Endogenous bacterial flora in pregnant women and the influence of maternal genetic variation


Dr MR Genc, Weill Cornell Medical College, Department of Obstetrics and Gynecology, 525 East 68th Street, New York, NY 10065, USA. Email mrg2003@med.cornell.edu


Please cite this paper as: Genc M, Onderdonk A. Endogenous bacterial flora in pregnant women and the influence of maternal genetic variation. BJOG 2011;118:154–163.

Preterm labour, premature prelabour rupture of membranes and low birth weight have all been associated with either specific maternal genital tract infections or an altered vaginal microflora during pregnancy. Factors that influence the variation in microbial–host interaction play an important role in individual susceptibility to adverse pregnancy outcomes. The innate immune responses at mucosal surfaces play a crucial role against microbial invasion. Multiple genes are responsible for the regulation of the innate immune system. Genetic polymorphisms that disrupt innate immune recognition or the responses to infectious microorganisms could explain the alterations in microflora and individual susceptibility to pregnancy complications.

Normal and abnormal vaginal flora in pregnancy and adverse pregnancy outcomes

The vaginal microflora in menarchal women (women in the reproductive age group between menarche and menopause) has been studied using direct microscopic observations, as well as quantitative and qualitative culture methods, for over 50 years. More recently, 16S rDNA deep sequencing methods have also been applied to this component of the human microbiome. In menarchal, overtly healthy women, microscopic examination of vaginal samples indicates that a predominantly Gram-positive microflora is present. Qualitative cultures indicate that the dominant genus is Lactobacillus. Lactobacillus sp., particularly H2O2-producing strains, have been cited as important indicators of a healthy vaginal environment.1 Decreased numbers of these organisms have been linked with the presence of bacterial vaginosis (BV), a ‘dysbiotic’ condition in which the typical repertoire of organisms is replaced by normally minor, often intermittent, components of the vaginal microflora. More detailed longitudinal quantitative assessments of the vaginal microflora of menarchal women have revealed that over 30 bacterial species can be isolated from healthy subjects on a regular basis, and that there may be multiple species of Lactobacillus present in the same individual.2 Equally as important is the observation that potentially pathogenic organisms, including Staphylococcus aureus, group B Streptococcus, Escherichia coli, Prevotella, Gardnerella, Mycoplasma and Ureaplasma, may also be present in low numbers in overtly healthy subjects. Under the normal acidic conditions that are present within the vaginal environment (pH 4.5 or less), these potential pathogens remain at low concentrations and do not cause obvious infectious complications. It has also been shown that the relative numbers and complexity of the microflora change during the menstrual cycle in a predictable manner, with the most complex microflora present just prior to menses and the least complex microflora present during menstrual flow.3,4 Pregnancy represents a unique environment in which the normal microbiological changes that occur during the menstrual cycle are suspended. Although a healthy vaginal microflora may remain intact during pregnancy, it is also possible for changes in this microflora to occur in an unpredictable manner that may interfere with normal gestational events.

