OUP user menu

Journal of Crohn's and Colitis: 10 (11)


Laurence J. Egan, Ireland

Associate Editors

Shomron Ben-Horin, IsraelSilvio Danese, ItalyPeter Lakatos, HungaryMiles Parkes, UKJesús Rivera-Nieves, USABritta Siegmund, GermanyGijs van den Brink, NLSéverine Vermeire, Belgium


Published on behalf of

Increased Gardnerella vaginalis urogenital biofilm in inflammatory bowel disease

Johannes Schilling, Vera Loening-Baucke, Yvonne Dörffel
DOI: http://dx.doi.org/10.1016/j.crohns.2013.11.015 543-549 First published online: 1 June 2014


Introduction Inflammatory bowel disease (IBD) is a systemic inflammatory condition that affects the entire organism, not only the bowel. An impaired interaction with microbiota has been shown to be important. We looked for bacterial factors, which may contribute to the well-known higher incidence of poor reproductive outcome in IBD.

Methods Urine specimen of patients with Crohn's disease (N = 42), ulcerative colitis (N = 46), and randomly selected patients attending the General Internal Medicine Outpatient Clinic of the Charité for non-IBD related medical conditions (N = 49) was analyzed for bacteria adherent to desquamated epithelial cells and diffusely distributed bacteria in the urine using fluorescence in situ hybridization.

Results The urine of IBD patients contained significantly more often Gardnerella vaginalis biofilms (CD 38%, UC 43%) than those of the control group (16%). There was no link between current disease activity, history of and present fistula and G. vaginalis biofilms, but the samples of patients with steroid refractory/dependent disease were significantly more often G. vaginalis biofilm positive. No significant differences in number of epithelial cells and leukocytes, and total bacterial counts were present.

Conclusions There is a significant link between IBD and G. vaginalis biofilm. This observation suggests an epithelial barrier dysfunction of the genital tract. Since G. vaginalis is believed to be one of the reasons responsible for bacterial vaginosis, it may be an important factor in the well-known higher incidence of poor reproductive outcome in IBD. Excessive G. vaginalis biofilms in steroid refractory/dependent disease suggests a need to avoid long-term steroid therapy.

  • FISH
  • Gardnerella vaginalis
  • Urogenital biofilm
  • Crohn's disease
  • Ulcerative colitis
A. vaginae,
Atopobium vaginae
bacterial vaginosis
colitis activity index
Crohn's disease
Crohn's disease activity index
cyanine dye, yellow fluorescence
cyanine dye, red fluorescence
fluorescence in situ hybridization
fluorescein isothiocyanate
G. vaginalis,
Gardnerella vaginalis
inflammatory bowel diseases
ulcerative colitis

1 Introduction

The structure of intestinal microbiota differs in patients with inflammatory bowel disease (IBD) and healthy and disease controls.1 We and others have previously reported on changes of the intestinal microbiota in patients with IBD.14 IBD affects the entire organism, not only the bowel. Till now, most studies on microbial structure in IBD patients were carried out on feces. Little is known about urogenital microbiota in patients with IBD.

Williams et al.5 showed that specific urinary metabolites, such as hippurate and formate levels related to gut microbial metabolism, differ between patients with Crohn's disease (CD), patients with ulcerative colitis (UC), and controls. It is possible that there are specific changes in the composition of the urogenital microbiota in patients with IBD. Bacteria are commonly found in relatively low concentrations in the urine sediment. Many epithelial cells are found in spontaneously voided urine and bacteria adherent to them can be investigated noninvasively. We found that urine samples from women with bacterial vaginosis gave similar results as vaginal biopsies.6 In addition, Dong et al.7 described the reliability of the urine samples of the initial void in the analysis of the microbial communities in the distal male urethra. They showed that the microbiota in male's initial voided urine and the urethral swab specimens were nearly identical. When we characterized the spatial distribution of bacteria adherent to the urethral epithelium of the initial voided urine from adult men using fluorescence in situ hybridization (FISH) analysis, we found that Gardnerella vaginalis biofilms are present in 7% of the samples of unselected hospitalized male patients.6

G. vaginalis is a gram variable rod that is believed to be the most important pathogen in bacterial vaginosis. It exists in a dispersed and in a biofilm mode. The dispersed form is unspecific, while the biofilm mode is indicative for bacterial vaginosis (BV).6,8

A vast number of data published over the last 30 years demonstrate that BV is associated with high risk preterm birth. We had shown previously that Gardnerella dominated BV biofilms and can spread up to the uterine cavity and infect the fetus, possibly explaining this association.9

Since a high preterm birth rate had been also described in CD women, we hypothesized that the disturbance of the vaginal flora may be an explanation for the higher preterm birth rate in CD patients.

