Intravesical Lactobacillus rhamnosus GG Alters Urobiome Composition and Diversity Among People With Neurogenic Lower Urinary Tract Dysfunction

Abstract

Background

Neurogenic bladder is associated with bacterial colonization and frequent urinary tract infections.

Objectives

To explore the effects of one to two doses of intravesical Lactobacillus rhamnosus GG (LGG) on the urobiomes of adults with spinal cord injury/disease (SCI/D) who manage their bladders with intermittent catheterization (IC).

Methods

This was a pilot substudy within an 18-month phase 1 clinical trial of self-instilled intravesical LGG for urinary symptoms as directed by the Self-Management Protocol using Probiotics (SMP-Pro). Urine samples were collected monthly when participants were asymptomatic. When SMP-Pro “trigger” symptoms (cloudier and/or more foul-smelling urine) occurred, urine samples were collected immediately pre-LGG instillation and 24 to 48 hours after LGG instillation. Urine was collected via a new catheter, immediately placed on ice/freezer, and processed within 12 hours. Genomic DNA was isolated, and the V4 region of the 16S rRNA bacterial gene was amplified and high throughput sequenced. Amplicon sequence variants were inferred and bacterial composition, community structure, and variation across clinical phenotypes were determined.

Results

126 urine samples were collected from 26 participants (SCI/D = 23; multiple sclerosis = 2; spina bifida = 1) between 20 and 57 years of age. The urobiomes were characterized by four dominant phyla (>1%): Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria, which were comprised of six dominant genera (>3%): Escherichia/Shigella (29.1%), Klebsiella (22.4%), Proteus (15.2%), Aerococcus (6.3%), Streptococcus (6.0%), and Pluralibacter (3.0%). Post-LGG samples were associated with a decline in Escherichia/Shigella predominance (p < .001) and altered bacterial diversity (p < .05).

Conclusion

Among people with SCI/D who use IC, intravesical LGG alters the bacterial composition and diversity of the urine ecosystem, potentially disrupting the uropathogenic urobiome.

Introduction

Advances in our understanding of the human microbiome and the role that microbiota play are altering our contextualization of human health and disease. Human microbial communities in various body niches have been described to differing degrees; however, descriptions of the human urobiome are fairly limited to date. While it has been widely accepted for years that neurogenic lower urinary tract dysfunction (NLUTD) and catheterization among people with spinal cord injury/disease (SCI/D) is associated with microbial colonization and symptomatic infection,¹ the 2012 discovery using molecular-based methods that a healthy urobiome exists and that urinary health is not synonymous with sterility of the urine^2,3 was transformative. Prior to that time, healthy urine was considered to be free of bacteria, and infection was conceptualized as due to growth and invasion of a single microbial organism. Currently, evidence indicates that the urobiome varies by gender,^4,5 age,⁶ and menopausal status,⁷ and associations between urine dysbiosis and urinary tract disease states (such as urinary incontinence and bladder cancer)⁸ and neurogenic dysfunction^2,9 are beginning to emerge.

Among adults with NLUTD, our group has shown that the NLUTD urobiome is characterized by a uropathogenic urotype, with predominance of Escherichia coli, Enterococcus faecalis, Pseudomonas aeruginosa, and Klebsiella pneumoniae even in the absence of symptoms. Among children with NLUTD who used intermittent catheterization (IC), the urobiome has a similar uropathogenic urotype, predominated by Enterococcus faecalis, Proteus mirabilis, and Klebsiella pneumonia, regardless of symptom status.⁹ In contrast, Lactobacillus, Streptococcus, Prevotella, and Veillonella species predominate in asymptomatic non-NLUTD control urobiomes,^10,11 a finding confirmed among healthy controls of both sexes.¹² Mechanisms for the observed NLUTD-related dysbiosis are likely translocation of bacteria from the gut and direct inoculation of bacteria into the urinary bladder¹³ by the multiple daily insertions of a urinary catheter and the permanent placement of an indwelling urethral catheter required to manage urinary retention. This is supported by Fouts and Groah's findings, demonstrating significant differences in the NLUTD urobiome by bladder management method (void, IC, indwelling catheterization).² Other factors related to NLUTD, such as bladder wall ischemia, mucosal disruption from bladder wall distention, poor compliance, bladder wall trabeculation, immunodeficiency, defective glucosaminoglycan layer, defective apoptosis, and impaired washout of bacteria, may also contribute to dysbiosis.^13,14

