Background
The urogenital microbiome is associated with urgency and mixed urinary incontinence symptoms and differential treatment responses to pharmacotherapy for urgency urinary incontinence.
Objective
This study aimed to describe whether the preoperative urinary and vaginal microbiomes were associated with surgical treatment responses at 12 months after a midurethral sling operation in women with mixed urinary incontinence.
Study Design
This cohort study compared the preoperative microbiome compositions of urine and vaginal samples from a subset of women undergoing a midurethral sling operation in the Effects of Surgical Treatment Enhanced With Exercise for Mixed Urinary Incontinence trial (NCT01959347) and compared the microbiota in women who were surgical responders vs surgical nonresponders. Twelve-month objective response was defined as a ≥70% reduction from baseline urinary incontinence episodes on a 3-day diary. Subjective response was defined as a change from baseline in the Urogenital Distress Inventory scores. Bacterial abundance and beta diversity were assessed using 16S ribosomal RNA sequencing. The primary differential abundance analysis described predominant bacterial operational taxonomic units associated with responders vs nonresponders using unadjusted and age-adjusted linear models.
Results
Objective nonresponders (n=28) compared with responders (n=72) were older (58.5±10.7 vs 51.6±10.2 years) and more likely postmenopausal without hormone use (odds ratio, 6.4; 95% confidence interval, 1.8–22.6). Vaginal and urinary microbiota beta diversities were associated with age ( P <.05) for both responders and nonresponders. Overall, predominant operational taxonomic units (genera) were Lactobacillus , Gardnerella , Tepidimonas , Escherichia , Streptococcus , and Prevotella . Operational taxonomic units from baseline urine samples were not significantly associated ( P threshold=.05) with surgical treatment responses. A greater abundance of baseline vaginal Lactobacillus was associated with an objective response ( P =.04) and Prevotella with an objective nonresponse ( P =.01). Adjusting for age, only a greater abundance of baseline vaginal Prevotella was associated with an objective nonresponse ( P =.01). Moreover, less abundant vaginal operational taxonomic units were associated with objective and subjective responses and persistent urinary incontinence symptoms ( P <.05).
Conclusion
Women meeting a 70% reduction of urinary incontinence treatment episodes (objective responders) had greater vaginal Lactobacillus at the time of the surgical procedure; however, controlling for age diminished this association. Women not meeting a 70% reduction of urinary incontinence episodes 1 year after a midurethral sling operation had greater vaginal Prevotella at the time of the midurethral sling operation. Further research is needed to determine whether therapy altering the vaginal microbiome may impact surgical treatment responses in women with mixed urinary incontinence.
Introduction
The urogenital microbiome is associated with lower urinary tract health, although the relationship is complicated. Although some investigators have found that urgency urinary incontinence (UUI) symptoms are associated with the urinary microbiome, a more complex relationship has been found in women with mixed urinary incontinence (MUI). Specifically in women <51 years of age, urinary microbiome community types with fewer Lactobacillus (≤61% of genera) was associated with MUI, whereas a higher proportion of controls had high Lactobacillus communities (89% Lactobacillus ).
Why was this study conducted?
This study aimed to evaluate whether preoperative urinary or vaginal microbiomes are associated with surgical treatment responses at 12 months in women with mixed urinary incontinence (MUI) undergoing a midurethral sling placement.
Key findings
Older and postmenopausal women without hormone replacement more commonly were surgical nonresponders. A greater abundance of preoperative vaginal Lactobacillus was associated with improved objective surgical response (decreased urinary incontinence episodes); this effect diminished after adjusting for age. A greater abundance of preoperative vaginal Prevotella was associated with worsened objective surgical responses even after adjusting for age. Urinary microbiota were not associated with surgical treatment responses.
What does this add to what is known?
Baseline vaginal microbiome findings were associated with surgical treatment responses to midurethral slings in women with MUI.
Recently, we demonstrated that the composition of the urinary microbiome is correlated with the vaginal microbiome in women with MUI. This complements previous genome-level work showing that vaginal and bladder microbiota are functionally distinct from gastrointestinal microbiota but similar to each other. The interrelatedness of the vaginal and urinary microbiomes may offer unique opportunities to optimize lower urinary tract health and surgical outcomes. For example, vaginal therapies could have beneficial downstream effects on the urinary microenvironment and surgical outcomes. Alternatively, just as differences in the urinary microenvironment are associated with differential treatment responses to medication for UUI, the vaginal microenvironment at the time of the midurethral sling operation in women with MUI could be related to differential treatment responses. This is plausible given that surgical outcomes following pancreatic and colon operations may be associated with specific gut microbiome compositions and targeting of the gut microbiome has been suggested as a means of improving surgical outcomes.