The leading cause of infant morbidity and mortality in the USA is spontaneous preterm delivery (SPD).4–8 More than 80% of premature births occur following premature prelabour rupture of membranes (PPROM) and/or spontaneous preterm labour. SPD may be a response to an induced inflammatory event, hypothalamic–pituitary–adrenal axis activation, decidual haemorrhage or pathological uterine overdistension.9 Microbial gestational tissue invasion is a predominant inducer of intrauterine inflammation. Evidence for microbial invasion of the amniotic cavity can be obtained in up to 50% of women with inflammation and intact membranes, and a higher rate is evident in cases of PPROM.10–18 Organisms implicated in SPD include specific agents, such as Bacteroides, Prevotella sp. and genital mycoplasmas.14,19–29 BV is a condition characterised by an apparent decrease in the concentration of lactobacilli, coupled with an increase in Gardnerella vaginalis, Ureaplasma urealyticum, Mobiluncus and the concentrations of obligate anaerobes, such as Prevotella, Bacteroides and Peptostreptococcus sp. In some apparently healthy women with a normal vaginal pH who lack appreciable Lactobacillus species, other lactic acid-producing bacteria, such as Atopobium, may be present. Bacteria that are typical components of BV have the potential to release inflammatory mediators, such as phospholipases, which induce prostaglandin production and myometrial contractions.30–36 Clinical symptoms consistent with a diagnosis of BV are an elevated vaginal pH, a thin watery discharge, the presence of a volatile amine odour and the microscopic observation of bacteria adhering to cell membranes (clue cells). The evaluation of vaginal smears by Gram stain, the Nugent score, is an alternative method to assess the health of the vaginal microflora and to detect BV.37 A normal Nugent score is 0–3, with an intermediate score in the range 4–6, and a score of 7 or higher being considered as abnormal. In one study of pregnant women sampled at 20 and 30 weeks of gestation, it was shown that the quantitative counts for Lactobacillus or the H2O2 production by the Lactobacillus strains isolated did not vary significantly among the three defined categories of vaginal microflora, nor was there a replacement of Lactobacillus sp. by other species. Instead, significant increases in several genera and species, including G. vaginalis, Prevotella sp. and Peptostreptococcus sp., were observed in subjects with an abnormal Nugent score. It is the overgrowth by these pathogenic bacteria that overwhelms the normal constituents of the vaginal microflora.38 More recently, quantitative data on the vaginal microflora during pregnancy have become available.39,40 These data have been used to develop statistical models to identify several key microbiological risk factors for preterm delivery.3,39,41–43 It is clear that the presence or absence of H2O2-producing and nonproducing Lactobacillus strains, elevated vaginal pH, Lactobacillus concentration and abnormal Nugent score are risk factors that yield an initial predictive model for preterm delivery.39 Among all the risk factors assessed to date, the presence of both H2O2-producing and nonproducing strains and race (a likely surrogate for other microbiological risk factors, such as total aerobic bacterial count) are the strongest predictors for outcome. It should be pointed out that each of these identified variables might be a surrogate marker for the actual risk factor. It has been shown, for example, that there is a difference in the total aerobic bacterial count based on race. Despite the recent success in the identification of specific microbiological risk factors, the actual mechanism(s) by which bacteria contribute to preterm birth is not well understood.

Vaginal pH is clearly associated with the presence or absence of high levels of certain organisms within the vaginal environment. At pH values of 4.5 or less, lactobacilli are the dominant species present, largely based on their ability to grow at low pH. Other organisms, such as G. vaginalis, Prevotella sp. and anaerobic streptococci, tend to be present at lower concentrations.2,44–46 In contrast, it has been shown that the organisms associated with BV are present in large numbers when the pH is >5.0.47–50 The association between BV and pH is well established. However, the association between an elevated vaginal pH, BV and preterm delivery is less clear, with some studies showing little increased risk in the presence of BV, and other studies showing significant risk of preterm delivery in the presence of an elevated vaginal pH and BV.9,10,12,51–55 Some of the discrepancy between studies may be related to how BV is diagnosed.56,57 One study has shown that routine screening of vaginal pH in pregnant women and subsequent treatment of BV decrease the number of preterm births.58 Quantitative studies of vaginal microflora during pregnancy have established vaginal pH as an important predictor for the number of specific microorganisms within the vagina of pregnant women.39 It has also been shown that there are clear synergistic relationships among specific organisms, such as Prevotella sp. and G. vaginalis, in vitro.59,60 Although pH is not an independent risk factor, based on published quantitative studies, it is highly correlated with the presence of large numbers of Prevotella sp. and other Gram-negative anaerobes, anaerobic streptococci and G. vaginalis, all thought to be contributory to the events leading to preterm delivery. pH is an important environmental regulatory mechanism for the vaginal microflora, and changes that may occur in an environment with an elevated pH may be the actual mediators of detrimental effects.

When considering adverse pregnancy outcomes, the limitation of the discussion to the abnormal vaginal microflora only excludes other important microbiological information. In a recent multisite prospective study of extremely low gestational age infants (ELGAN), placental chorion parenchyma samples were obtained from over 1000 pregnancies in which delivery was before the 28th week of gestation.61,62 It was documented that a significant percentage of placental samples were colonised with bacteria, and the rate of organism recovery decreased with increasing gestational age for both vaginal and caesarean delivery routes. The rates were higher for placentas delivered vaginally, but, even for placentas delivered abdominally, more than half the placentas delivered before the end of the 24th week, and one-third of those delivered before the end of the 27th week, harboured one or more microbial species. It was noted that a decline in microorganism recovery rates with increasing gestational age was seen for placentas from pregnancies that ended with spontaneous labour. The increase in microorganism recovery with increasing gestational age was also linear when placental samples were stratified by route of delivery, duration of membrane rupture and number of fetuses. The microbiological findings suggest that the source of many of the organisms isolated from chorion parenchyma is the vaginal microflora, presumably acquired through an ascending route from the vagina. When single- and multi-microorganism cultures were considered together, Prevotella bivia, group D Streptococcus, α-haemolytic Streptococcus, anaerobic Streptococcus, G. vaginalis, U. urealyticum and Mycoplasma sp. other than U. urealyticum were all present in more than 3% of placental specimens. In contrast, among caesarean-delivered placentas from pre-eclamptic pregnancies, only Propionibacterium sp. and coagulase-negative Staphylococcus sp., both considered members of the skin microflora, were present in >3% of samples.

