Methods
Clinical data
Data were acquired by retrospective analysis of all singleton cases with asymptomatic short cervix of less than 25 mm in the second trimester until 33 weeks of gestation admitted to the obstetric department of Mainz university medical center during a 6-year-period (2011–2016). All patients were followed up until delivery. Women admitted with preterm labor or PPROM were excluded. The following data were collected: age, parity, cervical length, gestational age at admission, and gestational age at delivery.
Microbiologic analysis
At admission, vaginal microbial smears were taken using a sterile swab (eSwab, Copan, Bresica, Italy). Vaginal bacterial colonization was identified by means of bacterial culture, except for Ureaplasma spp. and Mycoplasma spp., which were identified by PCR testing. In cases of bacterial colonization, antibiotic treatment took place based on the antibiogram. As Ureaplasma spp. and Mycoplasma spp. PCR testing did not deliver antibiograms, these bacteria were treated with Macrolide antibiosis, either Clarithromycin or Erythromycin.
Vaginal infection and pregnancy
The occurrence of the bacterial taxa found in women undergoing preterm birth before or after 34 weeks was compared to those women who were admitted with cervical incompetence but gave birth after 37 weeks of gestation. Mean cervical length, prolongation of pregnancy and preterm birth rates were compared in women with the two most found bacterial taxa, Ureaplasma spp. and E. coli, to women found with other bacterial colonization and unaffected women.
Antibiotic treatment was analyzed for the two most found extra-cellular species, E. coli and Group B streptococcus.
Data acquisition and analysis were performed with SPSS 23.0 (IBM®, 2018). Comparison between groups took place using non-parametric testing with Kruskal–Wallis test for linear variables, and post-hoc analysis was performed using Holm–Bonferroni correction. Pearson’s Chi-square test was used for nominal scaled variables.
Discussion
This study gives an overview over the vaginal bacterial colonization in pregnant women with cervical incompetence. In nearly two-thirds of our patients, a vaginal bacterial colonization was found. Most found species were E. coli, Ureaplasma spp. and Group B streptococcus. All patients had an antibiotic treatment according to an antibiogram. 37.5% of our patients did not show any vaginal colonization with potentially pathogenic bacteria. Irrespective of the microbiological result of the vaginal smear, the number of preterm deliveries did not differ significantly. However, a total of 93 patients (28.44%) were concerned and gave birth preterm.
It remains hard to distinguish, if the shortening of the cervix occurs because of a vaginal infection or not. Our results make seem two different conclusions possible: first: a vaginal colonization is the reason for a shortening of the cervix and an adapted antibiotic treatment is effective. Second: the vaginal bacterial colonization is not the cause of cervical insufficiency. As there are patients without any pathogenic vaginal colonization, there must be different causes for the shortening of the cervix. The etiology is supposed to be multifactorial.
In women with preterm birth before 34 weeks, Ureaplasma spp. was the by far most found species, and E. coli was the most found species in women undergoing preterm birth after 34 weeks but not for those with early preterm birth.
It is known from literature that
Ureaplasma colonizations and infections are linked to cervical insufficiency [
4], preterm labor [
5], and PPROM [
6], resulting in a higher incidence of preterm birth [
7]. Vaginal
Ureaplasma spp. lipoproteins are suspected to upregulate Toll-like receptor 2 (TLR 2) expression leading to induction of pro-inflammatory cytokines like IL-8 [
8]. This pathway is likely to be one of the mechanisms explaining increased preterm birth rates after
Ureaplasma spp. colonization.
In the light of this knowledge, it seems to be more surprising that there was no significant link between the diagnosis of short cervix and vaginal Ureaplasma spp. and preterm birth in our study population.
If it comes to vaginal infections with, for example,
E. coli, the insight concerning potential pathogenetic pathways turns out to be less clear. As some observational studies found evidence for the association of vaginal
E. coli colonization [
9,
10] and spontaneous preterm birth especially before 34 weeks, it would be expectable that this association would also be found in our survey. But this was not the case and looking a bit more into the depth beyond the scientific level of a pure statistical correlation clinical observation unveils that there is not too much known concerning the pathophysiological pathway leading from vaginal
E. coli colonization to preterm birth. Recently, Spencer et al. [
11] succeeded to identify an inflammatory pathway after vaginal
E. coli infection in a promising mouse model, comparable models for humans still turn out to be lacking.
Most notably, overall, none of the found bacterial taxa turned out to be “game changing” towards occurring typically in those women giving birth preterm. Overall, two explanations are plausible for these findings. First, a clear link seen between vaginal infection and preterm birth, like in the case of
Ureaplasma spp. infections, is not known for many bacterial taxa. Second, the question arises, whether bacterial colonization in the vagina coincides with amniotic infection and inflammation as a trigger for preterm birth. As we know, inter alia, from Kayem’s and Combs’ studies [
12,
13], intra-amniotic colonization may be a trigger for preterm birth, it seems doubtable, whether the finding of bacterial colonization in a vaginal fornix posterior smear correlates with intra-amniotic infection. If not, the proof of vaginal colonization in this smears may not be suitable to predict the risk of preterm birth or even the necessity of antibiotic treatment. Regarding to Daskalakis’ and Thomakos’ findings, alternatively, measuring the intra-amniotic levels of inflammatoric markers as Interleukin-6 and -18 or TNF alpha could be more accurate in detecting women at risk of preterm birth due to intra-amniotic bacterial colonization, but, on the other hand, more invasive as amniocentesis would be needed [14,15].