We and others have demonstrated the predominance of Lactobacillus species in the urine of asymptomatic individuals with normally functioning bladders, shedding light on urinary eubiosis.^2,3 In this study, we aim to determine whether intravesical instillation of Lactobacillus rhamnosus GG (LGG) can alter NLUTD-associated urinary dysbiosis. We conducted a first-in-human, 18-month prospective single-arm quasi-experimental pre-post phase 1 clinical trial of self-instillation of intravesical LGG in which we demonstrated that one to two LGG instillations are safe, well-tolerated,¹⁵ and associated with a reduction in urinary symptoms.^15-17 In this pilot project, we describe the effects of intravesical LGG on the urobiomes of a subset of the adults with NLUTD who manage their bladders with IC from this trial. We hypothesize that self-instillation of intravesical LGG will result in a measurable change in the bacterial composition and diversity of the NLUTD urobiome.

Methods

All study activities were conducted after Investigational New Drug (IND 16306) approval by the US Food and Drug Administration (FDA) of the therapeutic intervention and after institutional review board review and approval (#2014-211). This is a pilot substudy of an 18-month prospective single-arm quasi-experimental pre-post phase 1 clinical trial (ClinicalTrials.gov NCT02748356) of self-instilled intravesical LGG mixed with normal saline in response to urinary symptoms as directed by the Self-Management Protocol using Probiotics (SMP-Pro).^15,17 The study consisted of three 6-month phases: baseline, intervention, and washout. Participants with NLUTD due to SCI/D, inclusive of multiple sclerosis (MS) and spina bifida (SB), who were living in the community reported urinary symptoms weekly using the Urinary Symptom Questionnaire for Neurogenic Bladder-Intermittent Catheterization version (USQNB-IC).^18,19 During the 6-month intervention phase, participants followed the SMP-Pro to guide intravesical self-instillation of LGG in response to “trigger” urinary symptoms that do not meet the criteria for UTI according to the Infectious Diseases Society of America (IDSA)²⁰: cloudy urine or bad-smelling, stronger, fouler or more pungent smelling urine. Study procedures, including use of the SMP-Pro, are described elsewhere.^15,17,21 This study reports on a subset of participants who provided multiple urine samples for 16s sequencing.

Phenotypes.

Baseline (asymptomatic), pre-LGG (symptomatic with cloudy and/or malodorous urine), and post-LGG (all but one remained symptomatic) phenotypes (according to USQNB-IC reporting) are compared.

Timing of urine collection.

Urine samples were collected monthly during the baseline and intervention phases when participants were asymptomatic according to USQNB-IC reporting. When SMP-Pro “trigger” symptoms occurred during the intervention phase, a pre-LGG instillation urine sample was collected, and then the participant followed the SMP-Pro to instill LGG. Within 48 hours of LGG instillation, a post-LGG instillation urine sample was collected. Thus, the pre-instillation sampling occurred at the time of symptoms (cloudier/foul-smelling urine report by the participant) and prior to LGG instillation, and the post-instillation sample was collected within 48 hours after the instillation occurred (and for all but one, prior to symptom resolution). During the washout phase of the study, an additional asymptomatic urine sample was collected.

Urine collection and preparation.

For all urine collections, a new catheter and a sterile cup were used. The urine was immediately placed on wet ice for transport to a freezer, where it was then placed at 4°C and processed within 12 hours for sequencing. For 16S sequencing preparation, samples were centrifuged at 4°C, 5,000 x g for 20 minutes. The supernatant was aliquoted in 2 mL cryotubes and frozen at −20°C. Ten milliliters PBS was added to the pellet with the remaining supernatant and then centrifuged (5,000xg) at 4°C for 20 minutes. The pellets and aspirated PBS wash solution were stored at −20°C, and pellets were later transferred for DNA isolation and 16S rRNA sequencing.

DNA isolation.

Depending on the size of the pelleted material, genomic DNA was isolated either with the DNeasy Kit (Qiagen) using manufacturer's protocol for Gram-negative bacteria or with the QIAmp DNA Micro Kit (Qiagen) using manufacturer's protocol for DNA isolation from urine. Purified DNA was quantified using NanoDrop spectrophotometer (Thermo Fisher Scientific). Fractions of human and bacterial DNA in each sample were determined using Femto Human and Femto Bacterial DNA quantification kits (Zymo Research) according to manufacturer's instructions.

