Possible association between amniotic fluid micro-organism infection and microflora in the mouth


*Dr R. P. Allaker, The Department of Oral Microbiology, Barts and The London, Queen Mary's School of Medicine and Dentistry, Turner Street, London E1 2AD, UK.


Objective To determine whether oral bacteria are found in the amniotic cavity.

Design Laboratory based analysis of clinical samples.

Setting Royal London Hospital, Whitechapel.

Population Forty-eight women attending for elective caesarean section.

Methods Dental plaque, a high vaginal swab, amniotic fluid and chorioamnion tissue were taken from women with intact membranes.

Main outcome measures Samples were investigated using culture and microscopy for the presence of micro-organisms. Amniotic fluid was analysed by polymerase chain reaction (PCR) for the presence of the ubiquitous 16S rRNA gene specific to most eubacteria. Samples were analysed using PCR genus and species specific primers directed to bacterial taxa found as part of the normal oral microflora (Streptococcus spp. and Fusobacterium nucleatum). Levels of prostaglandin E2 and cytokines were measured in amniotic fluid.

Results Amniotic fluid was positive for universal bacteria PCR, Streptococcus spp. PCR and F. nucleatum PCR in 34/48, 20/48 and 7/48 of cases, respectively. Streptococcus spp. and F. nucleatum were cultured from the dental plaque, vagina and amniotic fluid of 48/48, 14/48, 0/48 and 29/48, 6/48, 0/48 subjects, respectively. A significant association was found between detection of microbial DNA (universal and F. nucletum) and complications in previous pregnancies including miscarriage, intrauterine death, neonatal death, preterm delivery and premature rupture of membranes (P < 0.05 and P < 0.01, respectively). Prostaglandin E2 and cytokine levels, with the exception of IL-1α, were not significantly different between women with and without evidence of infection.

Conclusions The results indicate that Streptococcus spp. and F. nucleatum in the amniotic fluid may have an oral origin.


Six percent of all live births in the United Kingdom are low birthweight, usually as a result of preterm labour or premature rupture of membranes. It is estimated that 60% of mortality among infants without anatomical or congenital defect is attributable to preterm low birthweight1. The prevention of low birthweight worldwide could be improved by a better understanding of the contributory risk factors. Although, many risk factors have been identified such as high (>34 years) or low (<17 years) maternal age, race, low socio-economic status, inadequate antenatal care, drug, tobacco and alcohol abuse, it is estimated that 25% of low birthweight infants occur without a suspected risk factor2.

In recent years attention has been paid to the association between low birthweight and infection within the amniotic cavity. This environment is considered to be sterile under normal conditions and thus identification, by whatever means, of any micro-organism at this site is of potential concern. In women with preterm labour and intact membranes bacteria have been detected using culture techniques in the amniotic fluid from 20% of these subjects3. The portals of entry for such micro-organisms into the amniotic fluid are thought to include direct spread from the genital tract, haematogenous transplacental infection and iatrogenic infection during operative procedures such as amniocentesis.

It has been postulated recently that opportunistic pathogens, from the oral cavity and other remote body sites, may play a role in premature labour via a haematogenous route. A number of bacteria, from a possible non-genitourinary tract source, have on occasions been cultured in amniotic infections. These include Fusobacterium nucleatum, Peptostreptococcus spp, Porphyromonas and Prevotella spp., Eubacterium spp. and Eikenella corrodens. Indeed, Fusobacterium nucleatum, a common oral species, is the most frequently isolated species from amniotic fluid cultures among women with preterm labour and intact membranes4. Infections of the oral cavity are now thought to be important with respect to preterm low birthweight (Plow birthweight). This is supported by a study in which periodontal disease was assessed to be a significant risk (odds ratio 7.0) for Plow birthweight5. Bacteria found in the mouth may therefore reach the amniotic fluid via the blood stream, particularly in the presence of gingivitis or periodontitis during pregnancy6.

