Bacterial vaginosis (BV), the most common cause of vaginal symptoms in reproductive age women, is characterised by decreased numbers of vaginal lactobacilli and increased concentrations of multiple anaerobic or facultative bacteria including Gardnerella vaginalis, Mobiluncus, and Prevotella species.1 These abnormal vaginal flora patterns have been linked to increased susceptibility for preterm delivery and low birth weight as well as the acquisition of HIV and herpes simplex virus type 2 infections.2^–4 Despite increasing evidence for the validity of these associations, the underlying aetiologies for these BV-associated adverse outcomes remain poorly understood.

Understanding of the vaginal immune response to BV is also incomplete. In contrast to the inflammatory responses induced by vaginal Trichomonas vaginalis or Candida infections, the large increases in Gram-negative bacteria characteristic of BV typically produce minimal overt inflammation. Multiple investigations have consistently shown BV to be associated with increased genital tract concentrations of the proinflammatory cytokine IL-1β.5^–7 The measurement of other proinflammatory cytokines, however, including tumour necrosis factor (TNF) and IL-6, yielded inconsistent results.8^–11

Similarly incomplete is our understanding of the effect of exogenous sex steroids on vaginal immunity. Although some earlier studies suggested that hormonal contraceptive use may alter susceptibility to a number of common genital tract infections, results have been inconsistent. Although prospective cohort studies have linked oral and injectable hormonal contraceptive use to the increased acquisition of Chlamydia trachomatis, Neisseria gonorrhoeae, HIV-1, and increased vaginal shedding of herpes simplex virus type 2,12^–15 a recent large multicentre prospective cohort study reported no association between hormonal contraceptive use and HIV acquisition.16

The first objective of the current study was to compare the cervical secretion concentrations of a number of cytokines/chemokines among women with BV with the levels detected in these same women after successful BV treatment. The second objective was to examine the relationship between current oral or injectable hormonal contraceptive use and the BV-associated inflammatory response. We hypothesised that BV would be related to increased concentrations of a number of pro-inflammatory cytokines/chemokines in the lower genital tract, and that the pro-inflammatory response to BV would be diminished among women using hormonal contraceptives.

METHODS

Women 14–45 years of age were recruited from adolescent and adult health clinics in Pittsburgh, PA and Seattle, WA into a randomised, double-blind investigation of the use of oral metronidazole and Lactobacillus crispatus intravaginal capsules for BV treatment. Approval for this investigation was obtained from Institutional Review Boards from the Children’s Hospital of Pittsburgh and the Universities of Pittsburgh and Washington. All eligible women received 2 g metronidazole by mouth, and were then given gelatin capsules containing either 108 colony-forming units of L crispatus or placebo. Study participants were instructed to insert these capsules intravaginally twice a day for three days and to return for follow-up evaluation in one month.

Women in the current investigation included those who fulfilled both Amsel and vaginal Gram stain criteria for the diagnosis of BV at enrollment and had no vaginal symptoms and normal vaginal flora at the one-month follow-up visit. Exclusion criteria for the randomised clinical trial included pregnancy, the presence of concomitant T. vaginalis or yeast vaginitis, vaginal bleeding, the use of any intrauterine devices, or the use of any antimicrobial agents (vaginal or systemic) in the seven days before enrollment. To minimise one potential confounding factor of an investigation of the local inflammatory response to BV, women diagnosed with T vaginalis, C trachomatis, and N gonorrhoeae at enrollment were also excluded from the current analyses.

After securing written informed consent, demographic and behavioural enrollment interview data were collected from each study participant. Vaginal fluid specimens were collected for the diagnosis of T vaginalis infection (InPouch; BioMed, White City, OR, USA) and for the diagnosis of BV using Gram-stained vaginal smears. Employing previously standardised criteria, normal (lactobacilli predominant) vaginal flora received a score of 0–3; intermediate flora (reduced lactobacilli), 4–6; and BV, 7–10.17 Cervical secretions were then collected for the quantification of cytokine/chemokine concentrations by the placement of a sterile polyester fibre swab (Dacron; Thermo Fisher Scientific Inc, Waltham, MA, USA) in the cervical introitus of each study participant for approximately 10 seconds. These swabs were then placed in cryovials containing 400 μl phosphate buffered saline and were stored at −20°C. A second cervical secretion swab specimen was then collected for the detection of C trachomatis and N gonorrhoeae (ProbeTec; Becton Dickinson, Sporks, MD, USA).18

