Article Text

Download PDFPDF

Short report
Cross-sectional analysis of Toll-like receptor variants and bacterial vaginosis in African–American women with pelvic inflammatory disease
  1. Brandie D Taylor1,
  2. Toni Darville2,
  3. Robert E Ferrell3,
  4. Roberta B Ness4,
  5. Sheryl F Kelsey1,
  6. Catherine L Haggerty1
  1. 1Department of Epidemiology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
  2. 2Department of Pediatrics, University of North Carolina Chapel Hill, Chapel Hill, North Carolina, USA
  3. 3Department of Human Genetics, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
  4. 4University of Texas School of Public Health, Houston, Texas, USA
  1. Correspondence to Dr Brandie DePaoli Taylor, Department of Epidemiology and Biostatistics, School of Public Health, Texas A&M Health Science Center, 211 SRPH Administration Building, TAMU 1266, College Station, TX 77843-1877, USA; Taylor{at}


Objective Bacterial vaginosis (BV) is a common condition associated with serious complications including pelvic inflammatory disease (PID). However, the pathogenesis of BV is poorly understood. Toll-like receptors (TLR) are responsible for microbial recognition and elimination through inflammatory responses. TLR variants have been implicated in infectious and inflammatory diseases and may be involved in BV pathogenesis. We conducted a cross-sectional study to determine if TLR variants are associated with BV.

Methods Logistic regression was used to test associations between 14 variants assayed in 6 genes (TLR1, TLR2, TLR4, TLR6, TIRAP and MyD88) and BV/intermediate flora among 192 African–American women with clinical PID from the PID Evaluation and Clinical Health (PEACH) Study. Additionally, we examined associations between variants and endometrial BV-associated anaerobes. To account for multiple comparisons a permutated p<0.003 was used to determine statistical significance.

Results African–American women with PID carrying the AA genotype for TLR2 SNP rs1898830 had a threefold increased rate of BV/intermediate flora (OR 2.9, 95% CI 1.2 to 7.3). This was not significant after accounting for multiple comparisons (p=0.0201). TLR2 variants rs1898830, rs11938228 and rs3804099 were associated with increased endometrial anaerobic gram-negative rods (p=0.0107, p=0.0076 p=0.0121), anaerobic non-pigmented Gram-negative rods (p=0.0231, p=0.0083, p=0.0044), and anaerobic Gram-positive cocci (p=0.0596, p=0.0640, p=0.1459).

Conclusions Among African–American women with PID, we observed trends between TLR2 variants, BV/intermediate flora, and BV-associated microbes. This provides some insight into BV pathogenesis. As not all BV-associated microbes may lead to pathology, future studies should focus on associations between TLR variants and individual BV-associated microbes.

  • Bacterial Vaginosis
  • Epidemiology (General)
  • Pelvic Inflammatory Disease
View Full Text

Statistics from


Bacterial vaginosis (BV) is a prevalent condition that occurs when hydrogen-peroxide producing lactobacilli are decreased and replaced by anaerobic and facultative aerobic bacteria including Gardnerella vaginalis, Ureaplasma urealyticum, Atopobium vaginae, Leptotrichia species, and three new species called BV-associated bacteria 1-3 (BVAB-1, BVAB-2, and BVAB-3).1

BV is associated with pelvic inflammatory disease (PID), the infection and inflammation of the fallopian tubes (salpingitis) and endometrium (endometritis).2 Additionally, clusters of BV-associated anaerobic bacteria are independently associated with histologic endometritis.2 BV is present in 40–53% of women with clinical PID,2 and complications may occur when BV-associated microbes ascend to the upper genital tract.

Toll-like receptors (TLR) protect against pathogens and induce inflammatory responses. TLRs 1, 2 and 6 are expressed in the fallopian tubes, endometrium, vagina and cervix.3 TLR4 expression is higher in the upper genital tract, and although results are controversial, TLR4 has been reported in epithelial cells of the endocervix and vagina.3 In women with PID, TLR variants that recognise BV-associated microbes may increase the risk of pathology, through increased inflammation or increased susceptibility to other pathogens. BV-associated microbes possess molecules recognised by TLRs including lipopolysaccharides and lipopeptides. Furthermore, genital mucosal fluids from women with BV induce TLR4 and TLR2 mRNA expression, NF-κB signalling, and inflammation in vitro.4 Studies among pregnant women primarily of European ancestry, have not found significant associations between variants in the TLR1, TLR2, TLR4 and TLR6 genes and BV.5 ,6 An association between the BV-associated microbe, A vaginae, and a TLR1 variant was reported.5 By contrast, variants in the TLR1, TLR4 and TLR9 genes were associated with BV among 159 HIV-1-positive African–American adolescents.7 It is well known that the risk of developing BV differs by race and is significantly higher in African–American women. We have previously found that African–American women with PID are significantly more likely to carry TLR variants that alter signalling compared to Caucasian women with PID.8

Our objective was to assess if TLR variants were associated with genital tract flora alterations determined by Nugent's criteria among women with clinical PID. Further, we determined if TLR variants were associated with endometrial BV-associated microbes.