Among placentas delivered by caesarean section, and therefore not contaminated by passage through the birth canal, aerobes and Mycoplasma were recovered much more commonly in those with fetal vasculitis, an indicator of inflammation, than in samples from placentas without this histological marker. The microorganisms recovered in pure culture most frequently from placentas with fetal vasculitis included: E. coli, group B Streptococcus, α-haemolytic Streptococcus, U. urealyticum and Mycoplasma sp. other than U. urealyticum. Microorganisms that were components of polymicrobial cultures that distinguished between placentas with and without fetal vasculitis included Actinomyces sp., P. bivia, Corynebacterium sp., Peptostreptococcus magnus and anaerobic Streptococcus sp. Among caesarean section-delivered placentas from pregnancies with preterm labour, Prevotella sp., a genus usually found as part of the vaginal microflora, but also recovered as an invasive pathogen from a variety of infections, tended to occur in association with E. coli. In contrast, Actinomyces sp. tended to occur with G. vaginalis, U. urealyticum and other Mycoplasma sp.

The association of increased rates of microorganism recovery from placenta parenchyma with gestational age, preterm labour and fetal vasculitis suggests that an appreciable proportion of the microorganisms recovered contribute to preterm labour and the fetal inflammatory response, which might be intermediary between the microorganism and preterm labour. In addition, the finding of presumed nonpathogenic microorganisms within the placenta parenchyma is provocative, and suggests that the placenta may routinely be colonised with organisms that will be encountered during passage through the birth canal, and low levels of these organisms are required to prime the fetal immune system for subsequent challenge by greater numbers of organisms encountered during delivery.

In a subsequent analysis of the data from the ELGAN study, the same group sought to characterise the relationship between histological patterns of inflammation and microorganism recovery from the placentas of live-born infants delivered before the 28th postmenstrual week.63 It was observed that microorganisms were recovered from 51% of placentas and were more frequently isolated when high-grade chorionic plate inflammation was present. Actinomyces, P. bivia, Corynebacterium sp., E. coli, Pe. magnus, multiple species of Streptococci (group B, group D, α-haemolytic and anaerobic), Mycoplasma sp. (other than U. urealyticum) and U. urealyticum were the most frequently isolated species, and were also associated with fetal vasculitis (neutrophilic infiltration of chorionic plate fetal stem vessels or umbilical cord vessels as observed by histological evaluation). Additional evaluation of the clinical data and follow-up on the neonates included in this study documented the relationship between placental bacterial colonisation and various neurological abnormalities.64,65

Although many questions remain, it is clear that the microflora of the vagina is an important mediator of adverse pregnancy outcomes and may serve as the source for organisms that not only colonise the vagina, but also the placental tissues of the developing fetus.

Microbial–host interaction in the lower genital tract

The shift in vaginal microbial flora and its relationship to health and disease are difficult to understand unless they are considered in the context of host–parasite conflict. Infectious diseases result not only from the ill effects of microbial invasion, but also from the nature of the host response. Clinicians examining the vaginal habitat should therefore be interested in the microbial–host interaction.