2 Materials and methods

2.1 Patients

Forty-two patients (30 females and 12 males) with histologically confirmed CD, 46 patients (31 females and 15 males) with histologically confirmed UC, and 49 patients (37 females and 12 males) (controls), who attended the General Internal Medicine Outpatient Clinic of the Charité for other reasons than IBD, were enrolled. Participants with current antibiotic treatment or current urinary tract infections (UTIs) were excluded.

The current disease activity in IBD patients was rated according to the Crohn's disease activity score (CDAI)10 and the Colitis activity score (CAI).11 A CDAI of ≥ 150 indicated active CD, while a CAI of ≥ 5 indicated active UC. IBD was counted as severe, when the CDAI score was > 450 or the CAI score was > 10. Patients with a CDAI of < 150 or a CAI of < 5 were considered to be in remission. We collected data from the files of all IBD patients including current medication, history of fistula, and primary sclerosing cholangitis (PSC).

Steroid refractory/dependent disease was defined following the recommendations of the European guidelines for CD and UC12,13 as active disease despite prednisolone up to 0.75 mg/kg/day for a period of 4 weeks.12,13 Steroid dependent disease was defined as the inability of reducing steroids below the equivalent of prednisolone 10 mg/day within 3 months of starting steroids without recurrent active disease or a relapse within 3 months after stopping the steroids.12,13

2.2 Urine samples

A 2 ml sample of the initial part of the first voided morning urine was obtained from every IBD patient and control. The subjects added 2 ml of this urine to a tube containing 8 ml of Carnoy solution (6/6/1 vol. ethanol/glacial acetic acid/chloroform), which was given to them by the investigator. The initial part of the urinary stream was used for the study to ensure that the concentration of epithelial cells was sufficient for analysis. The patients returned the samples to the outpatient clinic at their next appointment or delivered them directly to the Laboratory for Molecular Genetics, Polymicrobial Infections and Bacterial Biofilms in Berlin, Director: Dr. Alexander Swidsinski. The tubes were vortexed and then an aliquot of 1.5 ml was pipetted into an Eppendorf tube and then centrifuged for 6 min at 6000 g. The supernatant was removed from the sediment and 75 μl of Carnoy solution was added. The processed sample represented now is 300 μl of the initial urine sample and was stored in the Eppendorf tube at 4 °C until hybridization.

A 5 × 5 mm hybridization area was marked with a PAP Pen on a superfrost glass slide (Langenbrinck, Emmendingen, Germany). The sample was vortexed once more and a 5 μl aliquot (representing 15 μl of the initial sample) was pipetted within the area of hybridization. Four glass slides were prepared in this manner. The glass slides with the aliquots were dried for 30 min at 50 °C prior to hybridization.6 In addition, three of the four glass slides of every sample were incubated with 20 μl of 1% lysozyme for 15 min at 37° C.

2.3 FISH

Five microliters of the hybridization solution was pipetted onto each glass slide. The slides were hybridized for 60 min at 50 °C. The protocol used was previously reported by Swidsinski et al.6 The eight oligonucleotide probes listed in Table 1 were used.8,1420

View this table:

The oligonucleotide probes.a

Eub338Eubacteria (virtually all bacteria)FITCAmann et al.14
Ato291 Atopobium clusterCy3Harmsen et al.15
Bif164 Bifidobacterium spp.Cy3Langendijk et al.16
EbacEnterobacteriaceaeCy3Bohnert et al.17
Lab158 Lactobacillus and Enterococcus Cy3Harmsen et al.18
Bac303Most BacteroidaceaeCy5Manz et al.19
GardV Gardnerella vaginalis Cy5Swidsinski et al.8
Strc493 Streptococcus spp.Cy5Franks et al.20

All glass slides were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Bacteria, epithelial cells and leukocytes were assessed in a multi-color analysis using a mix of three 16 or 23 rRNA probes on each of the four superfrost glass slides. Four combinations of oligonucleotide probes were used and are shown in Table 2.

View this table:

Probe combinations.