Quantitation of cytokine/chemokine concentrations

Cervical secretion specimens were thawed at room temperature. Swab and diluent were placed in a centrifuge filter unit (Spin-X; Costar, Cambridge, MA, USA), and centrifuged at 12 000 rpm for 20 minutes. Commercially available multiplexed fluorescent bead-based immunoassays (Beadlyte Human Multi-cytokine Beadmaster Kit; Upstate, Temecula, CA, USA) were used to determine cytokine/chemokine (IL-β, TNF, INF-γ, IL-2, IL-4, IL-6, IL-8, IL-10, granulocyte macrophage colony-stimulating factor (GM-CSF), and macrophage-inflammatory protein 1 alpha (MIP-1α)) concentrations in accordance with manufacturer’s instructions. Briefly, 50 μl aliquots of a known standard concentration or sample were added in duplicate to individual wells of a 96-well multiscreen plate. Multiplexed fluorescent capture beads were added to each well, and plates were incubated on a plate shaker in the dark at 4°C overnight. Plate wells were washed with assay buffer, and the beads re-suspended in 75 μl of this solution. A biotinylated reporter solution was added to each well, and each plate was incubated in the dark at room temperature for 90 minutes. A streptavidin–phycoerythrin solution was then added for 30 minutes. Plates were washed again, and beads re-suspended in 125 μl of the assay buffer. Beads were analysed on Luminex 100 (Luminex, Austin, Texas, USA) multiplex instrumentation, and the cytokine/chemokine mean fluorescent intensities from each specimen were extrapolated from eight-point standard curves using four-parameter logistic algorithms (Applied Cytometry Systems, Sacramento, California, USA).

Statistical analyses

Statistical analyses were performed using SPSS 13.0.1 statistical software (SPSS Inc., Chicago, Illinois, USA). Whereas graphic displays of the cytokine/chemokine concentration data did not follow a normal distribution curve, log₁₀-transformed values of the data did. Therefore, log₁₀-transformed values were used in all subsequent analyses. Sample sizes were not informed by any type of power calculation as all women from the original investigation who met the eligibility criteria for inclusion in these subanalyses were included. Paired Student’s t-tests were used to compare cytokine/chemokine levels among women with BV at enrollment and normal vaginal flora at follow-up, and to determine whether the ratios of key cytokine concentrations differed between these two visits. The Wilcoxon test was used to evaluate whether the distribution of the geometric mean levels of the cytokine/chemokine concentrations differed between baseline and return visits. Unpaired Student’s t-tests were used to compare the cervical cytokine/chemokine response to BV, as well as the change in cytokine/chemokine levels between enrollment and follow-up, among women who used hormonal contraceptives with those who did not. Unpaired Student’s t-tests were also used to compare cervical cytokine/chemokine concentrations detected at follow-up among women who received intravaginal L crispatus capsules compared with those who received placebo. Because there were no detectable differences in the cervical cytokine/chemokine concentrations between the two treatment groups, data from both investigational arms were combined for subsequent analyses.

RESULTS

A total of 81 study participants had BV at enrollment and normal vaginal flora at their one-month follow-up visit. Among these women, 54 (67%) were self-described as black, 22 (27%) as white, and five (6%) were self-described as Hispanic, Asian, native American, or multiethnic. The mean age at enrollment was 23 years, whereas the majority of these women (65/81; 80%) reported sexual activity with only one male partner during the 30-day time period before their enrollment. Sixty per cent (49/81) of the study participants denied any type of hormonal contraceptive use during this investigation, whereas the remainder (32/81; 39.5%) reported the use of either an oral or an injectable hormonal contraceptive.

A comparison of the mean log cytokine/chemokine concentrations measured in the cervical secretions of women who had BV at enrollment compared with the concentrations detected from these same women who had normal vaginal flora at their one-month follow-up visit is shown in fig 1. Statistically significantly higher concentrations of IL-1β, TNF, interferon-γ, IL-2, IL-4, IL-10, and GM-CSF were associated with the diagnosis of BV compared with the concentrations measured when normal vaginal flora were detected (p = 0.007, p<0.001, p = 0.01, p<0.001, p<0.001, p<0.001, and p<0.001, respectively). The increases in cytokine/chemokine concentrations associated with BV were, however, relatively small compared with the constitutive levels detected in women with normal vaginal flora. In addition, women who had resolution of BV at their one-month follow-up had statistically higher levels of the chemokine IL-8 compared with the levels detected in the presence of BV (p<0.001). As seen in fig 1, no statistically significant differences were detected in the measured levels of IL-6 and MIP-1α between the two study visits.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Mean log cytokine/chemokine concentrations as measured by multiplexed fluorescent bead-based immunoassays from the cervical secretions of women with Gram stain diagnosis of bacterial vaginosis (BV) at enrollment and normal vaginal flora at follow-up. GM-CSF, Granulocyte macrophage colony-stimulating factor; MIP-1α, macrophage-inflammatory protein 1 alpha; TNF, tumour necrosis factor.