Sequencing.

V4 regions of 16S rRNA genes were amplified using primers 5′-TCGTCGGCAGCGTCA GATGTGTATAAGAGACAGGTGCCAGC MGCCGCGGTAA-3′ and 5′-GTCTCGTGGGCT CGGAGATGTGTATAAGAGACAGGACTACH VGGGTWTCTAAT-3′ and the following reagent concentrations: 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 1.5 mM MgCl2, 200 μΜ of each dNTP, 2 μΜ of each primer, 1% glycerol, 0.3 U AccuPrime Taq polymerase (Thermo Fisher Scientific), and 25 ng of template DNA in 20 μL total volume. Amplification conditions were 2 minutes at 95°C initial denaturation followed by 30 cycles of 20 s denaturation at 95°C, 15 s annealing at 55°C, 5-minute extension at 72°C, and a 5 -minute final extension at 72°C. Amplification products were purified with the AMPure XP system (Beckman Coulter), and their size was verified with the DNA 1000 Kit (Agilent). Indexing and pooling of amplification products were carried out according to Illumina's 16S Metagenomic Sequencing Library Preparation protocol. The resulting library was sequenced using the Illumina MySeq Reagent Kit v2 (500 cycles).

Analyses.

A computational biologist with expertise in microbiome research (M.P.-L.) performed all of the analyses of the high-throughput sequence data. Raw FASTQ files were processed in dada2.²² This pipeline offers improved taxonomic resolution and reproducibility compared to OTU-based methods.²³ Reads were filtered using standard parameters, with no uncalled bases, maximum of two expected errors, and truncating reads at a quality score of two or less. Forward and reverse reads were truncated after 220 and 200 bases, respectively. The standard dada2 pipeline was then applied to perform amplicon sequence variant (ASV) inference, merge paired reads, and identify chimeras. Taxonomic assignment was performed against the Silva v132 database²⁴ using the dada2-formatted training files for taxonomy and species-level assignment.²³ ASV sequences were aligned using MAFFT²⁵ and used to build a tree with FastTree and midpoint rooting.²⁶ The resulting ASV tables and phylogenetic tree were imported into phyloseq²⁷ for further analysis. All ASV singletons (n = 1) were eliminated. We normalized our samples using the negative binomial distribution as recommended by McMurdie and Holmes²⁷ and implemented in the Bioconductor package DESeq2.²⁸ This approach simultaneously accounts for library size differences and biological variability. Taxonomic alpha-diversity was estimated using Fisher, Shannon, and ACE indices, whereas phylogenetic alpha-diversity was calculated by the Faith's phylogenetic diversity index.²⁹ These indices measure different aspects of the microbial communities; taxonomic indices inform about the richness and evenness of the microbiotas, while the phylogenetic index considers the relatedness of the community members. Beta-diversity was estimated using phylogenetic Unifrac (unweighted and weighted), Bray-Curtis, and Jaccard distances. Similarly to alpha-diversity indices, Unifrac distances take into account microbial relatedness, whereas Bray-Curtis and Jaccard distances do not. Dissimilarity in distance matrices between samples was then visualized using principal coordinates analysis (PCoA).

We used linear mixed-effects (LME) models analysis, as implemented in the lmer4 R package,³⁰ to investigate associations between alpha-diversity indices and taxa (genera and phyla) abundances (response), study phase, and instillation status (predictors), while accounting for non-independence of subjects (random effect). We also tested LME models with random intercepts and random slopes and different orders of factors. Initial LME models including all the variables were compared using the function lmerTest, which performs automatic backward elimination of factors. Analysis of variance (ANOVA) type III tests with Satterthwaite approximation for degrees of freedom were also carried out for hypothesis testing. Model assumptions in final LME models were validated using residual versus fit plots and normal probability plots.

Beta-diversity Unifrac indices were compared using permutational multivariate analysis of variance (adonis) as implemented in the vegan R package.³¹ Adonis models were compared using the Akaike Index Criterion.³² Significance was determined through 10,000 permutations.

Only statistically relevant factors in the dataset under study were included in the final LME models to avoid subjectivity of choice and over-parametrization.

