The characteristics and individual variations of the bacterial microbiota in the human reproductive tract are well described but little is known about the influencing factors. Here, we showed in a relatively large cohort of Caucasian pregnant women that both the reproductive history as well as gestational age strongly affect the abundance and prevalence of the dominant vaginal bacteria, many of which have well-established associations to gynecological and reproductive health. We observed that nulliparity associated strongly with L. crispatus -dominated vaginal microbiota in term and late term pregnancies. Prior pregnancies ending in spontaneous or induced abortion did not alter this association. We also demonstrated that the mode of delivery might have an impact on the vaginal microbiota as among women with one previous delivery, in those with a history of elective CS, the vaginal microbiota corresponded to that of a nulliparous woman, while after one either vaginal delivery or delivery by emergency CS the microbiota diverged. On the other hand, increasing number of prior deliveries associated with decreasing prevalence of L. crispatus. Overall, the vaginal microbiota in the late third trimester varied according to the duration of gestation, and especially among women with no previous deliveries L. crispatus was more prevalent the longer the duration of the pregnancy was.
Interpretation of the results in light of previous evidence
We found that the vaginal microbiota differed profoundly between nulliparous and multiparous women, indicating that the reproductive history is reflected in the vaginal microbiota at or close to delivery. Our results confirm and extend similar findings reported earlier for the first trimester of pregnancy (between 8- and 12-weeks of gestation)11, and for non-pregnant reproductive aged women5. The depletion of vaginal lactobacilli postpartum compared to pregnancy is well established and coincides with the parturition-induced drop in estrogen levels19,29. However, the duration of Lactobacillus-deficient postpartum microbiota signature has not been thoroughly studied, though there are indications that it remains up to one year postpartum30. Recent research on US-based cohort suggests that such signature is also present in women who delivered by an unlabored cesarean surgery31, while another recent study identified birth mode-dependent differences on the vaginal microbiota of Chinese women sampled at comparable time 6 weeks postpartum32.
A successful pregnancy requires tightly coordinated and balanced interplay between host innate immune defenses, mucosal immune responses, and the resident microbiota33,34. Although we studied only vaginal microbiota, the inflammatory and/or adaptive and innate immunity profiles are also anticipated to differ based on pregnancy history. For instance, immunosuppressive regulatory T cells (FOXP3+ CD4 T lymphocytes), which recognize paternal antigens and are essential for maternal fetal tolerance, accumulate during the first pregnancy and persist to some level in the maternal circulation after delivery35,36. Later in the subsequent pregnancies these cell populations re-expand in a quicker manner compared with the initial pregnancy35. A pregnancy-induced memory cell response and pregnancy alloimmunization has also been proposed to be behind higher transplantation graft rejection in women with prior pregnancies37.
As a further support for immunological rather than direct and local microbiological effects, parity has been shown to affect not only the vaginal microbiota but also gut microbiota during subsequent gestation in a pig model38. This lends further support for the immunological effects as a mechanism for persisting microbiota changes as a result of previous delivery. Hence, our parity-related microbiota findings could reflect the state of the local inflammatory processes and potentially suggest some form of an immunological memory from prior labor. This is also supported by our suggestive finding that only previous vaginal delivery or emergency CS, both involving the physiological process of labor, were associated with changes in vaginal microbiota composition. We observed no differences in the relative abundances of L. crispatus or L. gasseri after elective CS, albeit the sample size in this subgroup was small. A recent study composed of healthy Chinese non-pregnant, reproductive age women5 showed that the vaginal microbiota is significantly associated with past reproductive events. Notably, questionnaire-based information on number of pregnancies, pregnancy history, mode of recent delivery, and number of deliveries were shown to be significantly associated with the microbiota, especially L. crispatus, corroborating our findings. Although intriguing, the hypothesis on the role of immunological memory warrants further research e.g. on the immunologic pathways activated during labor, and their possible differences between nulliparous and multiparous women. These studies would also help to elucidate whether the lower relative abundance of L. crispatus is a primary or secondary effect of previous delivery i.e., a persisting postpartum bacterial signature or reflecting an immunological imprint from the previous parturition process.
