Vaginal Microbiome Is Related To Adverse Pregnancy Outcomes For Elderly Parturient Women

BACKGROUD:The vaginal microbial ecosystem undergoes significant structural changes at various stages in a woman’s life that are directly linked to the level of estrogen and progestin in the body. Over the years, a number of different species of lactic acid bacteria, primarily species of Lactobacillus spp., have been shown to reside in the human vagina and have an important role in this process. These species effectively constitute an ecological guild-a group of species that have similar requirements and play a similar role within a community. Community state type (CST) is used in microbial ecology to describe a group of community states with similar microbial phylotype composition and abundance. It was reported that five CSTs (CST Ⅰ, CST Ⅱ, Ⅲ, CST CST Ⅳ and CST Ⅴ) was presented in the healthy reproductive age women. The CST Ⅰ, CST Ⅱ, Ⅲ and CST Ⅴ belonged to the genus Lactobacillus spp. While CST IV is characterized by the higher diversity of species in the vagina. The CST Ⅳ can be divided into two subcategories:CST Ⅳ-A and CST Ⅳ-B.Pregnancy is a special physical period in reproductive age women. The estrogen increased about 1000 folds during the pregnancy, which accompanied by the increasing cervical glands, higher permeability in the vaginal mucous membrane and edema of the vaginal mucosa. The estrogen have promoted the growth of epidermal cells of the vagina and higher concentration of glycogen and lactic acid, which maintain low pH in the vagina. The vagina microbiome, epidermal cells and vaginal discharge have composed a natural barrier to restrict the growth of pathogens and other opportunistis organisms. In addition, the other factors such as the mother’s immune status, the weight, the fetus’s growth, nutrition, medicine et al. have disturbed the vaginal microbiome. We want to know whether the vaginal microbiome will fluctuate in the different trimesters. What’s more, we want to explore the characteristic of the vagina microbiome in the postpartum. As we all known, the concentration of estrogen decreased sharply at the postpartum and the vaginal microbiome had gone through the delivery. The process of the delivery that would thoroughly washed the vagina with the amniotic fluid and blood, which forced the vaginal microbiome to reconstruct the microecological environment. All the mother’s status including the weight, uterus, immune system and cardiovascular system et al. will recover to the pre-pregnancy at about postpartum 6 weeks. Therefore, we wonder that whether the vaginal microbiome will recover to its non-pregnancy status.Bacterial Vaginosis (BV) is the most frequently cited cause of vaginal discharge and the most common vaginal symptoms prompting reproductive age women women to seek medical care. BV was characterized by low or non abundance in the lactic acid bacteria (Lactobacillus spp.) and higher microbial diversity. The facultative anaerobic bacteria or anaerobic bacteria in BV are cluding Gardnerella, Anaerococcus, Corynebacterium, Finegoldia, Streptococcus, Atopobium, Prevotella, Parvimonas, Sneathia, Mobiluncus, Peptoniphilus, BVAB1, BVAB2 and BVAB3. During pregnancy, BV is associated with several adverse pregnancy outcomes, such as premature rupture of membranes (PROM), spontaneous abortion, preterm birth (PTB), postpartum endometritis et al. Owing to the infection in the low genital tract had correlated closely with the adverse pregnancy outcomes, the researchers paid more attention to the vaginal microbiome in the pregnancy.PROM and PTB were the most common cause of neonatal mortality worldwide, which had closely associated with vaginal infection. The PROM was defined that the membrane was rupture before labor. The infection of the syndrome were composed by PROM, the amniotic cavity infection and premature birth, which is associated with serious complications in the pregnancy. At present, many researchers tried to correlate with the inflammatory cytokines and PROM. However, the misdiagnosis rate and the missing diagnosis rate for cytokines to predict the PROM are high. On the other side, traditional methods with puncturing can accurately diagnose the PROM, which is traumatic to the pregnant women. To date, and there is no gold standard in the noninvasive diagnosis of PROM. This is a challenge for the researchers to find one method that can be rapid, convenient, noninvasive and accurate to diagnose the PROM. Many factors have proved to correlate with the adverse pregnancy outcomes such as ethnic, immune status, nutrition, medicine et al. However, no consistent results have reached to associate with the factors and the adverse pregnancy outcomes. The association between the vaginal microbiome and adverse pregnancy outcomes need further investigations.Objectives:1. Whether the vaginal microbiome at different sampling sites (cervical, posterior fornix and vaginal canal) are homogeneous