The use of traditional microbiological methods, for example gram staining and culture, has been of limited clinical benefit in the diagnosis of intra-amniotic infection. This is due to the low yield of bacteria cultured, perhaps owing to the bacteriostatic properties of amniotic fluid and the difficulty of certain bacterial species to grow in culture (e.g. Chlamydia, Mycoplasma spp. and ‘uncultivatable’ bacteria). The value of indirect markers of infection, like amniotic fluid glucose or inflammatory cytokines, in clinical diagnosis is also still unclear. Application of up-to-date molecular microbiology techniques is now needed to obtain a precise aetiological diagnosis of intra-amniotic infection.

The aim of this study was to determine, using culture and molecular techniques, the measure of oral bacteria in the amniotic cavity.


Women with intact membranes booked for scheduled caesarean section at the Royal London Hospital, Whitechapel were approached at the booking-in clinic the day before. Informed written consent was sought for their participation in the study. Ethical clearance for the study was obtained from the East London and the City Health Authority (ELCHA).

General demographic details and known confounding factor information was collected from the women through a previously developed (ELCHA P/96/4 ‘Maternal periodontal disease and the risk of preterm low birthweight infants’) administered questionnaire and from Royal Hospitals Trust notes. Brief questions were asked relating to smoking, alcohol consumption, treatment/medication, dental treatment, diet, pan use, education, antenatal care and age of infant's father.

On the day before caesarean section dental plaque from the buccal gingival margin in each quadrant was taken using four sterile toothpicks. Immediately before caesarean section a high vaginal swab (Amies transport swab with charcoal) was taken. 10–20 mL of amniotic fluid was aspirated intra-operatively by trans-membrane amniocentesis after the uterine incision was made. After delivery, four tissue blocks of chorioamnion (two each from the maternal and fetal sides; 5 mm3 each) were taken with sterile forceps and scalpel blades. Dental plaque and tissue samples were placed into a reduced transport medium7 and transported immediately to the laboratory where all samples were processed within two to four hours.

Plaque samples were disaggregated by vortex mixing for 60 seconds. Chorioamnion samples were homogenised using a tissue grinder and sterile sand. Amniotic fluid was centrifuged (4000 rpm/15 minutes) and the resulting pellet resuspended in 2ml of phosphate buffered saline for culture and polymerase chain reaction (PCR) analysis.

Dental plaque, high vaginal swabs, amniotic fluid and chorioamnion tissue were investigated for the presence of aerobic and anaerobic bacteria, fungi, mycoplasmas and Trichomonas vaginalis using both microscopy and the relevant selective culture conditions (Table 1). Identification of micro-organisms to the species/subspecies level was carried out using established identification methods.

Table 1.  Culture media, incubation conditions and microorganisms cultured.
Culture mediaIncubation conditions (37°C unless stated)Microorganisms cultured
Blood agar base (Oxoid CM271) with horse blood (5% v/v)CO2/48 hTotal aerobes
Blood agar base (Oxoid CM271) with chocolated horse blood (5% v/v)CO2/48 hHaemophilus spp., Neisseria spp. and other
Columbia blood agar base No. 2 (Difco) with Gardnerella vaginalis selective supplement (Oxoid SR119E) and horse blood (10% v/v)CO2/48 hGarnerella vaginalis
Columbia blood agar base No. 2 (Difco) with Campylobacter selective supplement (Oxoid SR085E), growth supplement (Oxoid SR084E) and lysed horse blood (5% v/v)Microaerophilic/72 hCampylobacter spp.
Listeria enrichment broth base (Oxoid CM897) with Listeria selective supplementCO2/48 h (enrichment)Listeria spp.
(Oxoid SR140E). Listeria selective agar baseCO2/48 h (sub-culture)
(Oxoid CM856) with Listeria selective supplement (Oxoid SR140E)
Sabouraud dextrose agar (Oxoid CM41)Aerobic/48 h & 25°C/7 daysYeasts
MacConkey agar (Oxoid CM7)Aerobic/48 hEnteric bacteria
Tryptone soy agar (LabM) with horse blood (5% v/v) and clindamycin (5 μg/ml)CO2/72 hEikenella corrodens
‘New York City’ medium: G.C. agar base (Oxoid CM367) with VCAT selective supplement (Oxoid SR10B) and lysed horse blood (10% v/v)CO2/anaerobic/48 h/10 daysNeisseria spp. & Mycoplasmas
Fastidious anaerobe agar (LabM) with horse blood (5% v/v)Anaerobic/7 daysAnaerobic bacteria incl. Porphyromonas & Prevotella spp.
Fastidious anaerobe agar (LabM) with horse blood (5% v/v), 7.5 μg/ml vancomycin and 100 μg/ml neomycinAnaerobic/7 daysFusobacterium spp.
Rogosa agar (Oxoid CM627)Anaerobic/72 hLactobacillus spp.