We applied the Benjamini-Hochberg method at alpha = 0.05 to correct for multiple hypotheses testing.^33,34 All the analyses above were performed in R³⁵ and RStudio.³⁶

Results

A total of 126 urine samples were collected from 26 participants (SCI/D = 23; multiple sclerosis = 2; spina bifida = 1) between 20 and 57 years of age, enrolled in the clinical trial. Of these 26 participants, 20 (76.9%) completed the 18-month study. Urine samples were collected in three 6-month phases: baseline phase (N = 48 asymptomatic samples prior to any LGG instillation; range 1-6 samples from n = 25 participants); intervention phase (N = 58 total samples, range 1-4 samples from n = 13 participants: 25 = asymptomatic samples prior to any LGG instillation; 22 = symptomatic/pre-LGG instillation; and 11 = post-instillation); and washout phase (N = 20 asymptomatic samples, 1 sample/participant, collected an average of 21.7 weeks [range 13-40 weeks; SD 7.39] after LGG instillation). See Table 1.

Table 1.

Demographics of the 26 participants (A) and urine samples (B) by phase and phenotype

A. Demographics	n (%) or mean (range or count)
Mean age (range)	38.7 (20-57)
Female	23.1 (6)
SCI (n=23)	88.5 (23)
SB (n=1)	3.8 (1)
MS (n=2)	7.7 (2)
Level of SCI
Cervical	30.4 (7)
Thoracic	65.2 (15)
Lumbar	4.35 (1)
Completeness of SCI
Complete	30.4 (7)
Incomplete	69.6 (16)
Unknown	0 (0)
Mean years (range) since injury or diagnosis	11.8 (1-47)

B. Urine samples
	Baseline phase	Intervention phase			Washout phase
		Pre-LGG		Post-LGG
Symptom status	ASx	ASx	Sx	Sx	ASx
Subjects, n	25	13	11	9	20
Urine samples, n	48	25	22	11	20
Range of samples/subject	1-6	1-4	1-4	1-2	1

Note: ASx = asymptomatic; MS = multiple sclerosis; SB = spina bifida; SCI = spinal cord injury; Sx = symptomatic; LGG = Lactobacillus rhamnosus GG instillation

Taxonomic composition of NLUTD urine

We sequenced the variable region V4 of the 16S rRNA using the Illumina MiSeq platform. A total of 8,520,667 sequences ranging from 2,757 to 299,334 sequences per sample (mean = 65,043.3; median = 64,556) were obtained after quality control analyses. Two samples with less than 500 reads were excluded from subsequent analyses. From these data, we identified a total of 2428 ASVs.

The urobiomes across all 126 samples included sequences that corresponded to four dominant phyla (>1%): Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria. Those phyla comprised six dominant genera (>3%): Escherichia/Shigella (29.1%), Klebsiella (22.4%), Proteus (15.2%), Aerococcus (6.3%), Streptococcus (6.0%), and Pluralibacter (3.0%). Lactobacillus abundance was 2.4% across all samples and did not vary significantly by study phase (p = .513).

Urobiome varies by phenotype and LGG instillation status

In the baseline asymptomatic state, there was a predominance of uropathogenic genera (Proteus, Klebsiella, and Escherichia/Shigella). When the urobiomes were stratified by symptom presence and LGG instillation status (regardless of study phase), three of the four dominant phyla (Bacteroidetes, p = .009; Firmicutes, p = .025; and Proteobacteria, p < .001) and two genera (Escherichia/Shigella, p < .001; Aerococcus, p = .05) varied significantly, with Escherichia/Shigella having higher predominance at the time of symptoms and then decreasing after Lactobacillus instillation (see Table 2).

Table 2.

Abundance by phenotype (asymptomatic during any phase vs. symptomatic prior to LGG instillation vs post-LGG instillation)

Genus	Baseline asymptomatic, %	Symptomatic
		Pre-LGG, %	Post-LGG, %
Actinotignum	1.24	0.51	0.60
Aerococcus*	7.31	6.01	1.45
Enterococcus	3.40	2.56	1.76
Escherichia/Shigella**	16.82	40.74	22.59
Gardnerella	0.58	0.84	0.16
Klebsiella	18.41	31.46	26.81
Lactobacillus	3.79	0.52	1.53
Pluralibacter	2.21	5.70	0.72
Prevotella	1.27	1.00	1.27
Prevotella 7	1.19	0.50	0.26
Proteus	17.16	0.06	31.77
Staphylococcus	3.06	0.06	0.29
Streptococcus	8.83	4.36	5.92
Unclassified	2.18	0.51	1.01
Veillonella	3.07	0.40	0.25
All others combined	9.50	4.79	3.60

Note: LGG = Lactobacillus rhamnosus GG.