1. Glass slideBif164GardVEub338YesDAPI
2. Glass slideLab158Strc493Eub338YesDAPI
3. Glass slideAto291GardVEub338YesDAPI
4. Glass slideEbacBac303Eub338NoDAPI

Concentrations of epithelial cells and leukocytes were calculated per ml urine. The concentrations of the bacterial subgroups were estimated semiquantitatively in 5 × 105/ml steps by estimating the number of bacteria per field of view. This number was then used to approximate the concentration of bacteria per ml. Samples were counted as positive for G. vaginalis biofilm, if at least one desquamated epithelial cell showed a typical biofilm of adherent G. vaginalis (see Figs. 1 and 2). Microscopy was performed using a Nikon e600 fluorescence microscope. The images were photodocumented using a Nikon DXM1200 color camera and the accompanying software (Nikon, Tokyo, Japan). The glass slides were stored at room temperature after analysis.


The figure shows 4 desquamated epithelial cells (DAPI, blue fluorescence) with adherent G. vaginalis biofilm (GardV-Cy5, red fluorescence) at a magnification of 1000.


The figure shows a typical bacterial vaginosis biofilm in the urine sample of an IBD patient at a magnification of 400. The main part of the biofilm is composed of G. vaginalis (GardV-Cy5, red fluorescence), but there is also a large number of A. vaginae (Ato291-Cy3, yellow/orange fluorescence) present. The left figure shows the counterstain with DAPI (blue fluorescence).

2.4 Statistical analysis

The statistical analysis was performed using SPSS v. 16.0 (Chicago, Illinois, USA). A two-tailed p-value of < 0.05 was considered to be significant. The one-sample Kolmogorov–Smirnov test was used to analyze the deviation from a theoretical normal distribution of the concentrations of epithelial cells and leukocytes. We compared the concentration of the epithelial cells and the leukocytes using the nonparametric Mann–Whitney U-test when the p-values of the Kolmogorov–Smirnov test was < 0.05. The semiquantitative bacterial concentrations were compared using the Mann–Whitney U-test. A possible correlation of the frequency of G. vaginalis biofilms and IBD was tested using the Chi-square-test, while a correlation of current disease activity and G. vaginalis biofilms was analyzed with the Spearman's rank correlation coefficient. The concentrations are given as mean ± SD.

3 Ethical considerations

All subjects gave written informed consent according to the protocol approved by the Ethics Committee of the Charité Hospital in Berlin (number: EA1/170/08).

4 Results

4.1 Concentration of epithelial cells and leukocytes (Table 3)

View this table:

Mean ± SD concentrations of epithelial cells and leukocytes and presence of G. vaginalis biofilm.

Epithelial cells × 105/mlLeukocytes × 105/mlG. vaginalis biofilm positivep-Valuesa
CDN = 421.38 ± 1.752.98 ± 7.8616/42 (38%)0.019
Only femalesN = 301.66 ± 1.854.07 ± 9.1214/30 (47%)0.020
Only malesN = 120.66 ± 1.250.24 ± 0.232/12 (17%)
UCN = 462.06 ± 5.201.51 ± 2.7920/46 (43%)0.004
Only femalesN = 313.01 ± 6.152.17 ± 3.2015/31 (48%)0.030
Only malesN = 150.01 ± 0.010.14 ± 0.345/15 (33%)
ControlsN = 491.01 ± 1.911.09 ± 2.568/49 (16%)
Only femalesN = 371.31 ± 2.121.06 ± 2.068/37 (22%)
Only malesN = 120.09 ± 0.111.15 ± 3.830/12 (0%)

CD: Crohn's disease; UC: ulcerative colitis.

  • a The occurrence rate of G. vaginalis biofilm versus controls and testing of the female IBD groups versus female controls, all with the Chi-square test.

As previously described epithelial cells, leukocytes, cell debris and fungi could be easily recognized in DAPI (see also Fig. 3). 6 All urine samples contained enough epithelial cells for the analysis of the mean concentrations and standard deviations of epithelial cells and leukocytes (see Table 3), and the presence of G. vaginalis biofilm (see Tables 3 and 4).


The figure shows a cluster of desquamated epithelial cells as well as a large number of fungi (green arrows: hyphae; red arrow: spores) at a magnification of 400 (DAPI, blue fluorescence).

View this table:

Clinical data of the participants and G. vaginalis biofilms.