Amniotic fluid was analysed by PCR for the presence of the ubiquitous 16S ribosomal RNA gene specific to most eubacteria. Amniotic fluid samples were also analysed using PCR genus and species specific primers directed to Streptococcus spp. and F. nucleatum.

DNA was extracted by the CTAB method of Wilson. The sequences of the universal primers used to amplify most eubacteria8 were 5′-AGAGTTTGATC(A/C)TGGCTCAG-3′ (BACTF; forward primer) and 5′-AAGGAGGTG(A/T)TCCA(A/G)CC-3′ (BACTR; universal (n= 1) reverse primer) or 5′-TACGG(C/T)TACCTTGTTACGACTT-3′ (UNIRV; universal (n= 2) reverse primer). The two Streptococcus spp. primers used were 5′-AGTCGGTGAGGTAACCGTAAG-3′ (STR1F) and 5′-AGGAGG TGATCCAACCGCA-3′ (DG74)9 in a nested PCR based assay. In the nested assay, the BACTF and BACTR universal primers were used initially to flank a 1500 bp region in the 16S rRNA gene. Amplicons demonstrated with the universal primers were then further assessed with the genus specific primers (STR1F and DG74) in order to detect a 105 bp product. Streptococcus anginosus (NCTC 10,713) and Streptococcus oralis (laboratory isolate) were used to check primer specificity and PCR conditions. Fusobacterium nucleatum primers to amplify a region of the 16S rRNA gene were BACTF and 5′-GTCATCGTGCACACAGAATTGCTG-3′ (FN1)10. F. nucleatum subsp. nucleatum (ATCC 25,586 and labouratory isolate W1046), F. nucleatum subsp. polymorphum (NCTC 10,562), F. nucleatum subsp. vincentii (ATCC 49,256), F. nucleatum subsp. fusiforme (NCTC 11,326), F. periodonticum (ATCC 33,693 and laboratory isolate W1693) and F. necrophorum (laboratory isolate W3931) were used to check specificity of primers and PCR conditions.

5 μL of the DNA preparation, 10 μL of 10 × PCR amplification buffer (100 mM HCl, 500 mM KCl, 15 mM MgCl, Triton X-100 1% v/v), 1 μL of dNTPs (Appligene-Oncor), 0.5 μL of forward primer, 0.5 μL of reverse primer and 2.5 units of Taq DNA polymerase (Appligene-Oncor) were added together with sterile distilled water to make the final volume of 100 μL. For the universal and Streptococcus spp. assays the reaction tubes were placed on a thermal cycler (Hybaid) to equilibrate at 95°C for five minutes. The amplification stage then consisted of 30 cycles of 95°C for one minute, 55°C for one minute and 72°C for two minutes. The reaction was terminated at 72°C for five minutes. The PCR amplification products were analysed by agarose gel electrophoresis using 1.0% and 2.0% gels, respectively. For the F. nucleatum assays the reaction tubes were placed on the cycler to equilibrate at 94°C for five minutes. The amplification stage then consisted of 30 cycles of 94°C for one minute, 60°C for one minute and 72°C for 2.5 minutes. The reaction was terminated at 72°C for five minutes. The PCR amplification products (360 bp) were analysed using 2% w/v agarose gel electrophoresis. Appropriate controls and measures to minimise carry-over contamination were incorporated.

Enzyme-linked immunosorbent assay kits (R & D systems) were used according to the manufacturer's instructions to determine levels of IL-1α, IL-1β, IL-6, IL-10, TNF-α and PGE2 in amniotic fluid samples taken at caesarean section.