Vaginal mucosa, cervix and cervical mucus form a barrier against microorganism invasion of the intrauterine compartment. This is the result of the physical epithelial barrier with tight junctions, as well as the presence of continuous controlled inflammation. Such controlled inflammation is, in part, a function of the innate immune system. The cellular components of the innate immune system, including antigen-presenting dendritic cells, macrophages and natural killer cells, are the first to contact invading microorganisms. This arm of the immune response recognises bacteria mainly via pathogen-associated molecular patterns (PAMPs). Such recognition allows the immune cells to respond to a wide array of microorganisms using a restricted number of receptors. A major family of PAMP receptors is the membrane-associated Toll-like receptor (TLR) family, which bind to different bacterial products and transduce an inflammatory signal in the cells. Following activation of the innate immune system, dendritic cells activate T lymphocytes, which are part of the adaptive immune system, leading to the differentiation of naive T cells into effector and regulatory T cells. Following activation, T cells respond by secreting soluble mediators, including pro-inflammatory [e.g. interleukin-1 (IL-1), tumour necrosis factor-α (TNF-α), IL-6] and anti-inflammatory [IL-1 receptor antagonist (IL-1ra), IL-10] cytokines and other immunomodulators (Figure 1).

Figure 1.

 Components of the innate immune system. (1) Mannose-binding lectin binds to carbohydrate moieties on microbial surfaces, resulting in complement activation and opsonisation of the microorganisms by phagocytic cells. (2) Toll-like receptors (TLRs) recognise pathogen-associated molecular pattern (PAMP) structures uniquely present on microbial cell surfaces. (3) TLR activation triggers a cascade of intracellular events, resulting in the transcription of pro-inflammatory cytokine genes. (4) Intracellular expression of the 70-kDa heat shock protein (hsp70) gene, as a consequence of infection or other stressors, inhibits apoptosis and protein denaturation. (5) Extracellular hsp70 binds to specific receptors on phagocytic cells and stimulates the production of cytokines. (6) The activation of inflammatory cells, such as neutrophils, leads to phagocytosis and death of microorganisms.66 IL, interleukin; TNF, tumour necrosis factor.

Although the transition from a Lactobacillus-dominated to BV-like vaginal microflora is associated with variables such as sexual intercourse with specific male partners, vaginal douching, smoking, intrauterine devices and chronic stress, the factors that trigger this shift remain unknown. Recent observations suggest that this may, in part, be a result of alterations in innate immunity. The proposed mechanisms include: (1) diminished TLR activation; (2) increased extracellular heat shock protein 70 (hsp70) production; and (3) inadequate release and/or function of mannose-binding lectin (MBL), all of which may impair controlled inflammation that prevents bacterial overgrowth in the vagina.67,68 This hypothesis is consistent with the observation that most patients with a clinical diagnosis of BV lack vaginal leucocytosis, and hence the condition is called ‘vaginosis’ rather than ‘vaginitis’.

Yet, there is a subset of women with a BV-like microflora who manifest a local pro-inflammatory response. Cervicovaginal samples from such women contain large numbers of leucocytes and elevated levels of pro-inflammatory cytokines (IL-1β, IL-6, IL-8).69–71 The pro-inflammatory cytokine milieu in the cervix is enhanced in pregnant women with BV, compared with that in nonpregnant women. Smokers and those heavily colonised with anaerobic Gram-negative rods and G. vaginalis have higher levels of pro-inflammatory cytokines. This group of women is at greater risk of SPD. We have shown that the risk of SPD is three-fold greater among women with elevated IL-1β (>10 pg/ml) measured in cervicovaginal samples collected in mid-trimester.70 Almost all cases of elevated IL-1β had altered vaginal microflora. Simhan and Krohn72 made similar observations. In their population, women with high pro-inflammatory (IL-1α, IL-1β, IL-6) and low anti-inflammatory (IL-4, IL-10 and IL-13) cytokines in cervicovaginal secretions collected in the first trimester had a 7.7-fold greater risk for SPD before 34 weeks. In another study, Simhan et al.73 used a vaginal pH > 5.0 as a surrogate for altered vaginal flora and more than five leucocytes per oil field as a marker for inflammation. Only women with both elevated vaginal pH and leucocytosis were at an increased risk of PPROM in the early third trimester.

These studies suggest that there is individual variation in microbial–host interaction in the lower genital tract. Those who fail to control bacterial overgrowth (particularly anaerobic Gram-negative bacteria and G. vaginalis), but exhibit an exaggerated pro-inflammatory response, are at risk for SPD and PPROM. It is unclear whether inflammatory markers in cervicovaginal fluid correlate with those in the uterine compartment or simply represent a trait of overall hyperresponsiveness to certain infectious agents. It is, however, indisputable that factors that influence the variation in microbial–host interaction play an important role in individual susceptibility to adverse pregnancy outcomes.