Materials and methods

This is a cross-sectional analysis nested within the PID Evaluation and Clinical Health (PEACH) study, a randomised clinical trial comparing inpatient and outpatient treatment in preventing long-term complications among 831 women with clinical PID.9 In PEACH, gynaecological exams, baseline questionnaires and follow-up phone interviews were administered. We obtained data on TLR genotypes from a previous PEACH substudy of 205 African–American who had stored blood samples available. Only women who had data on Gram-stain results were included in our analysis for a total of 192 non-Hispanic African–American women. Among these women, 127 had BV, 39 had intermediate flora, and 26 had normal flora scored by Nugent's criteria.10 Additionally, 93 women had facultative and anaerobic isolate culture performed for Lactobacillus species, anaerobic pigmented and non-pigmented Gram-negative rods, G vaginalis, group B streptococcus, enterococcus species, Escherichia coli, candida species, Mycoplasma hominis and U urealyticum. The University of Pittsburgh Institutional Review Board approved this study.

For each TLR gene, HapMap (http// was used to identify SNPs with a minor allele frequency ≥0.05 and blocks of linkage disequilibrium were then identified. A total of 14 tagging SNPs (TagSNPs) from TLR1 (rs5743618, rs4833095), TLR2 (rs3804099, rs11938228, rs1898830), TLR6 (rs1039559, rs5743810, rs3775073), TLR4 (rs4986790, rs4986791, rs5030728), MyD88 (rs4988457), and TIRAP (rs3802813, rs7932976) were genotyped by fluorescence polarisation. PCR reaction included 2.5 μL of 1× PCR buffer (invitrogen), 1.0 μL of MgCl2, 4 μL of dNTPs, 1.5 μL of each primer, 0.1 μL of taq polymerase (invitrogen), and 13.4 μL of dH2O. PCR products were resolved by electrophoresis (3% agar gel) and visualised under UV light after ethidium bromide staining. Genotypes were assigned by direct comparison with controls of sequence-confirmed genotypes, and a 5% random resample was included for consistency of the genotyping. All SNPs were tested for deviations from Hardy–Weinberg equilibrium.

Logistic regression was used to examine associations between genotypes (exposure) and BV and intermediate flora which were collapsed into one category (outcome). To correct for multiple comparisons, permutation tests were used and a p value ≤0.003 was statistically significant. In a subset of 93 women, we examined the association between genotypes and frequencies of BV-associated microbes colonised in the endometrium. Possible confounders including chlamydia, gonorrhoea and douching were considered but did not change our results.


Women with BV were similar to women with normal flora; age <25 (32.3% vs 38.5%), married (5.7% vs 4.0%), and uninsured (48.3% vs 48.0%). However, chlamydia (45.7% vs 50.0%), gonorrhoea (36.0% vs 15.0%), and douching (51.2% vs 26.9%) differed between women.

Women who carried the TLR2 rs1898830 AA genotype had increased odds of BV/intermediate flora (OR 2.9, 95% CI 1.2 to 7.3; table 1). This was not statistically significant after accounting for multiple comparisons. No other TLR variants were significant.

Table 1

ORs and 95% CI for BV/intermediate flora and anaerobic bacteria by TLR2 genotypes in African–American women with clinical PID

Compared with other genotypes, women who carried the TLR2 rs1898830 AA genotype had increased odds of endometrial anaerobic Gram-negative rods (OR 5.4, 95% CI 1.4 to 19.6; p=0.0107) and anaerobic Gram-positive cocci (OR 3.7, 95% CI 1.0 to 14.4; p=0.0596). Women who carried the TLR2 rs11938228 CC genotype (OR 4.9, 95% CI 1.5 to 15.7; p=0.0076) and rs3804099 CC genotype (OR 3.3, 95% CI 1.3 to 8.3; p=0.0121) had increased odds of endometrial anaerobic Gram-negative rods. Non-pigmented Gram-negative rods showed the strongest association with all three TLR2 variants; rs1898830 (OR 8.1, 95% CI 1.3 to 49.5; p=0.0231), rs11938228 (OR 10.9, 95% CI 1.9 to 65.1; p=0.0083), rs3804099 (OR 4.6, 95% CI 1.6 to 13.0; p=0.0044). Results did not change when we considered possible confounders including chlamydia, gonorrhoea and douching in our statistical models.