in the pregnant women.2. What’s the characteristics and dynamic fluctuation of vaginal microbiome at different trimesters in Chinese pregnant women.3. Whether the vaginal microbiome at postpartum 6 weeks can recover to its pre-pregnancy status.4. Whether the vaginal microbiome combined with the metadata can predict adverse pregnancy outcomes.Methods:Part 1Volunteers were recruited between December 2012 and Janunary 2013 during a routine obstetrical visit at Southern Medical University in China, Guangzhou. All participants provided written informed consent. Thirty four subjects were divided into 5 groups, including non-pregnancy (5 subjects), T1 (gestational age< 14 weeks,6 subjects), T2 (14～28.6 weeks gestation,6 subjects), T3 (29～42 weeks gestation,12 subjects) and postpartum (7 weeks,5 subjects). We had numbered the subjects S1 to S34. The sampling process with a sterile speculum examination was performed by a single professional obstetrician. For each individual,9 sterile plastic swabs with triplicates were obtained from the cervix (C), posterior fornix (P) and vaginal canal (V). Three swabs were obtained from each site using the swab method. A total of 306 vaginal swabs were collected from 34 subjects.Swabs were frozen within 4 hours after collection and stored at -80℃ until usage.Using bacterial 16S rRNA universal primers, we amplified the bacterial 16S rRNA V46 conservative district. The PCR products were sequenced by platform of Illumina Hiseq 2000 with PE100-bp and then got the high-throughput primitive data. We used QIIME data analysis process to combine with clinical metadata. QIIME biological analysis process mainly included the operation classification unit (OTU) and principal component analysis (PCA), hypothesis testing, the evolutionary tree analysis and characteristic of the microbiome analysis, species richness, diversity of community structure analysis,. Differentially abundant features were determined using Linear discriminate analysis effect size (LEfSe). LEfSe is an algorithm for high-dimensional biomarker discovery and explanation that identifies genomic features characterizing the differences between two or more biological conditions. LEfSe determines the features most likely to explain differences between classes by coupling standard tests for statistical significance with additional tests encoding biological consistency and effect size. The threshold on the logarithmic LDA score for discriminative features was 4.0.Part 2During the period January to December 2013,340 pregnant women presenting for their first prenatal visit at Southern Medical University in Guangzhou, China were enrolled in our prospective longitudinal cohort study to assess how well biomarkers of the vaginal microbiota during the first trimester could predict adverse pregnancy outcomes. Pregnant women who agreed to participate in the study and signed written informed consent underwent a vaginal ultrasound examination at the first prenatal consultation to confirm the gestational age of the pregnancy, and the recorded gestational age was corrected if the menstrual and ultrasound dates differed by more than one week. Physical examinations and specimen collection were performed by an obstetrician with a sterile speculum. Vaginal pH was measured using pH strips with a pH range of 3.8-5.5 on glass slides. Clinical information including the pregnant women’s antenatal information, the odor, color and character of the vaginal discharge, previous genital tract disease and medical history, the husband’s healthy status were collected. PROM, PTB, acute histologic chorioamnionitis, delivery mode and asphyxia were abstracted from the labour and delivery or prenatal care medical records. A normal pregnant woman was defined as a woman with no obstetrical, medical or surgical complications and delivery at term (i.e.,37 to 42 weeks) without complications. On the other hand, adverse pregnancy outcomes involved the premature rupture of membranes (PROM), preterm birth (PTB), acute histologic chorioamnionitis, asphyxia andcaesarean delivery. PROM was defined with a sterile speculum examination and documentation of amniotic fluid leaving or pooling behind the cervix;in this condition, the fluid had a pH>6.5 and a vaginal posterior fornix smear exhibited a fern-like appearance under the microscope when dried. Preterm birth (PTB) was diagnosed as a birth prior to 37 complete weeks of gestation. Acute histologic chorioamnionitis was diagnosed based on the presence of inflammatory cells in the chorionic plate and/or amniotic membranes. Asphyxia was defined by an Apgar score of less than 7 within one minute.For this study, we selected only women who had a singleton pregnancy and who attended the hospital for an antenatal visit. Each individual had a speculum examination and samples of vaginal fluid were collected under direct visualisation from the posterior vaginal fornix using a plastic swab. Samples were collected at S1 (18.0 ± 2.7 