with data collection.
1. Aragón IM, Herrera-Imbroda B, Queipo-Ortuño MI, et al. The urinary tract microbiome
in health and disease. Eur Urol Focus. 2018;4:128–138.
2. Cuthbertson L, Craven V, Bingle L, et al. The impact of persistent bacterial bronchitis on the pulmonary microbiome
of children. PLoS One. 2017;12:e0190075.
3. Lynch SV, Pedersen O. The human intestinal microbiome
in health and disease. N Engl J Med. 2016;375:2369–2379.
4. Slattery J, MacFabe DF, Frye RE. The significance of the enteric microbiome
on the development of childhood disease: a review of prebiotic and probiotic therapies in disorders of childhood. Clin Med Insights Pediatr. 2016;10:91–107.
5. Sverrild A, Kiilerich P, Brejnrod A, et al. Asthma and lower airway disease: Eosinophilic airway inflammation in asthmatic patients is associated with an altered airway microbiome
. J Allergy Clin Immunol. 2017;140:407–417.e11. doi: 10.1016/j.jaci.2016.10.046.
6. Buttó LF, Haller D. Dysbiosis in intestinal inflammation: cause or consequence. Int J Med Microbiol. 2016;306:302–309.
7. Cortes-Penfield NW, Trautner BW, Jump RLP. Urinary tract infection and asymptomatic bacteriuria in older adults. Infect Dis Clin North Am. 2017;31:673–688.
8. Thomas-White KJ, Kliethermes S, Rickey L, et al.; National Institute of Diabetes and Digestive and Kidney Diseases Urinary Incontinence Treatment Network. Evaluation of the urinary microbiota of women with uncomplicated stress urinary incontinence. Am J Obstet Gynecol. 2017;216:55.e1–55.e16.
9. Karstens L, Asquith M, Davin S, et al. Does the urinary microbiome
play a role in urgency urinary incontinence and its severity? Front Cell Infect Microbiol. 2016;6:78.
10. Bao Y, Razvi H, Gloor G, Reid G, Denstedt J, Burton J. Mp88-19 urinary microbiome
patterns appears to be correlative to intravesical recurrence of non-muscle invasive bladder cancer. J Urol. 2016;195:e1136.
11. Pearce MM, Hilt EE, Rosenfeld AB, et al. The female urinary microbiome
: a comparison of women with and without urgency urinary incontinence. mBio. 2014;5:e01283–e01214.
12. Pearce MM, Zilliox MJ, Rosenfeld AB, et al.; Pelvic Floor Disorders Network. The female urinary microbiome
in urgency urinary incontinence. Am J Obstet Gynecol. 2015;213:347.e1–347.11.
13. Shrestha E, White JR, Yu SH, et al. Profiling the urinary microbiome
in men with positive versus negative biopsies for prostate cancer. J Urol. 2018;199:161–171.
14. Duel B, Dru C, Funari V. Pediatrics: urinary tract infection & vesicoureteral relux: MP44-19 Preliminary analysis of the vulvar microbiome
in prepubertal girls. J Urol. 2014;191:e449–e450.
15. Gorbachinsky I, Sherertz R, Russell G, et al. Altered perineal microbiome
is associated with vulvovaginitis and urinary tract infection in preadolescent girls. Ther Adv Urol. 2014;6:224–229.
16. Ching C, Lucas E, Dabdoub S, Kumar P, Justice S. Infections/inflammation/cystic disease of the genitourinary tract: Kidney & bladder I: pd12-03 normal perineal microbiome
in prepubertal females with dysbiosis if recurrent urinary tract infections. J Urol. 2017;197:e266.
17. Reaffirmation of AAP clinical practice guideline: the diagnosis and management of the initial urinary tract infection in febrile infants and young children 2-24 months of age. Pediatrics. 2016;138:e20163026.
18. Finnell SM, Carroll A, Downs SM. Diagnosis and management of an initial UTI in febrile infants and young children. Pediatrics. 2011;128e749–e770.
19. Higashihara E, Nishiyama T, Horie S, et al.; Working Group for the Creation of Hematuria Guidelines. Hematuria: definition and screening test methods. Int J Urol. 2008;15:281–284.
20. Caporaso JG, Kuczynski J, Stombaugh J, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods. 2010;7:335–336.
21. Andrews S. FastQC: A Quality Control Tool for High Throughput Sequence Data. Available at: http://www.bioinformatics.babraham.ac.uk/projects/fastqc
22. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for illumina sequence data. Bioinformatics. 2014;30:2114–2120.
23. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome
census data. PLoS One. 2013;8:e61217.
24. Wong WSW, Clemency N, Klein E, et al. Collection of non-meconium stool on fecal occult blood cards is an effective method for fecal microbiota studies in infants. Microbiome
25. Yatsunenko T, Rey FE, Manary MJ, et al. Human gut microbiome
viewed across age and geography. Nature. 2012;486:222–227.
26. Salminen S, Gibson GR, McCartney AL, et al. Influence of mode of delivery on gut microbiota composition in seven year old children. Gut. 2004;53:1388–1389.
27. Dominguez-Bello MG, Costello EK, Contreras M, et al. Delivery mode shapes the acquisition and structure of the initial microbiota across multiple body habitats in newborns. Proc Natl Acad Sci U S A. 2010;107:11971–11975.
28. Fouts Derrick E, Rembert P, Sebastian S, et al. Integrated next-generation sequencing of 16S rDNA and metaproteomics differentiate the healthy urine microbiome
from asymptomatic bacteriuria in neuropathic bladder associated with spinal cord injury. J Transl Med. 2012;10:174.
29. Debbie AL, eBrown R, eWilliams J, et al. The human urinary microbiome
; bacterial DNA in voided urine of asymptomatic adults. Front Cell Infect Microbiol. 2013;3.
30. Hourigan SK, Chen LA, Grigoryan Z, et al. Microbiome
changes associated with sustained eradication of Clostridium
difficile after single faecal microbiota transplantation in children with and without inflammatory bowel disease. Aliment Pharmacol Ther. 2015;42:741–752.
31. Skraban J, Dzeroski S, Zenko B, et al. Gut microbiota patterns associated with colonization of different Clostridium
difficile ribotypes. PLoS One. 2013;8:e58005.
32. Dethlefsen L, Huse S, Mitchell LS, David AR. The pervasive effects of an antibiotic on the human gut microbiota, as revealed by deep 16S rRNA sequencing. PLoS Biology. 2008;6:e280.
33. Bao Y, Al KF, Chanyi RM, et al. Questions and challenges associated with studying the microbiome
of the urinary tract. Ann Transl Med. 2017;5:33.