This study did not demonstrate an association between TLR variants and BV/intermediate flora after accounting for multiple comparisons. These results are similar to those reported among cohorts of pregnant women.5 ,6 In a non-pregnant cohort of 159 HIV-1-positive African–Americans, BV (Nugent's criteria) was associated with TLR1 rs5743612 (HRadjusted 1.56), TLR4 rs4986790 (HRadjusted 1.47), and TLR9 rs187084 (HRadjusted 1.52), while TLR2 rs1898830 was associated with BV diagnosed by Amsel's criteria (HRadjusted 1.89).7 We did find that the TLR2 rs1898830 AA genotype increased the odds of BV/intermediate flora threefold. However, our study had a small sample size that limited our power. Cytokine response differs depending on the microbe present, and microbes elicit different disease mechanisms (ie, TLRs may not be involved in the pathology of all BV-associated microbes). Each patient in our study who had BV/intermediate flora had a distinct vaginal flora, which was likely comprised of pathogenic and non-pathogenic bacteria. Therefore, associations between TLRs and BV may be biased towards the null.

Our observed associations between TLR2 variants and non-pigmented gram-negative rods had large effect sizes and borderline significant p values. TLR variants may play a role in the development of PID following BV through increased risk of endometrial colonisation of anaerobic bacteria. All women in our study had clinically suspected PID, although not all these women will have lasting pathology. In a PEACH substudy, associations between anaerobic bacteria and endometritis were observed after adjusting for chlamydia and gonorrhoea.2 This may suggest that BV-associated anaerobic bacteria contribute to upper genital tract inflammation and long-term sequelae in a subset of women with PID, possibly through altered immune responses via stimulation of TLR2 genes. However, long-term outcomes will need to be assessed in a larger cohort of women.

Unfortunately, little is known about the function of these TLR2 variants. rs1898830 and rs11938228 are both located in an intron, and rs3804099 results in a synonymous variation. Therefore, these variants may be in linkage disequilibrium with a functional variant. In a Chinese population, the rs1898830 G allele and the rs3804099 C allele increased production of anti-inflammatory IL-10 and pro-inflammatory TNF-α and IL-8.11 As the immunological mechanisms of these TLR2 variants are not well understood, future studies should examine the effect of TLR variants on genital tract cytokine expression.

Our data were obtained from a large, multicenter, prospective randomised clinical trial, with comprehensive demographic, clinical and obstetric measurements. Not all women in the PEACH study had blood samples available for analyses. However, important demographic and clinical characteristics between women with and without blood samples did not differ. All women had clinically suspected PID, and we relied on an internal control group, thus, our results may be biased towards the null. Furthermore, this limits our generalisability. We were able to examine associations between TLR variants and BV in a cohort of African–American women, who are at increased risk for BV and PID.

Among African–American women with PID, TLR2 variants may be associated with BV/intermediate flora and endometrial colonisation of BV-associated microbes. These results need to be replicated in larger cohorts. It is possible that BV-associated microbes alter the vaginal mucosa through TLR signalling and increase susceptibility to upper genital tract infection. It is unclear if these anaerobic bacteria actually increase inflammatory responses among women who carry specific TLR variants and ultimately influence outcome. As BV is polymicrobial, and the function of many TLR SNPs is unknown, the relationship between endometrial BV-associated microbes colonies and TLRs is complex and difficult to interpret. Future studies should focus on examining cytokine responses and reproductive outcomes of women who carry different TLR variants and are colonised with specific BV-associated microbes identified by molecular methods.

Key messages

  • In African-American women with clinical pelvic inflammatory disease, variants in the Toll-like receptor (TLR) 2 gene were associated with BV/intermediate flora and bacterial vaginosis (BV)-associated microbes colonised in the endometrium.

  • These results may suggest that TLR variants are involved in the pathogenesis of some BV-associated microbes. However, our sample size limited our power to detect significant associations after correction for multiple comparisons.

  • Relationships between BV-associated microbes determined by molecular methods, TLR variants, and genital cytokine expression should be explored in a larger cohort of women.


View Abstract


  • Handling editor Jackie A Cassell

  • Contributors BDT made substantial contributions to the concept, design and interpretation of the study and conducted the statistical analyses. CLH and TD made substantial contributions to the conception, design and interpretation of the study. RBN was the PI for the PEACH study and made contributions to the interpretation of the data. REF performed the TLR analysis and contributed to the interpretation of the data. SFK was involved in the original PEACH study and consulted on the statistical analyses for the current manuscript. BDT wrote the manuscript and CLH, TD, RBN, SFK, and REF critically revising the manuscript. All authors were involved in the final approval of this manuscript and agree to be accountable for all aspects of the work.

  • Funding This work was supported by the Agency for Healthcare Research and Quality (grant HS08358-05 for RBN) and the National Institute of Allergy and Infectious Diseases (grant AI054624 and grant U19 A1084024 for TD).

  • Competing interests None.

  • Ethics approval University of Pittsburgh.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.