Antibiotic treatment without clear evidence of its benefit should be seen with extreme precaution. First, the real benefit of antibiotic treatment is quite difficult to verificate as we did not include women with evidence of bacterial colonization without antibiotic treatment. As the role of vaginal bacterial colonization is not clear, it is hard to evaluate, if antibiotic treatment was effective in patients with cervical incompetence. Second, as we identified Penicillins with Betalactamase inhibitors as treatment of first choice in even in a majority in two of the most common bacterial taxa, the antibiotic regimes used have to be seen with a critical eye as it comes to side effects. As we learned from the ORACLE trial even more than 20 years ago, Penicillins with Betalactamase inhibitors (in the case of the mentioned trial co-amoxiclav) in prophylactic treatment after PPROM were seen as harmful regarding neonatal necrotizing enterocolitis [
16]. As the actual guidelines concerning the prevention and management of preterm birth do not recommend the use of these antibiotics [
17‐
19] and in our study the treatment intention was about reducing the risk of preterm birth, we assume these types of antibiotics not suitable for women with cervical incompetence as well. There is, once more limited, but nevertheless, encouraging evidence that preterm birth is linked to distinct changes in the vaginal microbiome in early pregnancy [
20,
21]. In 2019, Fettweis et al. delivered key findings on this issue in their comprehensive investigation concerning changes of the vaginal microbiome composition and preterm birth, concluding that the microbiome composition early in pregnancy may be most useful in the prediction of adverse pregnancy outcomes [
22].
In the light of this knowledge, the whole approach of identifying a shortened cervix in the second and early third trimester, taking vaginal microbial smears and trying to eliminate distinct bacterial taxa as a kind of pathology-causing target turns out to be at least doubtable. Of course, as preterm birth rates were as high as 28.44%, measuring the cervix in this phase of gestation succeeded to identify patients at risk of preterm birth. But as there were almost no significant differences in preterm birth rates and prolongation of pregnancy whether distinct bacterial taxa were found or not, this approach failed to be beneficial in the prevention of preterm birth. Therefore, as a key finding of our study, we estimate that the approach of measuring cervical length and searching for germs as being causative for a shortened cervix in late second and early third trimester is not an effective approach in the prevention of spontaneous preterm birth, because it is not sufficiently linked to intra-amniotic infection, and, when it comes to prevention in the narrow meaning of the word, because it is too late. Our findings support the hypothesis that, knowing that the ascension of certain bacteria as, for example, Ureaplasma spp., increases the risk of spontaneous preterm birth, the identification of women at risk should take place early in pregnancy and should rest on a sophisticated analysis of the vaginal microbiome rather than identifying single bacterial taxa alone.
Without doubt, our study has several limitations. First, it was a retrospective study with a limited number of patients. Sub-group analysis was not suitable as the number of patients would get too small in sub-groups. Likewise, the fact that the difference of rates of E. coli colonization pregnancies with and without preterm birth was not significant may also be biased by the limited number of cases.
Nevertheless, we are convinced that our study is helpful to induce a thorough reconsideration of daily clinical practice. The worth of the microbial smear as a diagnostic tool should be the identification of a potential target of treatment in our effort to prevent preterm birth. If the process of taking smears in patients with a shortened cervix is not effective in risk management or the influence of preterm birth rates, the identification of women at risk of spontaneous preterm birth should take place earlier in pregnancy (Table
4).
Table 4
Diagnostic findings, prolongation of pregnancy and preterm birth rates compared after diagnosis and treatment of E. coli, Group B streptococcus or Ureaplasma spp. *: lowest level of significance after Holm–Bonferroni correction +: Significance calculated by Pearson Chi-square testing
Cervical length admission (mm) | 15.67 | 17.91 | 18.14 | 16.49 | 17.67 |
Rank | 138.66 | 184.28 | 177.38 | 157.75 | 169.68 |
Significance* | p = 0.496 |
Prolongation of pregnancy (weeks) | 9.33 | 6.93 | 8.19 | 7.74 | 7.08 |
Rank | 158.69 | 117.96 | 138.75 | 133.33 | 118 |
Significance* | p = 0.084 |
Preterm birth < 34 weeks | 8% | 13.63% | 4.76% | 14.55% | 9.09% |
Preterm birth > 34 weeks | 24% | 16.67% | 14.28% | 14.55% | 17.27% |
Significance+ | X2 = 7.551, p = 0.478 |
More research is needed concerning how the knowledge about vaginal microbiome changes in women at risk of preterm birth can be transcripted in clinical practice in the means of clinical tools and preventive treatment efforts as preterm birth still turns out to be a major burden to perinatal health [
23,
24].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.