weeks), S2 (29.5 ± 3.5 weeks) and S3 (37.6 ± 0.7 weeks), and PP (postpartum 6.4 ±0.9 weeks). There were 340,298,205 and 73 subjects follow-up in S1, S2, S3 and postpartum respectively, which in a total of 916 samples. Vaginal swabs were placed in tubes without any buffer and immediately stored at -80℃ until assayed.Using bacterial 16S rRNA universal primers, we amplified the bacterial 16S rRNA V4 conservative district. The PCR products were sequenced by platform of Illumina Hiseq 2000 with PE100-bp and then got the high-throughput primitive data. We used QIIME data analysis process to combine with clinical metadata. QIIME biological analysis process mainly included the operation classification unit (OTU) and principal component analysis (PCA), hypothesis testing, the evolutionary tree analysis and characteristic of the microbiome analysis, species richness, diversity of community structure analysis, the different taxa identified between groups (LEfSe online statistical analysis tools). The threshold on the logarithmic LDA score for discriminative features was 2.0. In addition, we used the machine learning random forest model analysis during pregnancy to predict the adverse pregnancy outcomes.Results:Part 11. We obtained a total number of 720,601 high quality 16S rRNA gene sequences from 306 samples, with an average of 2,354 sequences per sample. Within them,35 samples were filtered because of having less than 1000 reads and 271 samples included in the subsequent analysis.2. The Community State Type (CST) of the vaginal microbiome were mainly divided into four types:CST Ⅰ (L.crispatus), CST Ⅱ (L.gasseri), CST Ⅲ (Liners), CST IV (diverse species). The CST IV was characterized by a relatively low abundance of Lactobacillus along with proportions of various anaerobic bacteria species.3. The Binary jaccard, Bray Curtis, Abund jaccard, Unweighted Unifrac distance, Weighted Unifrac distance and Euclidean distacne to show that the microbiome of the three subsites of cervical (C), posterior fornix (P) and vaginal canal (V) are closely clustered, which mean that the microbiome composition of the three subsites are similar.4. The alpha-diversity assessed by the Shannon diversity index and PD (phylogenetic distance) whole tree value have shown that no significant difference existed among the subsites of cervical (C), posterior fornix (P) and vaginal canal (V).5. Although 17.6% (6/34) of subjects had minor taxa differences among the three subsites of cervical (C), posterior fornix (P) and vaginal canal (V), the core community structure can be confirmed within each individual.Part 21. A total of 15,606,477 high-quality 16S rRNA gene sequences were obtained from the 919 samples, with an average of 17,037 sequences per sample. The sequence depth range from 3,070 to 71,839. All samples with more than 3000 reads per sample are included in the following analysis.2. The pregnant women’s age range from 19.0 to 40.0 years, with mean age of 28.5 years. The delivery gestational age was 39.1 weeks and the birth weight are 3297.2 grams on average. The pregnancy outcomes of 23 participants were unknown due to loss to follow up, and 317 individuals were confirmed through prenatal care medical records. The incidence of premature rupture of membranes (PROM), preterm birth (PTB), chorioamnionitis, neonatal asphyxia was 28.7% (66/230),3.5%(11/317), 4.4%(14/317),5.0%(16/317) respectively. The rate of cesarean delivery was 32.8% (106/317).3. Based on criteria of routine vaginal examination, the vagianl discharge can be divided into three grades:good, median and bad. "Good" is based on the criteria: cleanliness (Ⅰ/Ⅱ), fungi (negative), trichomonad (negative),white blood cells (+/++), BV (negative). "Median" is based on the criteria:cleanliness (Ⅱ/Ⅲ), fungi (negative/positive), trichomonad (negative/positive),white blood cells (++/+++), BV (negative/positive). "Bad" is based on the criteria:cleanliness (Ⅲ/Ⅳ), fungi (negative/positive), trichomonad (negative/positive),white blood cells (+++/++++), BV (negative/positive). The grade of "Good", "Median" and "Bad" accounted for 58.9%,15.1%,26.0% and the medicine use were 0.6%,16.0% and 36.3% respectively.4. Community states types (CSTs) in vagina were clustered into five groups. Four CSTs (Ⅰ, Ⅱ, Ⅲ and Ⅴ)were dominated by Lactobacillus spp., namely L.crispatus (CST Ⅰ,41.8%), L.gasseri (CST Ⅱ,3.7%), Liners (CST Ⅲ,38.3%) and L,jensenii (CST Ⅴ,5.5%). CST Ⅳ, which lacks Lactobacillus spp., was further divided into CST IV-A (9.5%) and CST Ⅳ-B (1.2%). CST IV-A was characterized by a low occurrence of strictly anaerobic bacteria, such as Gardnerella, Prevotella, Atopobium, Streptococcus, Megasphaera, Bifidobacterium, Anaerococcus, Finegoldia, Sneathia and Mobiluncus. In contrast, CST Ⅳ-B had a high proportion of Atopobium in addition to Prevotella, Peptoniphilus and Porphyromonas,etc. The pH of CST IV-A (pH 4.8) was higher than any other groups (one-way ANOVA, P=0.000).5. The vaginal bacterial