Although there were many statistically significant increases in the levels of both pro and anti-inflammatory cytokines in the lower genital tract of women with BV, the overall balance of these two types of molecules appears to be maintained. A comparison of the ratio of the sum of the log concentrations of TNF, interferon-γ, and IL-1 (three key pro-inflammatory molecules) with the sum of the log concentrations of IL-4 and IL-10 (key anti-inflammatory molecules) in the presence of BV or normal vaginal flora found no statistically significant differences between the two groups (1.71 (0.34) versus 1.75 (0.38); respectively; p = 0.4). Likewise, a comparison of the differences in ordinal ranking of the measured cytokine/chemokine levels between the enrollment and follow-up visits also demonstrated no significant differences between groups (p = 0.4).

Figure 2 compares the mean log cytokine/chemokine concentrations from the cervical secretions of study participants with either normal vaginal flora (A) or BV (B) stratified by hormonal contraceptive use. As demonstrated in part A of this figure, there were no statistically significant differences in the levels detected between hormonal contraceptive users and non-users for any of the chemokines/cytokines measured in the presence of normal vaginal flora. As shown in part B of fig 2, the presence of BV was associated with statistically significantly lower levels of TNF, interferon-γ, IL-2, IL-4, GM-CSF, and MIP-1α (0.005, 0.05, 0.009, 0.03, 0.008, and 0.04, respectively) in women who reported the current use of oral or injectable contraceptives compared with those study participants who denied hormonal contraceptive use. Further analysis of the data demonstrated that women who were not using any method of hormonal contraception had a statistically significantly greater change in the amount of TNF detected between enrollment and follow-up compared with the change in TNF concentration detected between these two study visits among women currently using hormonal contraceptives (0.61 (0.92) versus 0.13 (0.76) pg/ml; respectively, p = 0.02; data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Mean log cytokine/chemokine concentrations as measured by multiplexed fluorescent bead-based immunoassays from the cervical secretions of women with (A) normal vaginal flora or (B) bacterial vaginosis (BV) diagnosed by Gram stain analysis of vaginal smears stratified by hormonal contraceptive (HC) use. GM-CSF, Granulocyte macrophage colony-stimulating factor; MIP-1α, macrophage-inflammatory protein 1 alpha; TNF, tumour necrosis factor.

DISCUSSION

Our study provides important new information about the response to both normal and abnormal vaginal flora in the lower genital tract. First, the lack of an inflammatory response to endogenous flora in the normal lower genital tract, despite surprisingly high constitutive levels of many important inflammatory mediators, suggests that the balance, rather than absolute levels, of these inflammatory molecules may be an important regulator of the immune response. Second, we demonstrate statistically significant increases in a number of cytokines/chemokines in the lower genital tract of women with BV. Furthermore, a number of these increases, interferon-γ, IL-2, IL-4, and GM-CSF, had previously been unreported. These increases are, however, small in comparison with the constitutive levels detected in response to normal vaginal flora. The overall balance between pro and anti-inflammatory molecules is maintained in women with BV, despite increases in both types of mediators, and this may contribute to the absence of overt inflammatory signs in this condition. Finally, our results demonstrate that current oral or injectable hormonal contraceptive use is associated with a differential inhibition of the local immune response to BV in comparison with the response to BV among women not using hormonal contraceptives.

As mentioned previously, BV is characterised by a large overgrowth of facultative anaerobic Gram-negative bacteria. TNF is the major proinflammatory cytokine induced by the presence of such organisms,19 whereas increases in IL-1β are triggered by TNF and the Gram-negative bacterial cell wall component lipopolysaccharide.20 It was, therefore, not surprising to find increased levels of TNF and IL-1β in the cervical secretions of women with BV. Similarly, the detection of increased interferon-γ and IL-2 levels in the cervical secretions of women with BV, although novel, was not unexpected. Interferon-γ is a potent inducer of macrophage activation essential for the expression of inducible nitric oxide synthase, the enzyme responsible for generating antimicrobial nitric oxide, and IL-2 is an important inducer of T and B lymphocyte activation, proliferation, and differentiation in response to extracellular pathogens.21 22 Further research is needed to determine whether the increased levels of these potent inflammatory mediators in the cervical secretions of women with BV contribute to the numerous adverse events associated with BV.

