Urgency urinary incontinence (UUI), which is defined as involuntary urinary leakage accompanied by or immediately preceded by urgency (Haylen et al., 2010), affects up to 30% of women with increasing risk with age (Coyne et al., 2012). Women are twice as likely to be affected by UUI as men. Many etiologies have been proposed for UUI ranging from neurogenic to idiopathic (less understood) causes. The underlying pathophysiologic factors that contribute to non-neurogenic UUI are thought to range from abnormal sensory function at the level of the urothelium or urethra (de Groat, 1997; Yamaguchi et al., 2007) to involuntary myogenic (detrusor) contractions (Abrams et al., 2002), altered brain function (Griffiths et al., 2009; Nardos et al., 2014; Nardos et al., 2016), and more recently, a shift in urinary bladder microbiome (Brubaker et al., 2014; Pearce et al., 2014; Pearce et al., 2015; Karstens et al., 2016). Other factors such as metabolic disorders (Bunn et al., 2015), affective disorders (Vrijens et al., 2015) and hormonal changes (Cody et al., 2012) have also been shown to be associated with this condition. These pathophysiologic factors are likely not mutually exclusive.

Recent advances in both culture and culture-independent techniques have made it possible to evaluate the role of microorganisms more thoroughly in health and disease. With these breakthroughs came the discovery that microbes contribute to human health more extensively than initially thought (Haiser et al., 2013; Fulde and Hornef, 2014; Neuman et al., 2015). Recent studies demonstrate that resident bacteria in the bladder may have a role in healthy bladder function (Siddiqui et al., 2011; Fouts et al., 2012; Wolfe et al., 2012; Khasriya et al., 2013; Lewis et al., 2013; Hilt et al., 2014; Wolfe and Brubaker, 2015). An alteration in the resident bladder bacterial community, on the other hand, may be associated with bladder disorders such as overactive bladder (OAB) with or without UUI (Brubaker et al., 2014; Pearce et al., 2014; Pearce et al., 2015; Karstens et al., 2016; Drake et al., 2017; Aragón et al., 2018) and interstitial cystitis (Siddiqui et al., 2012).

It is generally understood that the gut is the main source of urinary uropathogens by way of an intermediary, i.e., vagina. The female urogenital tract is particularly amenable to this migration theory given the anatomical characteristics and proximity of these organs to each other. The vaginal microbiome is therefore thought to play a critical role in contributing to bladder health in addition to its well-established role in maintaining a healthy vaginal environment (Ravel et al., 2011; Greenbaum et al., 2019). However, little is known about how these proximal ecosystems, i.e., microbiomes in bladder and vagina, relate to each other to affect bladder health or disease. In this study, we use a microbial network framework to understand the shift in the dynamics of interactions or co-occurrences of bacteria in the vagina and urinary bladder in women with and without UUI. Such application of network analysis methods allow us to understand the larger microbial community structures of the bladder and vagina and how they differ in the urogenital tract of women with and without UUI.

This observational case–control study was conducted at the Oregon Health & Science University (OHSU) between 2016 and 2019. Study approval was obtained from the OHSU’s Institutional Review Board (IRB 00010729) as part of a larger effort to understand the role of urinary bladder microbiome in overactive bladder syndrome in women. Women between the ages of 45 and 85 were recruited both from the general population in the Portland area and through the Pacific Northwest Pelvic Floor Research Group, urogynecology clinical providers and researchers from the OHSU, Kaiser Permanente NW, and affiliated Portland-area hospitals. Participants were prescreened over the phone and those who met inclusion criteria completed their study visits at the OHSU’s Women’s Health Research Unit (WHRU).