^*p < .05;

^**p < .001.

Alpha-diversity

Taxonomic alpha-diversity was estimated using Fisher, Shannon, and ACE indices, whereas phylogenetic alpha-diversity was calculated by the Faith's index. Alpha-diversity varied across phase, but only Shannon (p < .001) and Fisher (p = .048) estimates, which showed greater diversity for the intervention phase and less diversity for the baseline and washout phases, were significantly different in our LME analyses (see Figure 1).

Figure 1.

Alpha-diversity indices (Shannon, ACE, Fisher, and Faith's) across study phases.

Beta-diversity

Beta-diversity was estimated using phylogenetic UniFrac (unweighted and weighted), Bray-Curtis, and Jaccard distances. PCoAs of UniFrac, Bray-Curtis, and Jaccard distances showed segregation of the microbiotas across phases. Our Adonis analyses detected significant differences in beta-diversity across phases (p < .05) with the exception of UniFrac unweighted (p = .11).

Effect of Lactobacillus instillation on the urobiome

Pre-post instillation symptomatic samples

Only participants who instilled LGG at least once were considered (22 pre-instillation and 11 post-instillation samples among 9 participants; all samples of people experiencing urinary symptoms except 1 post-instillation sample) for this analysis. The urobiomes across all 33 samples included sequences that corresponded to four dominant phyla (>1%): Actinobacteria, Bacteroidetes, Firmicutes, and Proteobacteria. Those phyla comprised three dominant genera (>20%): Proteus, Klebsiella, and Escherichia/Shigella. Abundance of Escherichia/Shigella (p < .001) and Aerococcus (p = .049) were significantly reduced post-LGG instillation, whereas there were no significant differences in other uropathogenic genus abundance related to instillation (Enterococcus, p = .850; Proteus, p = .499; Klebsiella, p = .732; Enterobacter, p = .200; and Pseudomonas, p = .230). There was no significant difference in pre- versus post-instillation abundance of Lactobacillus (p = .237), Prevotella (p = .61), Streptococcus (p = .172), Gardneralla (p = .72), and Citrobacter (p = .65).

Alpha-diversity

Alpha-diversity indices (Shannon, PD, Fisher, ACE) varied pre- versus post-instillation, with Shannon (p < .001), and phylogenic diversity (p = .045) estimates significantly lower post-instillation. ACE (p = .119) and Fisher (p = .088) were not significantly different.

Beta-diversity

PCoAs of UniFrac-unweighted (p = 0.03), unifrac-weighted (p < .001), Bray-Curtis (p < .001), and Jaccard (p < .001) distances showed segregation of the microbiotas pre- versus post-instillation. Our Adonis analyses detected significant differences in beta-diversity across phases (p < .03). See Figure 2 and Table 3.

Figure 2.

(A) Taxonomic (Shannon, ACE, and Fisher) and phylogenetic (Faith's) alpha-diversity (within-in sample) indices showing variation in richness, evenness, or relatedness between pre-instillation and post-Lactobacillus GG (LGG) instillation groups. (B) Principal coordinates analysis (PCoA) of beta-diversity (between-sample) distances (phylogenetic Unifrac, Bray-Curtis and Jaccard) showing dissimilarity between pre-instillation and post-LGG instillation samples.

Table 3.

Linear mixed-effects (LME) model analyses of alpha-diversity and dominant phyla and genera for pre-instillation and post-instillation samples