communities were highly conserved within individuals during pregnancy. More than 80% of individuals maintained the same CST through all gestation stages regardless of the CST the woman exhibited. On the other hand, nearly 20% of individuals changed their CST; this usually occurred from one Lactobacillus-dominated CST to another Lactobacillus-dominated one but rarely between Lactobacillus dominated CSTs and CST Ⅳ-A or IV-B. The vaginal suppository used during the pregnancy did no effect on the CSTs (Chi-square test, P> 0.05). Similar to the above results, approximately 80% of samples at the postpartum 6 weeks were classified as CST IV-A, which mean that only 20% of samples at the postpartum 6 weeks exhibited a CST dominated by Lactobacillus spp. The status of the postpartum recovery were not correlated with the choice of cerasean. (Chi-square test, P>0.05)6. The vaginal microbiome during pregnancy was dominated by Lactobacillus spp. with low diversity in the composition. While the microbial diversity of samples from the postpartum was unique. The median pH in postpartum samples (5.0±0.3) was significantly higher than that in gestational samples (4.1±0.2) (one-way ANOVA, P=0.000). In addition, the alpha diversity indices (i.e., Chao 1, PD whole tree, observed species and Shannon) were significantly higher in postpartum samples than in gestational samples (Kruskal-Wallis one-way ANOVA, P<0.05).7. No association observed between the CST and dverse pregnancy outcomes including:premature rupture of membranes (PROM), delivery mode (the rate of cesarean), chorioamnionitis, preterm birth (PTB), neonatal Asphyxia (Chi-square test, P>0.05).8. The area under the curve (AUC) for PROM, preterm, delivery mode, acute histologic chorioamnionitis and asphyxia was 0.803,0.714,0.65,0.672 and 0.652, respectively, indicating that the vaginal bacterial community is related to these adverse pregnancy outcomes and the vaginal microbiome has the potential to be used to predict these adverse outcomes. According to the model results, we selected the PROM with an AUC value of 0.803 for further exploration. Age was found to be the most import environmental factor correlated with PROM. We further divided the cohort into three different age categories (≤25 years,26 to 30 years and ≥ 31 years) and analyzed the rate of PROM with or without specific microbial taxa. The incidence of PROM in the ≥ 31 years age group with Megasphaera or Gardnerella was 3.13 and 2.15 fold higher than that in the ≤ 25 age group, respectively. Additionally, in the ≥ 31 years age group, the incidence of PROM with Megasphaera or Gardnerella was 2.39 and 2.10 fold higher than the incidence without these taxa (Chi-square test, P<0.05). The taxa Prevotella and Sneathia exhibited similar trends as Megasphaera and Gardnerella, but the results did not reach statistical significance (Chi-square test, P>0.05).Conversely, Lactobacillus spp. were negatively correlated with the incidence of PROM. The incidence of PROM without Lactobacillus spp. increased gradually with age. Compared with the≤25 years age group, the incidence of PROM without Lactobacillus spp. was higher in the≥31years age group (Chi-square test, P<0.05). When comparing the rate of PROM inthe oldest age group, the Lactobacillus spp.-positive groups, including L.reuteri, L.jensenii, L.iners and L.crispatus, had significantly lower rates than the Lactobacillus-negative group (Chi-square test, P<0.05).Conclusions:1. The vaginal microbiota varied substantially among subjects, even within CSTs, but was homogenous throughout the vaginal subsites (cervix, posterior fornix and vaginal canal) within individuals.2. Community states types (CSTs) in vagina were clustered into five groups. Four CSTs (Ⅰ, Ⅱ, Ⅲ and Ⅴ) were dominated by Lactobacillus spp., namely L.crispatus (CST Ⅰ), L.gasseri (CST Ⅱ), Liners (CST Ⅲ) and L.jensenii (CST Ⅴ). CST IV, which lacks Lactobacillus spp., was further divided into CST Ⅳ-A and CST Ⅳ-B.3. The vaginal microbiome during pregnancy was dominated by Lactobacillus spp. and highly conserved within individuals. While the postpartum was dominated by CST IV-A, which characterised by a high relative abundance of anaerobic bacteria, such as Gardnerella, Prevotella, Atopobium, Streptococcus, Megasphaera, Bifidobacterium, Anaerococcus, Finegoldia, Sneathia and Mobiluncus. The vaginal suppository used during the pregnancy did no effect on the CSTs. The status of the postpartum recovery were not correlated with the choice of cerasean.4. No association observed between the CST and dverse pregnancy outcomes including:premature rupture of membranes (PROM), delivery mode (the rate of cesarean), chorioamnionitis, preterm birth (PTB), neonatal Asphyxia.5. In elderly parturient women (i.e., older than 30 years), the existence of Megasphaera, Gardnerella, Sneathia and Prevotella was correlated with a higher risk for the occurrence of PROM, whereas Lactobacillus species exerted a protective role.