This was a planned secondary analysis in a subset of women participating in the effect of behavioral and pelvic floor muscle therapy combined with surgery vs surgery alone on incontinence symptoms among women with mixed urinary incontinence (Effects of Surgical Treatment Enhanced With Exercise for Mixed Urinary Incontinence [ESTEEM]) trial (NCT01959347). This study aimed to describe preoperative urinary and vaginal bacterial taxa associated with 12-month responses after a midurethral sling operation in women with MUI. It was hypothesized that the relative abundance of Lactobacillus among preoperative urinary or vaginal microbiota would be associated with surgical treatment responses.
Materials and Methods
Study population
After institutional review board approval and written informed consent were obtained, catheterized urine and vaginal bacterial samples were collected from a subset of women (126 of 480) with MUI enrolled in the ESTEEM trial at 1 time point, before surgery. As previously described, ESTEEM participants were randomized in a 1:1 ratio to either midurethral sling operation or midurethral sling operation plus perioperative behavioral and pelvic floor muscle therapies. Baseline and follow-up lower urinary tract symptoms were obtained using the 19-item Urogenital Distress Inventory (UDI), a validated questionnaire measuring symptom severity, and 3-day bladder diary characterizing incontinence episodes. Here, 2 subjective measures and 1 objective measure of response were evaluated. The binary subjective response was defined as a reduction in the UDI score from baseline to 12 months meeting the minimal clinically important difference (MCID) threshold (−26.1 points). The second subjective measure was a change from baseline in the UDI score (Δ UDI) at 12 months. For participants who underwent additional urinary incontinence (UI) treatment before 12 months, Δ UDI was determined using the participant’s UDI score from the last questionnaire completed before the initiation of additional treatments. Binary objective response to treatment was defined as a ≥70% reduction from baseline in UI episodes (UIEs) on the 3-day diary at 12 months and no additional treatment, as previously described.
Sample processing
Sample processing has previously been described. , Briefly, after DNA extraction, the 16S rRNA variable regions (V4–V6) were amplified using polymerase chain reaction primers 515F and 1114R and sequenced with 300 bp paired-reads on the Illumina MiSeq (Illumina, Inc, San Diego, CA). Sample duplicates in 10% of the samples were performed, and negative controls were included for quality control. Operational taxonomic unit (OTU) counts were generated using the Illumina BaseSpace 16S Metagenomics App (version 1.01) for sequencing data from urine and vaginal samples. The pipeline classified OTUs to the genus level using the Ribosomal Database Project Classifier and a curated version of the Greengenes database.
Statistical analyses
Clinical and demographic differences between surgical responders and nonresponders were assessed using unadjusted general linear and logistic models and Wilcoxon rank-sum tests implemented in Statistical Analysis System software (version 9.4; SAS Institute Inc, Cary, NC). OTU count data were normalized using total sum scaling to calculate relative abundances. To evaluate beta diversity differences associated with clinical and demographic characteristics and surgical treatment responses, Bray-Curtis dissimilarity metrics were calculated for urinary and vaginal sequencing data and analyzed using permutational multivariate analysis of variance (ANOVA). , These analyses were implemented in R (version 3.6.0, R Foundation for Statistical Computing, Vienna, Austria) using the vegdist and adonis functions in the vegan package (version 2.5-6). Beta diversity was examined for each clinical and demographic variable independently. Multivariate models were used to identify covariates to be included in subsequent differential abundance analyses.
The primary differential abundance analysis focused on characterizing the most prevalent genera, defined as being the predominant OTU in urine or vaginal samples. For this analysis, only the OTUs present in 3 or more subjects were considered. The differential abundance of the predominant genera (n=6) was compared between treatment responders and nonresponders, using the normalized and arcsine square-root–transformed OTU data with unadjusted and adjusted generalized linear models (GLMs) implemented with MaAsLin with a significance threshold of P <.05. The GLM implemented in MaAsLin was also used to evaluate the relationship between differential OTUs and participant characteristics selected as model covariates.
Moreover, the differential abundance analysis was expanded to include an analysis of less abundant OTUs, defined as the presence in at least 10% of the samples and a minimum abundance (normalized OTU count) of 0.0001. False discovery rate (FDR) adjustment was used as a multiple testing correction using the Benjamini-Hochberg procedure. For discovery purposes, an FDR-corrected P value threshold of .10 was applied. Differential OTUs were annotated using Medical Subject Headings ( https://www.nlm.nih.gov/mesh/meshhome.html , annotated files created 7-20).
