Background
According to the WHO (
https://www.who.int/news-room/fact-sheets/detail/preterm-birth), 15 million babies are born preterm (before 37 completed weeks of gestation) every year, and preterm birth complications are the leading cause of death among children under 5 years of age, accounting for approximately 1 million deaths in 2015 [
1]. In 30 to 40% of the cases, a preterm premature rupture of membranes (PPROM) is preceding the preterm deliveries [
2]. Such a rupture of the membranes provides an entry point for ascending microbes to the uterine cavity which may contribute to the development of chorioamnionitis, colonization of the neonate, and subsequent maternal and neonatal morbidities [
3,
4].
Neonatal sepsis is a frequently fatal condition affecting neonates and children worldwide [
5,
6]. Based on the time of infection, neonatal sepsis is classified as early-onset neonatal sepsis (EONS), occurring in the first 3 days of life and being caused by bacterial pathogens transmitted vertically from mother to infant before or during delivery [
5,
6] whereas late-onset neonatal sepsis (LONS) is sepsis occurring after 72 h, which may be caused by vertically or horizontally acquired pathogens. The incidence of EONS in a healthy term birth pregnancy is lower than one case per 1000 births [
7], whereas the rate of developing EONS after PPROM ranges from 14 to 22% [
4,
8].
The importance of the vaginal microbiome for the prevention of urogenital diseases in women and for the maintenance of health has been more and more realized [
9]. In healthy women, the vaginal microbiota is typically dominated by different
Lactobacillus species. By producing lactic acid,
Lactobacilli create a low pH in the vaginal environment, which inhibit growth of potential pathogens [
10]. However, microbiota composition and stability differ between women and are mainly dependent on the menstrual cycle phases, sexual activity, and ethnicity [
11]. A dramatic depletion of
Lactobacillus spp. is considered a vaginal dysbiosis and is characterized by an enhanced growth of pathogenic organisms such as
Gardnerella,
Prevotella,
Atopobium, or
Fannyhessea [
12,
13]. Vaginal dysbiosis has been associated with several pathological conditions such as bacterial vaginosis, increased risk for sexually transmitted infections, preterm labor, preterm premature rupture of membrane, or chorioamnionitis [
4,
14,
15].
Lactobacillus spp. stability in pregnancy was suggested to represent an evolutionary adaptation to enhance reproductive fitness and protect against ascending infections [
16].
The clinical management of PPROM is challenging and has to balance the prolongation of the pregnancy to enable fetal maturation and the risk of infection and subsequent poor neonatal outcomes. In most countries, and specifically when PPROM occurs at earlier gestational ages (< 34 weeks), antibiotic therapy is recommended during pregnancy prolongation to reduce neonatal morbidity [
17,
18]. However, a recent analysis indicated that erythromycin treatment resulted in a shift towards dysbiotic community structures and depletion of
Lactobacillus spp. [
3,
4] thereby disrupting optimal communities. It was thus suggested that it is time to reconsider the role of antibiotic therapy in PPROM [
3].
Several perinatal factors like maternal nutrition, antibiotic use, and maternal stress, as well as maternal age during pregnancy, gestational age, mode of delivery, and breastfeeding, are described frequently to modulate the acquisition and development of gut microbiota in early life [
19]. However, it was generally assumed that the neonate is born sterile and only after delivery populated by bacteria. In contrast, in the recent years, various reports indicated the presence of bacteria or at least of bacterial DNA in the placenta, uterus, and amniotic fluid [
20,
21]. With a rupture of the membranes, the barrier to the fetus is damaged which makes bacterial contamination of the amniotic fluid and finally of the fetus feasible. In fact, 25% of preterm infants are reported to be born to a mother with intra-amniotic infection, most commonly due to invasion of the amniotic cavity by
Ureaplasma species [
22]. Placental pathogenic colonization seeded from ascending vaginal infection [
16] and the post-birth meconium microbiome and other fetal microbiota might therefore mirror the in utero microbial environment [
23].
The present study aims to assess the vaginal microbial composition of PPROM patients and its development under standard antibiotic therapy and to evaluate the usefulness of the vaginal microbiota for the prediction of EONS. It moreover aims to decipher neonatal microbiota at birth as a possible mirror of the in-utero microbiota in the context of EONS after PPROM.