The ultimate physiological stimuli leading to the onset of labor remain unclear despite intensive research. Although Lactobacillus-depleted vaginal microbiota has been associated with increased risk of PTB39, studies on microbiota composition at term and especially at late term and prolonged gestation are scarce. The duration of parturition is usually shorter in women with prior deliveries compared to nulliparous40, and nulliparity is a known risk factor for prolonged gestation41. We also observed that L. crispatus dominance increased along the progression of gestational weeks at late pregnancy, especially among nulliparas. This could be explained by higher estrogen levels at late pregnancy as estrogen concentrations increase with gestational age42, and due to the higher estrogen levels during the whole pregnancy among nulliparas than multiparas43. Our results support the data of Romero et al.44 who showed in their longitudinal study that the relative abundance of Lactobacillus spp., including L. crispatus, increased as a function of gestational age. Dissimilar observations have, however, been presented as Avershina et al.21 first reported an increasing diversity of the vaginal microbiota at the onset of labor compared to samples taken at 36 weeks of gestation. Later, Rasmussen et al.20 observed that there was a gradual decline of Lactobacillus spp. from week 24 of pregnancy until birth while only genera Enterococcus and Granulicatella were associated with gestational age at birth. Their sample size (n=57) was small compared to ours and the last samples taken later during birth after rupture of membranes so the results might not be thoroughly comparable. Furthermore, the proportion of nulliparous women, in whom the association between higher gestational age and L. crispatus dominance was more pronounced in our study, was lower in their study (40% vs 58%).
Ripening of the cervix in normal parturition is characterized by inflammatory changes, including the activation of leucocytes and increasing levels of pro-inflammatory cytokines which remodel the extracellular matrix of the cervix45–47. An increased ratio of L- to D-lactic acid isomers may alter the cervical tissue integrity by activating matrix metalloprotein inducers29,48. Low concentrations of D-lactate versus L-lactate have been seen in association with L. iners and G. vaginalis48, whereas concentrations of D-lactate are higher in L. crispatus -dominated vaginal microbiota, reflecting the known metabolic characteristics of vaginal Lactobacillus species49. In in vitro studies an inflammatory response evoked by L. crispatus has been shown to be weaker than by L. iners or anaerobic bacteria50. Hence, our findings on the high abundance of L. crispatus in late and postterm pregnancies may reflect decreased inflammatory bacterial signals. While these findings from a cross-sectional study cannot address causality, it is tempting to hypothesize that the vaginal microbiota and its interactions with the host immune system could play a role in the maintenance of gestation and initiation of spontaneous term or late term labor.
Identifying women susceptible to prolongation of pregnancy and its associated complications would be of great clinical benefit. Prolonged pregnancy increases the risk of stillbirth, neonatal morbidity and mortality, and the risk of cesarean delivery associated to the induction of labor is higher than in inductions in earlier pregnancy weeks51. Nulliparity is one of the known risk factors for prolonged pregnancy52, and nulliparas form the majority of women undergoing induction of labor due to prolonged pregnancy28. Previously Lactobacillus abundances have been shown to decline gradually toward birth20. Prevalence of L. crispatus -dominated vaginal microbiota, however, was higher after the due date among nulliparas in our study. Although this finding alone is not sufficient to predict the duration of gestation, more detailed research in inflammatory and immunological pathway activation in addition to the microbiota composition could well help to identify women at risk of prolonged gestation. Also, as different clinical scoring systems to predict the success of labor induction have been proposed53,54, we may speculate that the vaginal microbiota could potentially be included in the evaluation to decrease the rates of unsuccessful labor inductions which could decrease maternal and neonatal complications and affect maternal birth experience.
The strengths of our study include the prospective setting and relatively high number of study subjects and the use of modern sequencing techniques for analyzing the microbiota. The study population was homogenic with comprehensive outcome and background information from medical records and the specific questionnaire designed for the study. To our knowledge, this study is the first to characterize the vaginal microbiota at late pregnancy stages in a large cohort of women with covering obstetric records. The limitations include sampling at a single timepoint for each participant. A longitudinal study with samples from early pregnancy until delivery would allow us to see the potential intraindividual changes in the microbiota with advancing gestation. The homogeneity of our study population, despite it being also an asset, does not let us interpret whether the results can be generalized to women with different ethnic and biogeographical backgrounds knowing that ethnicity-related differences in the vaginal microbiota exist also during pregnancy29,55.
In conclusion, previous pregnancy history and mode of previous delivery have a strong association with the composition of vaginal microbiota close to parturition, while among nulliparous women, gestational age associates with the microbiota. The findings are novel and add to the few currently known factors associated with the duration of gestation (e.g. the relationship of maternal age, nulliparity and obesity to postterm pregnancy)41. Whether the microbiota changes just reflect or actively contribute to the underlying immunological processes and mechanisms remain to be studied. As the vaginal microbiota is, at least conceptually, modifiable, our findings highlight the importance of future studies to understand the nature of implicated host-microbiota interactions and to investigate its potential for diagnostic and therapeutic approaches.