An exploratory analysis was performed to characterize whether the relationships between genera and surgical treatment responses were more closely associated with the stress or urgency components of UI. The relative abundance of 4 OTUs of interest were compared among women who had persistent stress urinary incontinence (SUI), UUI, MUI, or none (defined as reporting only SUI, only UUI, both, or neither in the 12-month diary) using unadjusted and adjusted ANOVA and the Tukey Honest Significant Difference tests. All analyses and plots were performed in R (version 3.6.0) unless otherwise noted.
Results
Of 126 women with MUI who contributed urine and vaginal samples, 100 provided objective treatment response data, and 104 provided subjective response data at 12 months. Nearly all participants (93%) met the MCID criteria for the reduction from baseline UDI score at 12 months. Therefore, because of the small number of nonresponders, this binary subjective outcome was not analyzed further, and the change in UDI (Δ UDI) score was assessed as the subjective outcome.
Compared with responders (n=72), objective nonresponders (n=28) were older and more likely to be postmenopausal not on hormones. Other clinical and demographic variables are noted in Table 1 . Mean UIEs at 12 months were higher in nonresponders by 2.9 incontinence episodes per day (95% confidence interval [CI], 2.0–3.8). Responders had a greater reduction in leaks from baseline with a mean difference of 2.2 incontinence episodes per day (95% CI, 0.8–3.6). Objective responders had greater improvements in UDI score at 12 months overall and by UI subtype ( Table 2 ).
| Variable | Category | Total (N=100) | Objective nonresponders (n=28) | Objective responders (n=72) | Unadjusted mean or median difference or odds ratio (95% CI) |
|---|---|---|---|---|---|
| Age (y) | 53.5 (10.8) | 58.5 (10.7) | 51.6 (10.2) | 6.9 (2.2–11.6) | |
| Body mass index | 32.3 (7.2) | 34.2 (7.2) | 31.6 (7.1) | 2.6 (−0.6 to 5.8) | |
| Race | White | 71 (71.0) | 16 (57.1) | 55 (76.4) | Ref |
| >1 race | 1 (1.0) | 1 (3.6) | 0 (0.0) | — a | |
| Asian | 2 (2.0) | 1 (3.6) | 1 (1.4) | 3.4 (0.2–60.3) | |
| Black or African American | 12 (12.0) | 5 (17.9) | 7 (9.7) | 2.5 (0.7–8.9) | |
| American Indian or Alaskan Native | 1 (1.0) | 0 (0.0) | 1 (1.4) | — a | |
| Other | 13 (13.0) | 5 (17.9) | 8 (11.1) | 2.1 (0.6–7.6) | |
| Ethnicity | Not Hispanic or not Latina | 72 (72.0) | 18 (64.3) | 54 (75.0) | Ref |
| Hispanic or Latina | 26 (26.0) | 10 (35.7) | 16 (22.2) | 1.9 (0.7–4.9) | |
| Unknown or not Reported | 2 (2.0) | 0 (0.0) | 2 (2.8) | — a | |
| Smoking status | Never smoked | 58 (58.0) | 16 (57.1) | 42 (58.3) | Ref |
| Quit smoking ≥6 mo ago | 29 (29.0) | 9 (32.1) | 20 (27.8) | 1.2 (0.4–3.2) | |
| Quit smoking <6 mo ago | 2 (2.0) | 0 (0.0) | 2 (2.8) | — a | |
| Currently smoking | 11 (11.0) | 3 (10.7) | 8 (11.1) | 1.0 (0.2–4.3) | |
| Current UTI symptoms | 2 (2.0) | 0 (0.0) | 2 (2.8) | — a | |
| Vaginal deliveries, median (min–max) | 2 (0–8) | 2 (0–6) | 2 (0–8) | 0 (−1 to 1) | |
| Menopause and hormone status | Premenopausal | 31 (31.0) | 4 (14.3) | 27 (37.5) | Ref |
| Postmenopausal with hormones | 19 (19.0) | 4 (14.3) | 15 (20.8) | 1.8 (0.4–8.4) | |