Genetic susceptibility to disease

In any two unrelated individuals, 99.9% of the genome is identical. Differences in DNA sequence and structure account for the remaining variation. A specific sequence variation that is common in the population is referred to as a polymorphism. Typically, a frequency of 5% or more is considered as common. Alternative versions of genes containing different sequences are known as alleles. The resulting total pattern of variation in a gene or a chromosome forms what is known as a haplotype.

Single nucleotide polymorphisms (SNPs, pronounced ‘snips’) are variations in a single nucleotide, and are the main source of human genetic variation. By 2007, the HapMap project had documented more than 3.1 million SNPs.74 It is estimated that SNPs average 1 in every 300 nucleotides among the three billion nucleotide base pairs that constitute the human genome.

Microsatellites are another common form of genetic variation scattered almost at random throughout the genome. They are defined as DNA sequences consisting of relatively short repeats of 1–5 bp units. The distinct biological functions of microsatellites in eukaryotes and prokaryotes are unclear.

Genome-scanning technologies have revealed an unexpectedly large extent of structural variations in the human genome. Genomic alterations that involve segments of DNA larger than 1 kb are operationally defined as structural variants. They include deletions, duplications, insertions, inversions and translocations, and copy-number variants (CNVs), segments of DNA that are present at a variable copy number.75 The Database of Genomic Variants is the most up-to-date repository for results of comprehensive genome-wide screening for CNVs from peer-reviewed studies.76 In March 2010, the database contained 14 478 loci of CNVs. It is estimated that there are some 100 CNVs per individual, each more than 50 kb in size when compared with the reference sequence.

Most polymorphisms have no biological effect. Others are functional and alter the stability and/or expression of the resulting protein. SNPs in coding regions that involve amino acid substitutions may affect the activity, stability and/or substrate specificity of the protein. SNPs in noncoding regions can influence the rate of transcription and translation. Gene duplication results in elevated protein production. Deleted gene sequences or splice variants result in absent, truncated or abnormal protein.

Interindividual genetic variation common in the population may be associated with altered susceptibility to disease. The individual risk associated with such genetic variations is often low. However, because of their high population frequency, they potentially have greater public health relevance (i.e. population-attributable risk). Malaria and the haemoglobinopathies are classical examples of this phenomenon.77,78 The severity of malaria and the related risk of death are reduced among carriers of sickle cell disease. This is probably because of decreased growth of the parasite in infected erythrocytes of such individuals. Similarly, both the thalassaemias and glucose 6-phosphate dehydrogenase deficiency protect the host against malaria. Another example is the influence of the CCL3L1 chemokine receptor gene containing segmental duplications on HIV-1/AIDS susceptibility. Individuals carrying fewer CCL3L1 gene copies than average are at a greater risk of HIV and AIDS.79

Genetic influence on alterations in vaginal flora and adverse pregnancy outcomes

Earlier in this article, we highlighted the importance of the innate immune response against microbial invasion at mucosal surfaces, such as the female genital tract. Multiple genes are responsible for the regulation of the innate immune system. Hence, genetic polymorphisms that disrupt innate immune recognition or the response to infectious microorganisms could explain the genetic variability to vaginal flora alterations and pregnancy complications, such as SPD and PPROM.

Polymorphisms in the IL-1 gene complex

The IL-1 family of molecules consists of two agonists, IL-1α and IL-1β, a specific receptor antagonist called IL-1ra, and two different receptors, IL-1R type I and IL-1R type II. There are numerous SNPs and microsatellites in the IL-1 gene complex, but only a few have so far been studied in relation to lower genital tract microflora. We studied a microsatellite in intron 2 of the IL-1ra gene (gene symbol IL1RN). This polymorphism results in five distinct alleles. Alleles 3, 4 and 5 are present in <5% of the population. The frequency of allele 2 (IL1RN*2) is 4–26%.80 This allele has been associated with various chronic inflammatory conditions. IL1RN polymorphism seems to influence both IL-1 and IL-1ra gene expression (reviewed in Gençet al.67). IL1RN*2-positive monocytes show increased production of IL-1ra and decreased production of IL-1β. The T allele of an SNP at position −31 (IL1B31T) in a transcriptional start site is also associated with a decrease in IL-1β production in response to endotoxin stimulation-induced protein secretion. This may be a consequence of linkage equilibrium between IL1RN*2 and IL1B31T.