Discussion
Preterm premature rupture of the fetal membranes (PPROM) precedes one-third of all spontaneous preterm births. Vaginal bacterial communities depleted in
Lactobacillus species and high in diversity have been reported as a risk factor for subsequent PPROM [
36]. In the present study, only roughly two-thirds of the pregnant women harbored a vaginal microbiota dominated by
Lactobacillus species at the time of hospital admission, in accordance with previous reports on women suffering from PPROM [
4,
36]. Typical vaginal pathogens such as
S. agalactiae,
Fannyhessea vaginalis,
G. vaginalis, or
U. parvum/urealyticus were frequently observed. As the uterine cavity, placenta, and fetus are subsequently exposed to these bacteria, the risk of chorioamnionitis, funisitis, and colonization of the neonate increases. Thus, antibiotic treatment was recommended [
38], also based on a clinical trial where antibiotic treatment had been shown to be associated with the prolongation of pregnancy, and fewer positive blood cultures [
39]. However, it was assumed as unlikely that for example treatment with erythromycin prevents ascending infections, as erythromycin concentrations in the vagina may reach levels effective against
Lactobacillus species but not against most vaginal pathogens [
3,
40]. A recent cohort study of mother-infant couples who delivered after PPROM showed an increased incidence of antimicrobial-resistant Gram-negative organisms on placental swabs and all cases of neonatal early-onset sepsis occurred in those who received erythromycin [
41]. Based on a clinical trial performed by Mercer and colleagues, ampicillin and amoxicillin together with erythromycin are widely used, with erythromycin being substituted by azithromycin due to fewer side effects [
42] and a recent study [
43] showed that women under such therapy had a longer median latency from time of rupture of membranes to delivery than women prescribed erythromycin. However, the microbiota analysis performed here shows, in agreement with previous reports [
4], that antibiotic treatment resulted in an increase in the vaginal microbial diversity and
Lactobacillus was eliminated from vaginal communities previously dominated by species of that genus. In contrast, various communities became dominated by
U. parvum with possible negative consequences [
44]. The commonly used single 1 g maternal azithromycin dose may not be optimal to maintain sufficient antibiotic concentrations for the expected 7-day course in the setting of PPROM [
45]. Interestingly, only 4 of 29 pregnancies where antibiotic treatment had occurred for < 48 h resulted in an EONS case after delivery (14%) whereas 10 of 34 (29%) extensively treated pregnancies gave rise to EONS cases. Even though this difference is not statistically significant, it is addressing the central problem of pregnancy prolongation after PPROM weighting the benefits of maturing and the potential risk of exposition time to ascending bacterial contamination of the amniotic cavity. It should also be noted that upon the termination of antibiotic treatment, thirteen out of fourteen communities remained dysbiotic and only one community recovered to the initial
L. crispatus dominance. Clearly, alternative methods to antibiotic treatment need to be evaluated such as the simultaneous use of probiotics [
46] or individualized management including amniotic fluid and vaginal microbiota analysis and targeted treatment [
47].
We had indicated above that
Anaerococcus obesiensis,
Anaerococcus lactolyticus,
Campylobacter ureolyticus, and
Howardella trended to be underrepresented in samples of mothers where a child suffers later from EONS, whereas
Escherichia/Shigella,
Enterococcus faecalis,
Facklamia,
Winkia neuii,
S. aureus, and
Eremococcus trended to be overrepresented. There had been no significant difference in the abundance and prevalence of
Ureaplasma between mothers where the child suffered later from EONS and those who did not, even though a case of
Ureaplasma caused EONS and severe bronchopulmonary dysplasia of the preterm infant was documented [
44]. In fact,
Ureaplasma species, the most common microbes found in amniotic fluid and placenta after preterm birth, have previously been correlated with chorioamnionitis and preterm labor [
48], and especially in the context of PPROM they have to be considered as a harmful pathogen for the preterm infant. Typical pathogens involved in chorioamnionitis, besides
Ureaplasma and
Gardnerella, comprise
Fusobacteria and specially
E. coli [
49] in accordance with
E. coli as a potential risk factor for EONS when detected in the vaginal microbiota before delivery. The organisms most frequently involved in EONS are
Streptococcus agalactiae and
E. coli, with additional pathogens comprising
Staphylococcus aureus,
Enterococcus spp., and
Haemophilus influenzae among others [
5] in accordance with the identification of
E. faecalis and
S. aureus as further potential risk factors for EONS when present in the vaginal microbiota.
Already two decades ago,
W. neuii has been characterized as a pathogen causing not only chorioamnionitis but also neonatal sepsis [
50]. Since then, this species has been identified as a cause of chorioamnionitis and neonatal sepsis also in other cases [
51,
52].
W. neuii (and
E. faecalis) were also shown to enhance
G. vaginalis virulence in bacterial vaginosis [
53] and may be a more important pathogen than previously thought.
Facklamia strains (obviously often misidentified as viridans group
Streptococcus) are reported as emerging pathogens and were isolated from human infections, including sepsis, genitourinary infections, or wound infections [
54].
Facklamia hominis has been described as a cause of chorioamnionitis and maternal sepsis and may have been responsible for an episode of sepsis in the neonate [
55]. Importantly, resistance against erythromycin and other antibiotics seems to be spread among
Facklamia isolates [
54‐
56], and their importance for chorioamnionitis and neonatal sepsis clearly needs to be further evaluated.