Number of patients424649
Age range (mean age)20–73 (44.1)19–88 (48.4)20–79 (52.0)
Past or present perianal fistula14 ptNANA
G. vaginalis biofilm positive50%, p = 0.261
Primary sclerosing cholangitisNA12 ptNA
G. vaginalis biofilm positive 58%, p = 0.227
Current medication with corticosteroids25 pt24 pt6 pt
G. vaginalis biofilm positive52%, p = 0.02467%, p = 0.0010%, p = 0.248
Steroid refractory/dependent disease19 pt15 ptNA
G. vaginalis biofilm positive68%, p < 0.00180%, p < 0.001
Current medication with corticosteroids except patients with steroid refractory/dependent disease6 pt9 ptNA
G. vaginalis biofilm positivea 0%, p = 0.27044%, p = 0.129
Remission (CDAI < 150, CAI < 5)22 pt27 ptNA
Mild to moderate activity (CDAI 150–450; CAI ≥ 5–10)18 pt15 ptNA
Severe activity (CDAI > 450; CAI > 10)2 pt4 ptNA
  • a p-Values as compared to patients without current corticosteroid medication.

There was no statistically significant difference of epithelial cell concentrations or leukocyte concentrations between the CD, UC and the control group. Analyzing the data of the female participants alone, we found no significant difference in the concentrations of epithelial cells and leukocytes.

The samples of the female IBD patients and female controls contained a significantly higher number of epithelial cells (mean: 2.3 × 105/ml) and leukocytes (mean: 3.1 × 105/ml) than the samples of male IBD patients and male controls (mean epithelial cells: 0.4 × 105/ml; mean leukocytes: 0.2 × 105/ml) (p < 0.001).

4.2 Total bacterial concentration

Since the identification of bacteria is often difficult, we followed the requirements for the identification of bacteria as suggested by Swidsinski.21 The urines of the CD patients contained 2.69 ± 9.03 × 106 bacteria per ml, the urines of the UC patients contained 3.17 ± 10.13 × 106 bacteria per ml and the urines of the controls 2.05 ± 6.81 × 106 bacteria per ml. The bacterial concentrations were not significantly different between the groups (CD versus controls p = 0.091; UC versus controls p = 0.181).

4.3 Semiquantitative concentrations of bacterial subgroups

The statistical analysis with the Mann–Whitney U-test of the semiquantitative concentrations of Atopobium vaginae, Enterobacteriaceae, Lactobacillus/Enterococcus, Bacteroidaceae, Bifidobacterium spp. and Streptococcus spp. showed no statistical significant differences between CD patients, UC patients and controls.

4.4 G. vaginalis biofilm

A typical G. vaginalis biofilm attached to desquamated epithelial cells is shown in Fig. 1, while the data regarding G. vaginalis biofilms is shown in Tables 3 and 4. G. vaginalis biofilm was significantly more common in the CD patients (p = 0.019) and in the UC patients (p = 0.004) than in controls (see Table 3). There was no significant difference in the occurrence rate of G. vaginalis biofilms between CD and UC patients (p = 0.608). There were significant differences between female CD patients and female controls (p = 0.020) and female UC patients and female controls (p = 0.030).

The composition and spatial distribution of the biofilm attached to epithelial cells in IBD patients did not differ from the biofilm typical for bacterial vaginosis. The biofilm was composed of several bacteria, such as A. vaginae, Lactobacillus spp. and Bacteroides spp. with G. vaginalis as the predominant bacterium. Urines positive for G. vaginalis biofilm were significantly more often positive for A. vaginae than samples without evidence of G. vaginalis biofilm (p < 0.001).

We could not verify a connection between current disease activity assessed by CDAI or CAI and the occurrence of G. vaginalis biofilm using Spearman's rank correlation coefficient (CD: p = 0.385; UC: p = 0.492). There was also no evidence for a link between current medication and G. vaginalis biofilms. There was no correlation between current steroid intake analyzing the patients without steroid refractory/dependent disease only and G. vaginalis biofilm. There was a significant correlation in CD patients between younger age and G. vaginalis biofilms (p = 0.042), but no significant correlation in UC patients (p = 0.053) or in the controls (p = 0.539). The urines of IBD patients with steroid refractory/dependent disease were more often positive for G. vaginalis biofilms than those of patients without steroid refractory/dependent disease. We found that the urines of IBD patients with steroid refractory/dependent disease (either CD or UC) showed in 68 to 80% a G. vaginalis biofilm, while only 16% of the control urines were positive for such a biofilm.

Thirty-three percent of CD patients had a history of a prior or present fistula. Fifty percent of the fistula-positive CD patients and 35% of fistula-negative patients were G. vaginalis biofilm positive (p = 0.261).