In our study, IL-1β levels rose in proportion to vaginal concentrations of Gram-negative anaerobic rods and G. vaginalis in homozygous carriers of IL1RN*1.81 Elevated mid-trimester vaginal IL-1β has consistently been associated with increased susceptibility to SPD. Not surprisingly, we also observed a higher rate of preterm births among homozygous carriers of IL1RN*1, a genotype associated with elevated IL-1β levels in our population.

Although IL1RN*1 homozygosity was associated with elevated IL-1β, an increase in pH in the vagina was related to IL1RN*2 carriage. This association was most striking among IL1RN*2-positive African American women Other features characteristic of BV—diminished colonisation with lactobacilli, a Nugent score of ≥7, elevated levels of anaerobic Gram-negative rods and Mycoplasma—were also increasingly prevalent in African American women who possessed the IL1RN*2 allele. In another study of pregnant women, homozygous IL1RN*2 carriage has also been associated with an increased rate of U. urealyticum isolation.82 Homozygous IL1RN*2 carriers also have the highest vaginal IL-1ra concentration.

Thymine to cytosine (T > C) substitutions in the IL-1β gene (gene symbol ILB) at positions −511 and +3954 relative to the transcriptional start site of the IL-1β gene have also been studied in relation to susceptibility to BV in pregnant women. The IL1B511*T 83,84 and IL1B+3954*T 85,86 alleles were associated with increased production of IL-1β in studies of lipopolysaccharide (LPS) stimulation of cells in vitro. In a study of pregnant women examined before 30 weeks of gestation, women with BV were less likely to be homozygous for IL1B+3954*C [odds ratio (OR), 0.5; 95% confidence interval (CI), 0.3–0.9].87 Cauci et al.88 arrived at similar conclusions in a study of nonpregnant women. In their population, BV was associated with homozygous carriage of the IL1B+3954*T allele (OR, 2.8; CI, 1.4–5.9). In addition, they showed that there was a significant association between BV and homozygous carriage of the IL1511*C allele (OR, 1.5; CI, 1.0–2.1). The influence of these polymorphisms on vaginal IL-1β was not reported in either study. We and others have studied the association between ILB+3954*C and SPD. Although maternal carriage of this polymorphism may influence the vaginal microflora, it does not seem to increase the risk of SPD.89–91 In contrast, homozygous carriage of the IL1B511*T allele, which seems to be protective against BV, was found to increase the risk of SPD by six-fold.92

Polymorphism in the TNF-α gene

The TNF-α gene (gene symbol TNFA) is located within the major histocompatibility complex on chromosome 6, the most gene-dense and polymorphic region of the entire genome. The most extensively investigated polymorphisms in relation to infectious diseases are in the promoter region of TNFA at positions −308 and −238 relative to the transcription start site. Both of these are guanosine to adenosine (G > A) substitutions. The A alleles of both polymorphisms have higher transcriptional activity when compared with the G alleles. These alleles are not more frequent in pregnant women with BV.87 However, they seem to influence the local TNF-α response to the vaginal microflora. We studied TNF-α levels in vaginal secretions of pregnant women in mid-trimester in relation to BV and TNFA308G > A.93 Among women with BV, the vaginal TNF-α levels were significantly higher in carriers of allele A than in carriers of allele G. Such a difference was not observed in women without BV, highlighting the fact that the altered genetic capacity is manifested at the phenotypic level only in the presence of an environmental exposure—BV-like flora in this case. This study was underpowered to investigate the association between SPD, gene polymorphism and the vaginal TNF-α concentration. Macones et al.94 addressed this question. Interestingly, they found that the risk for SPD increased almost six-fold in those women with BV who were also positive for TNFA308*A. The association between the TNFA308*A allele and SPD is predominantly observed in the African American population.

Polymorphism in the IL-6 gene

IL-6 is a pro-inflammatory cytokine that regulates the presence of macrophages and other helper cells. Its production is stimulated by IL-1, TNF-α and LPS. Cytosine to guanine (C > G) substitution in the IL-6 gene at position −174 results in a higher transcriptional activity in response to IL-1 and LPS.95 The −174*C allele was found to be significantly more common in black women with BV.87 This was not observed among white women. Although the −174*C allele was associated with BV, homozygous carriage of this allele was protective against SPD before 34 weeks of gestation.96 In contrast, Gómez et al.97 found that the risk of SPD increased slightly (1.6-fold) in carriers of the −174*C allele. The risk was increased three-fold if these carriers were also diagnosed with BV in the second trimester, suggesting a gene–environment interaction, the nature of which is unclear.