	Type			Pre-post
	F	df	p(>F)	F	df	p(>F)
Alpha-diversity
ACE	2.1	4	0.084	2.5	1	0.12
Fisher	2.5	4	0.048	2.9	1	0.09
PD	2.3	4	0.07	4.1	1	0.05
Shannon	7.6	4	<0.0001	14.9	1	<0.0001
Beta-diversity
Unifrac-unw	1.2	4	0.106	2.3	1	0.03
Unifrac-w	3.0	4	0.003	10.8	1	<0.0004
Bray-Curtis	1.4	4	0.013	2.7	1	<0.0003
Jaccard	1.3	4	0.017	2.3	1	<0.0003
Phyla
Actinobacteria	1.5	4	0.207	2.6	1	0.11
Bacteroidetes	3.6	4	0.009	0.65	1	0.42
Firmicutes	3.0	4	0.025	12.5	1	<0.0001
Proteobacteria	5.3	4	<0.0001	14.9	1	<0.0001
Genus
Aerococcus	1.8	4	0.140	3.92	1	0.05
Enterococcus	1.8	4	0.141	0.04	1	0.85
Escherichia/Shigella	5.5	4	<0.0001	12.2	1	<0.0001
Klebsiella	0.8	4	0.542	0.12	1	0.73
Lactobacillus	0.8	4	0.513	1.41	1	0.24
Proteus	1.4	4	0.255	0.46	1	0.50
Pluralibacter	1.1	4	0.355	0.06	1	0.80
Streptococcus	2.1	4	0.099	1.89	1	0.17

Note : Significance of LME models was estimated using analysis of variance of type III with Satterthwaite approximation for degrees of freedom. For each test, we report the relevant F statistic (F), degrees of freedom (df), and significance [p(>F)]. Permutational multivariate analysis of variance (adonis) of beta-diversity indices.

Discussion

In this first-in-human trial of self-instillation with an intravesical LGG and normal saline mixture, we demonstrate that LGG bladder instillation can disrupt the uropathogenic NLUTD urobiome. Specifically, compared with pre-Lactobacillus instillation, we observed a significant decrease in phylogenic diversity (Shannon and Simpson indices), significant differences in beta-diversity, and reduced abundance of Escherichia/Shigella and Aerococcus, without significant change in the predominance of other uropathogenic (Klebsiella, Proteus, Pluralibacter, and Enterococcus) or commensal (Lactobacillus, Veillonella, Staphylococcus, Streptococcus) genera. These results were after one to two doses of self-instilled L. rhamnosus GG and were not associated with any change in urinary symptom status. While these results are considered preliminary due to the small sample size, they are encouraging for the potential clinical application of this live biotherapeutic approach to positively impact NLUTD-related urobiome dysbiosis and potentially urinary symptoms and urinary tract infection (UTI).

In this pilot work, intravesical Lactobacillus instillation significantly reduced abundance of Escherichia/Shigella and Aerococcus. This finding is consistent with our evolving understanding of the mechanistic effects of Lactobacillus strains. Lactobacillus has been shown to have bacteriostatic effects on uropathogens related to nutrient and attachment sites,³⁷ they downregulate the expression of uropathogen virulence genes,³⁸ and they have a direct bactericidal effect on uropathogens through bacteriocins.³⁹ As such, Thomas-White et al. have suggested that health-associated commensals, such as L. crispatus and L. iners, can reside in the bladder and provide protection against UTI.⁴⁰ However, it has also been suggested that L. iners potentially contributes to UTI⁴¹ among renal transplant patients. Therefore, although this early evidence around the potential effects of Lactobacillus strains in the urinary tract are encouraging, there is a suggestion that not all Lactobacillus strains behave similarly in the genitourinary tract of varying patient phenotypes. Clearly, more work is needed.

Our finding of a significant change in diversity after self-instilled intravesical Lactobacillus is encouraging for further clinical exploration into this biotherapeutic approach. Bacterial diversity of the urobiome may vary with age and/or hormonal status and body mass index.⁴² Among women without NLUTD, species richness, but not evenness, has been shown to be associated with urinary incontinence, and higher bacterial diversity in the absence of Lactobacillus dominance has been shown to be associated with urge incontinence and resistance to anticholinergic treatment.⁴³ In our initial work in which we cross-sectionally compared the urobiomes of healthy non-NLUTD controls to those of people with NLUTD (regardless of bladder management method), we found much greater phylogenic diversity among the Lactobacillus and Streptococcus branches of the phylogenic trees compared with those of Aerococcus and Enterococcus.² Given these early discoveries and the shift in diversity we note here after LGG instillation, it is suggestive of a substantive change in the urobiome as a result of the Lactobacillus being instilled into the bladder.