We recruited twenty women with UUI (cases) and thirty women with normal bladder function (controls). The UUI group included women with daily urge-predominant incontinence confirmed on a three-day voiding diary, with urge-predominant leakage as determined by a Patient Perception of Intensity Urgency Scale (PPIUS) (Cartwright et al., 2010) score ≥3 (severe urgency that I could not postpone voiding) for >50% of total incontinence episodes on diary. The control group included female participants without a history of any UUI symptoms or frequent (>once a week) stress incontinence symptoms based on screening questionnaire and confirmed on three-day voiding diary. Participants were excluded if they had urinary retention with a baseline need for intermittent self-catheterization, known neurological diseases that could affect bladder function (stroke, multiple sclerosis, brain or spinal cord injury, myasthenia gravis), current pregnancy or lactation, history of pelvic radiation, current pelvic or bladder malignancy, symptomatic urinary tract infection detected on screening urinalysis and confirmed with culture (growth of >10⁵ colonies per ml), symptomatic pelvic organ prolapse (sensation of vaginal bulge), prior or current diagnosis of painful bladder syndrome, or a history of antibiotics in the previous two months.

All participants provided written consent and completed a demographic and health questionnaire and a three-day bladder diary. Participants were asked to score their urinary urgency on the bladder diary using the PPIUS. Participants also completed the International Consultation on Incontinence Questionnaire (ICIQ) (Avery et al., 2004), the Pelvic Floor Distress Inventory Urogenital Distress Inventory (UDI) (Barber et al., 2005) and the Overactive Bladder Questionnaire (OAB-q) (Coyne et al., 2002). These validated questionnaires assess urinary incontinence symptoms, impact of pelvic floor disorders on daily function, quality of life, symptom bother, and health-related quality of life, respectively. Finally, participants completed the Patient Global Impression of Severity (PGI-S) (Yalcin and Bump, 2003), the Patient Perception of Bladder Condition (PPBC) (Coyne et al., 2006) and the Beck’s Anxiety Inventory (Fydrich et al., 1992). During their study visit, urine was collected from the bladder using an aseptic technique with a urethral catheter by a trained and licensed practitioner. The total volume of urine was emptied, and urine specimens were aliquoted into sterile 50 ml conical tubes and stored at −20°C until further processing. Mid-vaginal swabs were also collected by a trained and licensed practitioner from the same participants on the same study visit. This was done by inserting a sterile cotton-tipped swab into the vagina, rotating the swab 360° five times, and letting the swab sit in mid vagina for 20 s. All specimens were handled in a sterile biosafety cabinet after collection.

The study included twenty women with UUI and thirty women without UUI (controls). Women in both groups were similar in age, menopause status, estrogen use, number of vaginal deliveries, and race (p >0.05) ( Table 1 ). There was also no significant difference in clinical history such as IBS, history of pelvic surgery, diabetes, or current tobacco use. However, women with UUI were more likely to have had a history of recurrent UTIs compared to controls (p = 0.007). Women with UUI were also significantly more likely to score higher on Beck’s Anxiety Inventory (10.6 ± 12.0) compared to controls (2.6 ± 2.6) (p <0.001) even though they did not report a higher incidence of anxiety diagnosis in their medical history.

As expected, the UUI group had higher scores on clinical symptom and symptom bother questionnaires related to incontinence (UDI, OAB-q, ICIQ, p <0.001, Table 2 ). The higher UDI-6, OAB-q symptom bother, and ICIQ scores for the UUI cohort indicate the severity of symptom bother and disability in this particular population. Similarly, the lower OAB-q health-related quality of life scores for the UUI cohort indicates a lower quality of life in the UUI population. Objective measures of incontinence are captured by a three-day bladder diary which showed that 60% of the UUI population had daily urge leaks and 15% had daily stress leaks while the control group did not have any urge or stress leaks. There was no significant difference in nocturia or daytime urinary frequency between the two groups.

The 16S rRNA amplicon sequencing resulted in a mean sequencing read depth of 45,005 reads per sample (range 15,201–69,787) for vaginal samples and 38,941 reads per sample (range 2,319–137,986) for urine samples. There were no significant differences in the number of reads per sample between the UUI and control groups (p = 0.7 vaginal samples, p = 0.2 urine samples). After processing and filtering as described in the methods, the vaginal sequencing resulted in classification of 8 phyla, 12 classes, 18 orders, 31 families, and 66 genera. Urine sequencing resulted in classification of 14 phyla, 21 classes, 43 orders, 70 families, and 131 genera.