Polymorphism in the IL-8 gene

IL-8 is a small peptide (6–8 kDa) that is produced in response to LPS, TNF and IL-1. It induces chemotaxis and a respiratory burst in neutrophils, and promotes vascular leakage. The promoter region of the IL-8 gene contains polymorphic residues that influence the level of IL-8 expression. In a study of 34 pregnant women with asymptomatic BV, matched for gestational age with 38 pregnant women without BV, neither the frequencies of IL-8 polymorphic alleles nor the levels of IL-8 in vaginal fluid were associated with BV. Goepfert et al.87 studied a T > C substitution at position −845. In this study, the C allele was associated with increased IL-8 production and severe inflammatory conditions, such as systemic lupus erythematosus nephritis,98 and was less prevalent in pregnant women with BV. The influence of the IL-8 polymorphism on pregnancy outcome is not known.

Polymorphism in the TLR-4 gene

The gene coding for TLR-4 (gene symbol TLR4) is polymorphic in humans. Two SNPs result in the replacement of aspartic acid with glycine at amino acid 299 (TLR4 Asp299Gly) and threonine with isoleucine at amino acid 399 (TLR4 Thr399Ile) in the final protein. These alterations in the amino acid sequence of TLR-4 result in a marked reduction in responsiveness to endotoxin.99 We observed that the TLR4 Asp299Gly polymorphism influences the vaginal immune response to G. vaginalis and anaerobic Gram-negative rods.100 The IL-1β response to bacteria was greater in TLR4 299*Asp homozygotes than in carriers of the TLR4 299*Gly allele. The blunted pro-inflammatory cytokine response seen in TLR4 299*Gly carriers may allow a greater proliferation of these microorganisms. This observation becomes even more striking when considering that the concentration of G. vaginalis and anaerobic Gram-negative rods was 10-fold greater in TLR4 299*Gly carriers when compared with TLR4 299*Asp homozygotes. Goepfert et al.87 observed an association between the maternal TLR4 399*Ile allele and BV among pregnant women in North America. Neither of the maternal alleles, TLR4 299*Asp or TLR4 399*Ile, associated with vaginal flora shifts was linked to an increased risk of prematurity, PPROM or SPD.100,101 Although maternal carriage of the studied polymorphisms in TLR4 influences the vaginal microflora, it does not seem to have a major effect on pregnancy.

Polymorphism in the MBL gene

MBL (gene symbol MBL2) plays an important role in innate immunity. This circulating protein binds to mannose, N-acetylglucosamine and other carbohydrate residues on microbial surfaces, resulting in complement activation and opsonisation of the microorganisms by phagocytic cells. The variant allele (G > A) in codon 54 in exon 1 of the MBL2 gene produces an unstable MBL protein, which has been associated with increased susceptibility to infections.102 In a group of women sampled in mid-trimester, allele A carriage was associated with heavier colonisation with Peptostreptococcus species, but not with BV-associated microorganisms or Candida.66 This allele was more frequent in predominantly nonwhite women with a history of BV,103 but not in white women with the same condition.104 Maternal carriage of this allele was not associated with recurrent miscarriages in white women.105

Concluding remarks

Genetic association studies suggest that interindividual differences in the composition of vaginal microflora and susceptibility to adverse pregnancy outcomes appear to be a result, at least in part, of polymorphisms in genes encoding modulators of innate mucosal immunity. The available data do not allow us to make any recommendations for clinical management at the present time, because the effect of most individual SNPs on phenotype is at best moderate and has not been verified by multiple studies. Such inconsistency is primarily a result of methodological problems inherent to genetic association studies, and has been discussed in detail elsewhere.106,107 Future studies should follow the established guidelines for gene association studies and make use of inexpensive high-throughput genotyping techniques that will facilitate the analyses of complex interactions between multiple genetic and environmental variables. The ultimate goal is to modify the environmental factors, for example vaginal microflora, that may cause disease among individuals with a particular genetic susceptibility.

Disclosure of interest

The authors have no relevant financial, personal, political, intellectual or religious interests to disclose in relation to this article.

Contribution to authorship

AO wrote the section entitled, ‘Normal and abnormal vaginal flora in pregnancy and adverse pregnancy outcomes’, and MRG wrote the remaining sections. Both authors reviewed and approved the final version of the manuscript.

Details of ethics approval