Student's t test was used to compare gestation times, birthweight and cytokine values between women with and without molecular evidence of infection. Detection frequencies were compared using χ2 analysis. Regression analysis was used to determine possible relationships between gestation time and birthweight with cytokine levels.


Forty-eight women were recruited to the study from the London Borough of Tower Hamlet, a deprived area with a population of low socio-economic status. The ethnicity mix comprised 23 Caucasian, 16 Asian and nine Afro/Caribbean women with a mean age of 31 years. The reasons for caesarean section were previous caesarean section (n= 26), breech (n= 8), choice (n= 3), small pelvis (n= 2), previous neonatal death (n= 2), hypertension (n= 1), breech/cord around neck (n= 1), Perthes disease (n= 1), slipped disc (n= 1), perforated uterus (n= 1), kidney transplant patient (n= 1) and hip fracture (n= 1). The mean gestation period was 39 weeks and the mean weight at delivery was 3247 g (86). All were singleton births with 28 males and 20 females. Nine women were prmigravidae. No evidence of infective morbidity was found among the newborns. No cases of possible clinical chorioamnionitis were identified.

Bacterial DNA was detected by 16S rDNA directed PCR, with either one or both sets of universal primers used, in 34/48 of the amniotic fluid samples (Table 2). Streptococcus spp. and Fusobacterium nucleatum were detected by 16S rDNA directed PCR in the amniotic fluid of 20/48 and 7/48 of patients, respectively. Four of 48 of patients demonstrated multiple infection with both Streptococcus spp. and F. nucleatum in the amniotic fluid.

Table 2.  Association between detection of Fusobacterium nucleatum (Fuso) and Streptococcus spp. (strep) in the amniotic fluid and clinical history of women attending for ceasarean section. AF = amniotic fluid; SA = spontaneous abortion; NND = neonatal death; IUD = Intrauterine death; Uni 1/2 = universal PCR primers; PREM = premature; PROM = preterm/prelabour rupture of membranes; IUGR = Intrauterine growth restriction. Shaded area = correlation between detection of amniotic fluid infection by PCR and previous pregnancy complications.
Subject No.CulturePCRPrevious Complications
Plaque*VaginaUniv 1Univ 2StrepFuso
  1. *All plaque samples positive for culture of Streptococcus spp.

10++++SA, 2 PREM
6++++3 SA, NND
18++++4 SA, IUD
33++++3 SA
48+++2 PREM
13+++++6 SA, PROM
39++NND, 3SA
51+++++2 SA
24+++IUD, NND
36+++IUD, SA
40++9 SA
9, 43, 7, 11, 15, 42+++++0
44, 45+++0
5, 30+++0
17, 49+++0
50++1 SA
28, 34, 37, 38+0
25, 27, 290

All women demonstrated Streptococcus spp. in their dental plaque using standard culture techniques, while 29/48 demonstrated F. nucleatum in their dental plaque by culture. Streptococcus spp. and F. nucleatum were detected in the vagina, by culture, of only 14/48 and 6/48 women, respectively. However, the detection of Streptococcus spp. and F. nucleatum by PCR in the amniotic fluid did not correlate significantly with detection by culture in either the vagina or dental plaque. Candida spp., Aerococcus viridans, Staphylococcus aureus and Staphylococcus capitis were the only micro-organisms cultured from amniotic fluid (subjects 15, 21, 33 and 36). However, 21/48 of amniotic fluid samples were positive for bacteria by microscopy (Table 3). All samples were negative for protozoa and mycoplasmas. No micro-organisms were recovered from chorioamnion samples apart from occasional skin and environmental contaminants including Corynebacterium spp., Staphylococcus spp., Micrococcus spp., Propionibacterium acnes and coliforms. These bacteria were possibly acquired within the operating theatre.

Table 3.  Comparison of obstetric and other parameters (standard error) between women with and without microbial DNA, as detected by 16S rDNA PCR, in the amniotic fluid.
Parameter recordedMicrobial DNA detected (n= 34)Microbial DNA not detected (n= 14)P
  1. *Levels of cytokines in amniotic fluid (AF) at time of caesarean section.