It is generally accepted that the vagina acts as an intermediary between the gut and the bladder and may play a role in the pathogenesis of bladder conditions such as UTI (Czaja et al., 2009). It is also known that disruption of a normal microbiome environment in the vagina is associated with colonization by pathogenic organisms leading to disorders like bacterial vaginosis (Srinivasan et al., 2010), sexually transmitted infections (STIs) (van de Wijgert, 2017; Eastment and McClelland, 2018; Ziklo et al., 2018), and genital herpes infection (Shannon et al., 2017). It is however unclear if and how dysbiosis in the vagina plays a role in more chronic bladder conditions such as UUI.

In this study, we provide a detailed characterization of the differences in both bladder and vaginal microbiomes in a cohort of well-characterized women with and without UUI. More importantly, we go beyond traditional approaches to microbiome analysis by leveraging network-based analysis to look at underlying microbial community dynamics or interactions in these adjacent ecosystems. Network analysis provides a mathematical tool to understand complex systems such as identifying central bacteria that contribute to stability and resilience of ecosystems and their interactions with others in the network (Pielou et al., 2007).

In our study, women with UUI were similar to women without UUI in all clinical and demographic variables except for history of recurrent UTIs. Women with UUI were more likely to have a history of recurrent UTI which is not surprising given the known overlap in symptoms between UTI and UUI and the frequent misdiagnosis of UTI as a result (Nik-Ahd et al., 2018). There is also some evidence that overactive bladder may be associated with chronic low-grade colonization by bacteria that are commonly missed on routine cultures (Balachandran et al., 2016). These observations are the impetus for investigators, us included, to look at the role of bladder microbiomes in health and disease (Karstens et al., 2016; Brubaker and Wolfe, 2017; Drake et al., 2017; Aragón et al., 2018).

In our study, we did not see a significant difference in alpha or beta diversity in bladder or vaginal microbiomes of women with and without UUI. This is consistent with our own prior reporting (Karstens et al., 2016) but differed from others who reported increased diversity in UUI (Pearce et al., 2015; Thomas-White et al., 2016). Using silhouette analysis, we showed that the bladder and vaginal microbiomes formed two clusters of bacteria. In the vagina, one cluster was dominated by Lactobacillus and the other with mixed bacteria. Presence or absence of UUI did not impact clustering. The only clinical factor that was associated with this clustering was use of “any vaginal product” the week before collection of the vaginal specimen, with more vaginal product use noted in the Lactobacillus-dominated group. This observation is important to note and highlights the importance of environmental factors such as vaginal product use (douching, lubrication, estrogen cream etc.) that influence vaginal microbiome. This is consistent with the literature highlighting the dynamic nature of vaginal microbiomes and how these ecosystems can be perturbed by factors like menstruation, vaginal products, vaginal infections, hormonal status such as pregnancy, menopause, etc. (Srinivasan et al., 2010; Gajer et al., 2012; DiGiulio et al., 2015). Interestingly, the clustering in the vaginal microbiome was not associated with estrogen use per se. The small size of our study (with an even smaller number of women on estrogen therapy) does not allow us to make any definitive conclusion about the association of vaginal or systemic estrogen use and changes in vaginal or bladder microbiomes. This certainly needs further investigation given the promising evidence that vaginal estrogen improves OAB symptoms in post-menopausal women (Cardozo et al., 2004), reduces symptoms associated with genitourinary syndrome of menopause including the risk of UTI (Rahn et al., 2014), and that vaginal estrogen use may increase Lactobacillus in the urine of post-menopausal women (Thomas-White et al., 2020). This suggests that modulation of microbiomes is one likely mechanism of action for vaginal estrogens.

Similar to the vagina, silhouette analysis in the bladder microbiome identified two clusters. One cluster contains bladder microbiomes that have a higher proportion of Lactobacillus and Gardnerella, while the second cluster contains microbiomes with a mixed population of bacteria. This clustering in the bladder microbiome was associated with menopausal status in which post-menopausal women were more likely to have microbiomes without a dominant genus while premenopausal women tended to have microbiomes that were dominated by Lactobacillus or Gardnerella. This is consistent with prior report that shows higher Lactobacillus in the urine of premenopausal women compared to post-menopausal women (Curtiss et al., 2018). In our study, these dominant clusters in the bladder were not associated with presence or absence of UUI. This is in agreement with some prior work using similar-sized cohorts (Pearce et al., 2014), but different from other studies that found several clusters within UUI samples that were dominated by single bacteria such as Lactobacillus and Gardnerella (Pearce et al., 2015) and studies that identified bacterial community types that were different between women with and without mixed urinary incontinence (Komesu et al., 2018).