| Postmenopausal without hormones | 33 (33.0) | 16 (57.1) | 17 (23.6) | 6.4 (1.8–22.6) | |
| Not sure | 17 (17.0) | 4 (14.3) | 13 (18.1) | 2.1 (0.4–9.8) |
| Variable | Total (N=100) | Objective nonresponders (n=28) | Objective responders (n=72) | Unadjusted mean difference (95% CI) |
|---|---|---|---|---|
| Daily UIEs from a 3-d diary | ||||
| Total UIEs | ||||
| Baseline | 5.2 (3.0) | 5.7 (3.2) | 5.0 (2.9) | 0.7 (−0.7–2.1) |
| 12 mo | 1.0 (1.8) | 3.1 (2.3) | 0.2 (0.4) | 2.9 (2.0–3.8) |
| Change from baseline to 12 mo | −4.2 (3.0) | −2.6 (3.2) | −4.8 (2.8) | 2.2 (0.8–3.6) |
| Stress UIEs | ||||
| Baseline | 2.3 (1.7) | 2.1 (1.7) | 2.4 (1.6) | −0.3 (−1.1 to 0.5) |
| 12 mo | 0.2 (0.7) | 0.7 (1.2) | 0.0 (0.1) | 0.7 (0.2–1.1) |
| Change from baseline to 12 mo | −2.1 (1.8) | −1.3 (2.0) | −2.4 (1.6) | 1.0 (0.2–1.9) |
| Urge UIEs | ||||
| Baseline | 2.4 (2.4) | 3.1 (2.7) | 2.2 (2.3) | 0.9 (−0.3 to 2.0) |
| 12 mo | 0.6 (1.4) | 1.9 (2.0) | 0.1 (0.4) | 1.8 (1.0–2.6) |
| Change from baseline to 12 mo | −1.8 (2.3) | −1.1 (2.4) | −2.1 (2.3) | 0.9 (−0.1 to 2.0) |
| Urogenital Distress Inventory | ||||
| Total score | ||||
| Baseline | 179.2 (42.4) | 181.3 (39.3) | 178.4 (43.7) | 2.9 (−15.5 to 21.2) |
| 12 mo | 46.3 (60.2) | 118.0 (60.3) | 19.5 (31.5) | 98.5 (73.7–123.3) |
| Change from baseline to 12 mo | −132.9 (69.0) | −63.3 (55.3) | −159.0 (54.0) | 95.6 (70.7–120.6) |
| Stress subscore | ||||
| Baseline | 86.0 (17.4) | 83.3 (18.5) | 87.0 (17.1) | −3.7 (−11.9 to 4.5) |
| 12 mo | 19.5 (28.5) | 48.8 (31.7) | 8.6 (17.5) | 40.2 (27.1–53.3) |
| Change from baseline to 12 mo | −66.5 (33.5) | −34.6 (33.6) | −78.5 (24.5) | 43.9 (29.5–58.3) |
| Irritative subscore | ||||
| Baseline | 66.9 (18.6) | 72.0 (19.0) | 65.0 (18.2) | 7.0 (−1.6 to 15.5) |
| 12 mo | 20.0 (27.2) | 53.3 (27.0) | 7.6 (13.2) | 45.7 (34.7–56.8) |
| Change from baseline to 12 mo | −46.9 (28.1) | −18.7 (24.7) | −57.5 (21.3) | 38.8 (27.9–49.6) |
Significant differences in microbiome composition (beta diversity) were found in vaginal specimens concerning objective response, along with age and menopausal status (all P <.05). Significant differences in microbiome composition were found in urine specimens concerning age ( P =.001) ( Table 3 ). When age, menopausal status, and hormonal status were analyzed together, only age remained significant ( P <.05). Therefore, only age was included in the adjusted models.
| Variable | Urine | Vagina | ||
|---|---|---|---|---|
| R 2 | P value | R 2 | P value | |
| Age | 0.04 | .001 a | 0.06 | .001 a |
| Body mass index | 0.01 | .15 | 0.01 | .30 |
| Race | 0.01 | .83 | 0.01 | .67 |
| Ethnicity | 0.01 | .80 | 0.01 | .64 |
| Current smoking | 0.01 | .10 | 0.01 | .23 |
| Current UTI symptoms | 0.004 | .96 | 0.01 | .73 |
| Ever pregnant | 0.01 | .46 | 0.01 | .60 |
| Menopause and hormone status | 0.03 | .08 | 0.05 | .01 a |
| Oral or patch hormone use | 0.01 | .72 | 0.01 | .14 |
| Vaginal cream or tablet hormone use | 0.01 | .76 | 0.01 | .46 |
| Any hormone use | 0.01 | .47 | 0.01 | .19 |
| Δ UDI at 12 mo | 0.02 | .09 | 0.02 | .07 |
| Objective response | 0.01 | .39 | 0.02 | .03 a |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