Previous problem183<0.05
Gestation (days)268.00 (1.38)278.71 (3.82)<0.01
Birthweight (g)3180.59 (91.10)3408.71 (192.30) 
Gram stain +ve AF174 
Culture +ve AF40 
Current infection179 
PGE2 (pg/ml)*302.28 (54.67)452.54 (128.80) 
IL-1α (pg/ml)82.36 (6.86)118.31 (19.69)<0.05
TNF-α (pg/ml)3.43 (0.24)3.60 (0.42) 
IL-6 (pg/ml)235.99 (14.20)270.51 (13.96) 
IL-10 (pg/ml)00.41 
IL-1β (pg/ml)00.016 

A significant association between detection of microbial DNA (universal and F. nucleatum) in the amniotic fluid and previous pregnancy complications including miscarriage, intrauterine death, neonatal death, preterm delivery and premature rupture of membranes (P < 0.05 and P < 0.01, respectively) was found (Table 2). In all cases when F. nucleatum was detected, one or more miscarriages had occurred in previous pregnancies. With reference to possible confounding factors, the only factor approaching significance (P= 0.056) was smoking in relation to previous pregnancy problems.

A comparison of obstetric parameters revealed a significantly increased period of gestation (P < 0.01) and an increased mean birthweight in the absence of infection as detected by PCR (Table 3). In these women, only 3/14 had a history of problems in previous pregnancies, while 18/34 of those with PCR evidence of infection had such a history. Infections during the index pregnancy included urinary tract infection, bacterial vaginosis/vaginal discharge, thrush, pericoronitis (inflammation of soft tissue surrounding partially erupted tooth) and respiratory tract infection, and these were noted in 26/48 women (Table 3). However, no association between any such infection during pregnancy and PCR detection of bacteria in amniotic fluid was found. Mean levels of the cytokines IL-1β, IL-6, IL-10 and TNF-α, and prostaglandin E2 measured in amniotic fluid samples, taken at the time of caesarean section, were not significantly different between those women with and without current clinical or PCR evidence of infection. Levels of IL-1α were found to be significantly higher (P < 0.05) in those women with no evidence of infection. No significant association between cytokine levels with either gestation period or birthweight was found.


The actual onset of labour arises due to a sudden change or instability in the maternal system. The mechanisms associated with labour are not fully understood, although prostaglandins especially PGE2 appear to play an important role11. The initiation of preterm labour has been attributed to various clinical findings: normal physiological processes happening early, infection and inflammation, haemorrhage, placental ischaemia and stress12. The view that products of gram negative infection, such as endotoxin and the subsequent release of inflammatory mediators, are sufficient to initiate pregnancy complications is gaining momentum and has lead to agreement that Plow birthweight may occur as a result of gram negative infection mediated indirectly, principally by the translocation of bacterial products. Prostaglandins (PGE2) normally rise throughout pregnancy until a critical threshold is reached to induce labour, cervical dilation and delivery. These levels may also be affected by the normal host response to an infectious agent, which may in turn lead to premature labour and preterm rupture of the membranes13. However, it remains unclear whether infection results in elevated levels of these inflammatory mediators, which then cross the fetal membranes and influence parturition, or if a more direct mechanism involving entry of the infectious agent into the amniotic cavity followed by local elevation of mediators occurs.

The infected periodontium is considered to act as a reservoir for both microbial products and inflammatory mediators such as PGE2 and TNFα, which have been shown to increase in periodontal disease14. Animal experiments on pregnant hamsters with Porphyromonas gingivalis, a gram negative oral micro-organism, show a significant association between increasing levels of PGE2 and TNFα and fetal growth retardation. It is thus possible that the involvement of P. gingivalis and other presumptive periodontal pathogens, including F. nucleatum, in human periodontal infection may affect pregnancy outcome15.