One of the unique aspects of this study was the use of network analysis to understand microbial interactions in the bladder and in the vagina. We identified two major microbiome interaction networks in both the vaginal and bladder samples. In the vaginal samples, we identified similar numbers of genera in each network for women with and without UUI. They also have similar numbers of clusters of subgroups in the microbiome-wide networks. However, women with UUI tend to have more unique bacterial associations or co-occurrences compared to controls. In the bladder on the other hand, cases have less than half as many network clusters compared to controls, suggesting that loss of smaller and more specialized microbiome networks may be a characteristic of UUI.

As part of our network analysis, we looked further to identify central bacteria that are defined by the highest betweenness score (centrality measure) and potentially play important roles in stabilizing the whole community. In network analysis, a node (which in our study represents a genus) that has maximum centrality can be thought of as a keystone taxon which maintains the network structure and potentially the function of the ecosystem. This is irrespective of whether or not it is the most abundant bacteria. In our study, we chose the threshold of 0.10 for our betweenness score because 90% of the values for both microbiome networks were lower than this threshold. In the vaginal microbiome network analysis, we found that Bacteroides and Lactobacillus were the two top central genera in control samples while Aerococcus was the central bacteria in women with UUI. Bacteroides as a central bacterium in women with normal bladder control is consistent with what we know about the role of Bacteroides in maintaining health in the urogenital tract. For example, delay in appearance or absence of Bacteroides has been observed in the guts of infants born by cesarean section compared to those delivered by the vaginal route, and this has been proposed as one of the etiologies for the higher predisposition of children born by cesarean section to disorders related to poor immune system maturation (Grönlund et al., 1999; Huurre et al., 2008). This seems to indicate that Bacteroides in the maternal birth canal (vagina) may have a key role for developing a healthy immune system that may persist in adults. Lactobacillus, the second most central vaginal bacteria in our control group, have been known to be a characteristic of a healthy vagina (Ravel et al., 2011; Drell et al., 2013). Our finding shows that beyond being the most abundant bacteria in the healthy vagina, Lactobacillus are also the central bacteria in the microbiome-wide network and thus may play a key role in the stability and function of this ecosystem. In our study, we did not perform a species-level analysis and thus are unable to explain the relevance of our findings on known vaginal community state types dominated by various species of Lactobacillus, such as L. crispatus, L. gasseri, L. iners, bacterial vaginosis-associated bacteria and L. jensenii.

In women with UUI, we found that the most central bacteria in the vaginal microbiome network is Aerococcus. Some species within this genus, such as A. urinae, are increasingly being recognized for their pathologic role in urinary tract disorders such as UTI, OAB, and UUI (Pearce et al., 2014; Kline and Lewis, 2016; Hilt et al., 2020). Among the top genera most associated with Aerococcus were Actinomyces, Gardnerella, Prevotella, and Bacteroides. Actinomyces species have been reported as associated with genitourinary actinomycosis (particularly in the setting of IUD use) and urinary tract actinomycosis (Huang and Al-Essawi, 2013; García-García et al., 2017). Gardnerella and Prevotella are well-known pathogens involved in bacterial vaginosis (Randis and Ratner, 2019). These findings seem to suggest that dysbiosis in the vagina characterized by changes in microbiome-wide community structure may be associated with urinary disorders like UUI.

In the bladder microbiome network analysis, Lactobacillus was the central genera in both UUI and control groups. Both groups shared Lactobacillus association with six common genera, but they each had several unique genera that did not overlap. Interestingly, many of the unique associations seen in women with UUI were with known uropathogens associated with UTI. It is important to note that in the bladder microbiome network analysis, what differentiates women with and without UUI is not the type of central bacteria (Lactobacillus for both) but rather its association with other bacteria in the network. This emphasizes the role of community network structures in health and disease and the need to go beyond quantification of relative abundance or diversity measures of microbiomes to understand dynamic ecosystems.