5 Discussion

We investigated spontaneously voided urine for bacteria attached to desquamated epithelial cells in order to characterize urogenital microbiota in IBD patients which may explain the higher preterm birth rate in CD patients. All urine samples from IBD patients and controls contained large numbers of desquamated epithelial cells. The concentrations of desquamated epithelial cells were markedly lower in males than in females.

We found that significantly more CD and UC patients were positive for G. vaginalis biofilm compared to controls. Forty-seven percent of females with CD, 48% of females with UC, and 22% of female controls were positive for G. vaginalis biofilm. The occurrence of a G. vaginalis biofilm was not associated with current medication (except for patients with steroid refractory/dependent disease) or current disease activity, indicating that those differences in the urogenital flora are not only short-term changes but are stable ones. The urines of patients with steroid refractory/dependent disease were significantly more often positive for G. vaginalis biofilms than those without steroid refractory/dependent disease.

We could only show a significant difference in the occurrence rates of G. vaginalis in females. We did not see a difference in males, but we unfortunately did not recruit enough males to show significance.

A polymicrobial biofilm predominantly consisting of G. vaginalis, Lactobacilli and A. vaginae is characteristic for bacterial vaginosis.6,8,22 The abnormal microbiota in bacterial vaginosis is associated with poor reproductive outcomes and vaginal discharge. In the USA, 800,000 pregnancies are affected by bacterial vaginosis every year.23 Several studies demonstrated an association of bacterial vaginosis and spontaneous abortion, intrauterine infections, premature ruptures of membranes, and preterm labor and delivery.2427 A study using data from the Swedish national registry showed an increased risk of preterm delivery and low birth weight among pregnancies with CD and UC.28 Dominitz et al.29 provided evidence of an association between maternal IBD and risk of adverse neonatal outcomes. Mahadevan et al.30 showed that IBD disease activity, even moderate to high activity, was not predictive for poor newborn outcome. So far, there has not been a clear explanation why CD and UC are associated with preterm delivery since medication seems to be ruled out.3134 The significant higher rate of G. vaginalis biofilms in IBD patients suggests a possible link between bacterial vaginosis and IBD and may partially explain why preterm delivery is more frequent in patients with IBD. Still, further studies are required to investigate the frequencies of bacterial vaginosis and G. vaginalis biofilms in pregnant IBD patients.

We could not show a difference in the concentrations of the bacterial subgroups other than G. vaginalis and no significant difference in the total bacterial concentration. We found no significant association with current fistulas or a history of fistulizing disease and G. vaginalis biofilms. It is probable that the changes in the urogenital microbiota in IBD patients are not only due to simple overgrowth of perianal bacteria. It is likely that there is a more specific lack of the epithelial urogenital immunity in patients with IBD.

It appears that immunomodulatory medication is an important factor for the development of G. vaginalis biofilms. We could show that current medication with steroids (except the patients with steroid refractory/dependent disease) or immunomodulatory therapy like azathioprine is not associated with G. vaginalis biofilm. We showed that a steroid refractory/dependent disease is significantly linked to the occurrence of such a biofilm. Those results may suggest that longer lasting steroid treatment predisposes the patient to G. vaginalis biofilms. We need to point out that the number of patients with refractory/dependent disease is small and we are not aware of another study that showed that long-term steroid intake is associated with G. vaginalis biofilms and/or bacterial vaginosis.

The origin of G. vaginalis in patients with IBD remains unclear.

The conclusion of this study is that IBD and urinary G. vaginalis biofilms are associated, suggesting a lack of immunity of the urogenital epithelium. G. vaginalis biofilms are believed to be one of the reasons responsible for bacterial vaginosis and could be an important factor to explain the higher incidence of poor reproductive outcome found among patients with IBD. Furthermore, the impact of long-term steroid intake remains to be determined. Assessing the frequency of G. vaginalis biofilms in patients with other diseases that require long-term steroid medication, such as rheumatoid arthritis or other autoimmune diseases may help.

Conflict of interest

The authors certify that there is no conflict of interest with any financial organization.


The authors thank Alexander Swidsinski, MD for the use of his laboratory, his helpful discussions, insights and expertise.

The study was not supported by any sponsor.

JS and YD carried out the studies and the sample analyses and data analysis and drafted the manuscript. VLB participated in the design of the study and helped to draft the manuscript. All authors read and approved the final manuscript.


View Abstract