Cytokine levels were measured at the time of caesarean section in this study and mean levels increased with time of gestation. This has previously been demonstrated in other studies16,17. Analysis of amniotic fluid may well reveal a connection between evidence of infection, as determined by either molecular methods or culture, and levels of inflammatory mediators during pregnancy. Indeed, in women with preterm labour elevated levels of IL-6, IL-1α, IL-1β, TNF-α, IL-10 and PGE2 have been found when bacteria were cultured from amniotic fluid16,17. It remains to be determined whether elevated levels are found in cases when infection can only be determined using sensitive molecular methods.

The results in this study indicate that Streptococcus spp. and F. nucleatum found in the amniotic cavity may have an oral origin. In 86% of cases in which F. nucleatum was detected by PCR in the amniotic fluid this organism was also detected by culture in dental plaque. In the case of F. nuleatum, absence from the vaginal samples by culture was also noted in those women who were PCR positive in their amniotic fluid. Although not significant, this association helps support a possible link between the oral cavity and the amniotic fluid. Molecular analysis of vaginal samples could also be undertaken to help confirm absence at this site. F. nucleatum previously has been isolated from amniotic fluid in women with preterm labour and intact membranes18, but F. nucleatum is not considered to be part of the normal vaginal microflora19. It is commonly found in the mouth and could spread, via the blood stream, to amniotic fluid in women who have transient bacteraemia. Indeed, one of the subjects with pericoronitis, with which F. nucleatum is associated, was positive for this organism in the amniotic fluid. Molecular typing techniques to establish similarity between actual isolates of bacteria, or possibly ribosomal gene sequence typing, from various sites will help to confirm a possible link between infection of the amniotic fluid and the oral flora. In an investigation of F. nucleatum to determine the heterogeneity within this species at different body sites, amniotic fluid and healthy gingiva isolates from women with preterm labour were found to be genetically different20. They therefore suggested that extraoral isolates, perhaps from the gastrointestinal and vaginal tract, form a distinct genetic group when they are compared with isolates of oral origin. Simultaneous testing for genetic similarity of F. nucleatum isolates in individual patients from infection sites, the oral cavity, the gastrointestinal tract and the genital tract remains to be carried out.

In this study 71% of the women at caesarean section were found to be positive for evidence of micro-organisms in the amniotic fluid using 16S rDNA PCR. In a study by Jalava et al.21 control amniotic fluid samples were taken by amniocentesis during the second and third trimesters from 16 Finnish women showing no signs of clinical infection and with intact membranes. These were found to be negative by 16S rDNA PCR, using similar primers to this study, and culture to detect bacteria. While, in another group of 20 women21 with prelabour rupture of the fetal membranes 25% of amniotic fluid samples, taken at between 22 and 40 weeks of gestation, were positive using PCR. In a similar study, Hitti et al.22 detected bacteria by PCR in 36% of culture negative women in premature labour whose membranes were intact. It is suggested that current infections or previous pregnancy complications with a possible infectious aetiology as demonstrated in this study may account for this discrepancy. It also has been hypothesised recently that the uterine cavity has a ‘normal flora’ which may possibly be associated with some common gynaecological and obstetric enigmas. Therefore bacteria, many of which may be unculturable or in low numbers, may be detected by PCR and not by culture23. A correlation between detection of microbial DNA (universal and F. nucleatum) and complications during previous pregnancies including miscarriage, intrauterine death, neonatal death, preterm delivery and premature rupture of membranes was found. It is well recognised that DNA can persist in tissues for some time and may well have remained from a previous pregnancy. Amniotic fluid infection may not be detected by culture because of unculturable micro-organisms, a low infectious inoculum, or previous antibiotic treatment. The possible role of infection in such circumstances may thus have been previously under-estimated due to inappropriate microbial detection methods. It is suggested that these individuals may be more predisposed to infection and subsequent complications during pregnancy.


This work was funded by a grant from the Royal London Hospital Special Trustees No. RAC 366 Sample collection involved collaboration with the Royal London Hospital Midwifery and Obstetric theatre team. The authors would like to thank Dr C. Barnick and Dr G. Ayida (consultant obstetricians) for performing the caesarean sections, and Professor W. Wade for providing Fusobacterium spp. Statistical advice was obtained from Ms B. Boucher.