Looking at the similarities across the bladder and vaginal microbiomes, our study shows that women with UUI have a larger number of shared bacteria (43%) compared to women without UUI (29%). Interestingly, we found that although the two top bacteria that were most abundantly shared between these two ecosystems were the same between UUI and control samples (Lactobacillus and Gardnerella), Gardnerella was the most shared in the UUI subjects (24% overlap vs. 8.3%) while Lactobacillus dominated in the controls (17.3% overlap vs. 6.7%). Komesu et al. showed an overlap of 60% of the bacterial taxonomic units (genera) between the vagina and the bladder with the most abundant being the genus Lactobacillus (2020). Their study used younger participants (average age 53 compared to 61 in our study) and did not distinguish between disease and non-disease states. Others who performed species-level analysis comparing bladder genome samples with publicly available vaginal strains showed that there was an overlap of 23 species between the vagina and the bladder (Thomas-White et al., 2018). However, the majority of the samples in their study were from unrelated individuals in a less well-defined clinical population. Our observation that there is higher overlap in bacterial genera between bladder and vagina in women with UUI compared to controls and that the dominant bacteria involved in this overlap differs between these two groups indicates the possibility for more seeding of potentially pathogenic bacteria from the vagina into the bladder in women with UUI compared to controls. How this contributes to dysbiosis of the entire ecosystem is unclear.

The main limitation of our study is the small sample size that made it difficult to identify all relevant clinical and demographic factors that could influence microbiome community structure. For example, use of various vaginal products could potentially influence vaginal microbiome in different ways. Although there were no women who reported symptoms of vaginal infection in our study, we did not do objective screening for the presence or absence of infection in asymptomatic participants. We also used amplicon sequencing of the V4 region of the 16S rRNA gene to investigate the bladder and vaginal microbiomes. While this method is widely used for microbiome studies, it has known biases and limitations (Caporaso et al., 2012; Karstens et al., 2018; Knight et al., 2018). With this approach, we were limited to genus-level information, and could not robustly assess species-level differences or interactions, which may be of importance for understanding the associations to bladder health of the urogenital microbiome. One strength of our study is the use of a clinically well-characterized and demographically well-matched population of UUI and control subjects and the use of catheter-collected samples. Another, perhaps more important strength of our study is the application of network analysis methods to better understand the microbial community structures and how they differ in the urogenital tract of women with and without UUI. To our knowledge, this is the first time this analysis method has been applied to understand microbiome-wide networks in the urogenital tract of women with and without UUI. This approach provides a powerful tool to understanding the role of microbial communities as a whole in bladder health and disease and how ecosystems may be perturbed by environmental factors.

The original contributions presented in the study are publicly available. This data can be found here: https://www.ncbi.nlm.nih.gov/sra, PRJNA793927. Code for the analyses presented in this manuscriipt can be found at: https://github.com/KarstensLab/urogenital_microbiome_networks_in_uui.

The studies involving human participants were reviewed and approved by the Oregon Health & Science University, Institutional Review Board. The patients/participants provided their written informed consent to participate in this study.

RN Conceiving and designing experiment, writing manuscript. EL Data analysis, writing manuscript, editing manuscript. WG Assisting in study design, editing manuscript. LK data Analysis, conceiving and designing experiment, assisting in writing manuscript, editing. MA (Deceased): Assisted in designing and performing experiment. SD Performing experiment. ED Analyzing Data. All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

This project was supported by the Society of Urodynamics, Female Pelvic Medicine and Urogenital Reconstruction Foundation – OAB Urgency Incontinence Grant 2015 (RN); the National Institutes of Health (NIH) funded Oregon BIRCWH K12 award number K12HD043488 made possible through the Eunice Kennedy Shriver National Institute of Child Health and Human Development and the Office of Research on Women's Health (LK); the NIH National Institute of Diabetes and Digestive and Kidney Diseases K01 award number K01DK116706 (LK) and the NIH National Library of Medicine training award T15LM007088 (EL). The content is solely the responsibility of the authors and does not necessarily represent the official views of the any of the funders or National Institutes of Health.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.