When symptom status was not considered (all samples pooled), the urobiomes of people with NLUTD and who manage their bladders with IC were largely uropathogenic with a predominance of Proteus, Klebsiella, and Escherichia, consistent with our previous work.^2,10 In our previous work, we described a predominance of Enterococcus, Klebsiella, and Escherichia among people with NLUTD due to SCI.² However, that study was not restricted to IC users and included those with NLUTD who void and who use indwelling catheters. In a recent review of the pediatric urobiome, the body of evidence around the urobiome among both adults and children with NLUTD was noted to be uropathogenic with a predominance of Enterococcus, Pseudomonas, Enterobacteriaceae, Staphylococcus, Escherichia, Veillonella, Prevotella, and Streptococcus (three studies).^9,44 Even though these early studies (including this study) provide evidence of a uropathogenic urotype among both adults and children who manage their NLUTD with IC, greater depth of understanding is needed of the varied urotypes and their relationship to clinical phenotypes.

Other work suggests that Lactobacillus may be a key contributor to urinary health. Brubaker and Wolfe have explored Lactobacillus species in urine and found that L. crispatus is associated with asymptomatic controls whereas L. gasseri is associated with urinary urge incontinence (UUI) symptoms.⁴⁵ In a case control study of healthy women with and without overactive bladder (OAB) symptoms, Wu et al. found that bacterial diversity (Simpson index) and richness (Chao1) were lower in OAB samples compared with non-OAB controls.⁴⁶ Further, some genera were increased (Proteus and Aerococcus) while others were reduced (Lactobacillus and Prevotella) in the OAB samples.⁸ Similarly, Curtiss et al., in a study of 63 women with OAB and 35 healthy controls, found that Proteus species were more commonly isolated whereas Lactobacillus was less commonly present in subjects with OAB.⁴⁷ In another study of asymptomatic women, Curtiss et al. found that Lactobacillus species were more commonly identified in the urinary tracts of premenopausal compared with postmenopausal women,⁴⁸ an interesting and potentially corollary finding given that OAB symptoms tend to emerge after menopause. Our population of individuals with NLUTD likely had varying degrees and combinations of urinary retention, overactive bladder, and detrusor sphincter dyssynergia; the population similarly had an absence of Lactobacillus dominance and a predominance of uropathogens, consistent with this emerging evidence base linking dysbiosis and various types of bladder dysfunction. Further work may benefit from phenotyping of NLUTD dysfunction utilizing urodynamics.

This is a pilot study; as such, the primary limitation is the small sample size, which limits the ability to analyze across covariates and clinical phenotypes. The latter point is important because the relevance of detecting organisms associated with specific phenotypes is unknown. For example, what are the host–bacteria and bacteria–bacteria interactions, and do these induce (or ameliorate) symptoms that define phenotypes? In our future work utilizing shotgun (and other) sequencing methods, we hope to advance the evidence regarding these relationships. Complicating the issue further, although we have advanced our ability to describe the NLUTD phenotype according to symptoms with the development of our USQNBs,^16,18,19,49 characterization of a NLUTD-UTI phenotype (important to clinicians, researchers, and patients) is elusive due to a lack of a standardized definition for UTI among this population.^20,50,51 We have mixed-methods work ongoing to develop complicated UTI diagnostic guidelines, which will offer a definition of complicated UTI (cUTI) that can be used in clinical practice and research (as inclusion/exclusion criteria and/or as an outcome measure).⁵² We anticipate an improved understanding of the uropathogenic urobiome that accompanies NLUTD and a broadening of our conceptualization beyond the presence or absence of “urinary symptoms” and “UTI” to a spectrum of NLUTD-associated eu- and dysbiosis. As Brubaker and Wolfe suggest, “As opposed to overt clinical infection, the concept of urinary dysbiosis would more usefully describe the clinical spectrum of altered microbial states.”⁵³ Indeed, this reconceptualization, as we and others have described, supports use of “[symptom or phenotype]-associated urinary dysbiosis” and jettisoning terms such as “asymptomatic bacteriuria,” which indirectly perpetuate the now disproven dogma that healthy urine is sterile urine. Simply put, our new understanding of the urobiome demands reconceptualization of urinary health and disease.

In conclusion, this preliminary evidence indicates that self-instilled intravesical LGG alters the bacterial composition and diversity of the urine ecosystem, potentially disrupting the uropathogenic urobiome of people with NLUTD who manage their bladders with IC. There are numerous approaches for future research, which include identifying optimal dosing, conducting a larger randomized trial, comparing different Lactobacillus strains, and exploring other sequencing techniques that will provide more functional information.

Funding Statement

Financial Support This project was supported by PCORI award AD1310-08215.