Inflammatory signs are minimal in BV despite considerable colonisation by facultative or strictly anaerobic microorganisms, so it is possible that increased levels of IL-4 and IL-10, concomitant with the above-mentioned increases in a number of pro-inflammatory cytokines, contribute to this lack of overt inflammation. Many of the proinflammatory actions of interferon-γ are inhibited by IL-4, whereas the synthesis of IL-1β is inhibited by both IL-4 and IL-10. Cauci et al.23 have previously demonstrated higher concentrations of antibodies specific for a constitutively expressed G vaginalis toxin among women with BV, and IL-4 is an important mediator of humoral immunity. Therefore, levels of IL-4 may increase in women with abnormal vaginal flora, in association with the establishment of the humoral response needed to combat the extracellular pathogens associated with BV. In addition to the minimal signs of inflammation in women with BV, there are typically no increases in the number of vaginal neutrophils. Consistent with this observation, local levels of the neutrophil chemoattractant, IL-8, are not increased in women with BV.5 24 Our detection of significantly lower levels of IL-8 in women with BV is still somewhat surprising. Further research is needed to elucidate the mechanisms by which IL-8 production, induced in nearly all cell types that encounter TNF, IL-1β, or lipopolysaccharide, is either inhibited or selectively degraded among women with BV.

When our results were stratified by current hormonal contraceptive use, we found no statistically significant differences in the cervical secretion levels of any of the measured cytokines/chemokines in the presence of normal vaginal flora (fig 2A). The levels of a number of these cytokines (including TNF and interferon-γ) and chemokines (GM-CSF and MIP-1α) were, however, significantly lower among women with BV who reported current hormonal contraceptive use compared with women with BV who denied such use (fig 2B). Conversely, hormonal contraceptive use did not significantly reduce levels of the key anti-inflammatory cytokine IL-10. As TNF, interferon-γ, and GM-CSF are critical for the activation of the innate immune response to extracellular pathogens, our data suggest the possibility of an association between hormonal contraceptive use and impaired genital tract immunity. Hormonal contraceptive use was also associated with a statistically significant lesser change in TNF levels between the enrollment and follow-up visits compared with the amount of change detected between visits among women who denied their use. Although similar trends existed for several other pro-inflammatory molecules, including IL-2, interferon-γ, and GM-CSF, these other changes did not achieve statistical significance (data not shown). Therefore, a more definitive conclusion regarding the effects of hormonal contraceptive use on the local immune response to genital tract pathogens will require larger studies specifically designed to answer this question.

Although our study provides important new information on the immune response to BV and the modulation of this response by hormonal contraceptive use, it cannot elucidate which cell types in the lower genital tract produced the various cytokines/chemokines measured. This information is needed for a better understanding of the mechanisms of BV pathogenesis, and may have important clinical implications for the treatment of BV as well as for decreasing the frequency of BV-associated adverse events. Furthermore, as the objectives of the parent investigation did not include an examination of the immunomodulatory effects of hormonal contraceptive use, our data are underpowered to delineate any distinct effects of oral or injectable hormonal use on reproductive tract immunity. It seems likely that oestrogen and progestin-containing oral contraceptives and progestin-only-containing injectable contraceptives will have both similar and disparate effects on both innate and adaptive immunity, but further investigation will be needed to clarify these distinctions. We are also unable to determine the clinical significance of the altered local immune response to BV that was associated with hormonal contraceptive use. As hormonal contraceptives are used by more than 100 million women worldwide,25 more research is needed to provide more definitive answers to this important research question.

Our investigation does provide an example of how an immune response can be regulated through a complex network of pro and anti-inflammatory mediators, and how the overall response to most pathogens is the result of the net effect of these interactions. Although BV is not associated with signs of overt inflammation, it does invoke a significant local inflammatory response. Further elucidation of the local immune response to this common condition, associated with so many adverse sequelae, represents both an important opportunity to obtain a better understanding of the net effects of these interactions and to improve public health.