It is well documented that the extensive colonization of the human gut begins postpartum [
57]. Moreover, it was generally assumed that the neonate is born sterile, and only after delivery the neonate is populated by bacteria [
58], which would mean that the meconium is sterile in utero but rapidly colonized after birth [
59]. However, various publications have shown that meconium is not sterile [
60]. Moreover, the meconium microbiota was shown to share more features with the amniotic fluid microbiota than the maternal fecal and vaginal microbiota suggesting that the amniotic fluid microbiota contributed significantly to the seeding of the meconium microbiota [
23]. Given that the fetus swallows amniotic fluid throughout the second and third trimesters, such a sharing of a large portion of the microbiota is expected [
21]. Considering that the cohort analyzed here consisted exclusively of PPROM patients, where ruptured membranes offer the possibility of ascending bacterial colonization from the maternal vagina, it is also probable that in various of them a colonization of the amniotic fluid had occurred. Accordingly, typical pathogens of chorioamnionitis such as
Ureaplasma [
49] were highly abundant in various meconium samples analyzed here, and one sample, where the neonate later suffered from EONS, was dominated by
Fusobacterium nucleatum, previously described as responsible for various adverse pregnancy outcomes and even neonatal sepsis [
61]. However, besides these organisms, meconium commonly contains Gammaproteobacteria such as
E. coli and Bacilli such as
Enterococcus,
Staphylococcus, and
Streptococcus but also
Bifidobacterium and
Phocaeicola vulgatus [
59,
62], organisms also identified here as major members of the meconium microbiota. Importantly, the microbiota of the meconium was not influenced to a significant extent by the mode of delivery, and comparisons of the gut microbiota at delivery revealed that in both Cesarean and vaginally delivered infants,
Bifidobacterium and
Bacteroides/Phocaeicola were appreciably present [
60]. Notably, meconium samples harbored bacteria predicted to originate from the maternal stool or oral cavity. Thus, meconium communities may originate from a vertical ascension from the vagina and a hematogenous route through the placenta after translocation from the digestive tract [
63,
64] and postpartum from environmental sources.
It was already early suggested that the type of bacteria detected in meconium may influence childhood health [
65], and as an example, an association between neonatal jaundice and the meconium microbiome was observed [
66] with a higher abundance of
Bifidobacterium pseudolongum associated with a lower risk. Colonization with
Bifidobacterium early in life has also been associated with protection from necrotizing enterocolitis [
67] and late-onset neonatal sepsis [
68]. In fact, the early colonization with
Bifidobacterium may be crucial for human health status. They are important for shaping the immune system and interact with the host through cell surface-associated exopolysaccharides, fimbriae or pili, and secreted serine protease inhibitors, but they are also important metabolically through the degradation of diet-derived glycans and host-provided carbohydrates [
63]. It is, thus, not astonishing that
Bifidobacterium, here
B. longum, could be identified as putatively protective against EONS. It might be speculated, however, that Bifidobacteria are not protective per se but that they are indicative for a more mature gut. In fact, the gestational age of neonates the meconium of which harbored
B. longum was slightly higher compared to that of neonates where no
B. longum was detected (32.4 ± 2.0 versus 30.9 ± 3.2 weeks,
p = 0.028) and further studies with larger cohorts are necessary to define better the effect of
B. longum.
A. rectale was indicated as a second organism with a probably protective effect. In contrast to
B. longum, there was no difference in gestational age of neonates harboring
A. rectale and those that do not harbor this organism in their meconium (32.0 ± 2.04 versus 31.4 ± 3.0 weeks,
p = 0.377).
Conclusions
The vaginal microbiota analysis reveals complex bacterial communities in PPROM patients, and the analysis is capable of differentiating between Lactobacillus species. It can identify bacterial genera and species such as Facklamia spp. or Winkia neuii, which are not captured in conventional diagnostics but may be relevant for PPROM and/or EONS (or neonatal complications) and therefore relevant for adequate risk assessment and individual therapy.
The standard antibiotic treatment applied to PPROM patients contributes to the loss of potentially protective Lactobacillus species in EONS after PPROM leading to a rise in community diversity and persistence of potential pathogens and alternative methods to antibiotic treatment need to be evaluated.
The neonatal microbiota differs in composition by sampling site between the pharynx/rectum (surface) and meconium (processed amniotic fluid). Identification of potentially beneficial or harmful species in this specimen, where culture-based conventional diagnostics are often limited, may help to identify individual treatment strategies in preterm neonates after PPROM and an individual risk assessment.
Biomarkers, which have the potential to be used to predict EONS, could be identified from both the vaginal microbiota at birth as well as from the meconium. Microbiota analysis may therefore be a valuable diagnostic tool in the risk assessment of EONS after PPROM and has to be further evaluated in clinical settings.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.