Introduction
Recent studies using next-generation sequencing of the 16S rRNA gene revealed the existence of an endometrial microbiota represented by
Lactobacillus and other bacteria [
1,
2]. Also, recent reports have suggested that human uterine microbiota is related to implantation success and that non-
Lactobacillus-dominated microbiota (NLDM), defined as < 90%
Lactobacillus spp., was associated with significant decrease in implantation, pregnancy, ongoing pregnancy, and live birth rates [
3]. Meanwhile, in our previous study, pregnancy rate of IVF was higher in patients with ≥ 80%
Lactobacillus-dominated endometrial microbiota compared to that with < 80%
Lactobacillus [
4], and
Bifidobacterium-dominant endometrium was suspected to be an acceptable environment for implantation [
4]. On the other hand, we also experienced a few cases who achieved pregnancy in spite of the non-
Lactobacillus-dominated endometrial status [
4].
To what extent the human uterine microbiomes are involved in implantation failure, and whether or not dysbiotic endometrium really have an impact on embryo implantation, is still not clear. This present study aimed to analyze the pregnancy outcomes of IVF patients presenting eubiotic or dysbiotic endometrium at the time of embryo transfer and to analyze what bacterial profiles are suitable for embryo implantation.
Discussion
This may be the first report in Japan analyzing endometrial microbiota concurrently at the time of embryo transfer and analyzing the direct impact of uterine microbial environment on pregnancy. Previous reports, including our previous study, indicated that
Lactobacillus dominancy in the endometrium was favorable in terms of pregnancy outcome [
3,
4]. Moreno et al. reported that the adverse effect of NLDM on pregnancy was more evident in subjects presenting dominant
Gardnerella and
Streptococcus genera [
3]. Also, bacterial genera such as
Enterococcus,
Enterobacteriaceae,
Streptococcus,
Staphylococcus,
Gardnerella,
Mycoplasma,
Ureaplasma,
Chlamydia, and
Neisseria are reported to be responsible for chronic endometritis (CE) [
19] and are suspected to adversely affect implantation. But in the present study, embryo implantation was not affected by the non-
Lactobacillus-dominated endometrium, and
Atopobium,
Gardnerella, and
Streptococcus dominancy was acceptable for implantation in a subset of patients (Table
2, Fig.
1). There were patients who achieved FBT pregnancies with 0%
Lactobacillus and 95.5%
Streptococcus, or 0%
Lactobacillus, 60.8%
Atopobium, and 21.9%
Gardnerella (Fig.
1); these patients currently have ongoing pregnancies beyond 16 weeks. Meanwhile, there were patients who could not achieve FBT pregnancy with 0%
Lactobacillus and 97.3%
Atopobium, or 0%
Lactobacillus, 54.3%
Gardnerella, and 13.7%
Atopobium (Fig.
2). Patient NH (Fig.
2) was pathologically diagnosed as CE 1 month after the microbial analysis;
Streptococcus was dominantly detected at the time of initial transfer (Fig.
2) and thus suspected to be the possible pathogen of CE and one of the causes of implantation failure. As the number of cases was limited in this study, we cannot conclude from these inconsistent results, but some of the bacteria other than
Lactobacillus or
Bifidobacterium spp. detected in the uterine cavity may be simply residents, not pathogens, of the upper female reproductive tract [
20]. Pathogenicity may differ by bacterial species;
Streptococcus agalactiae and
Streptococcus anginosus are classified in the same bacterial genus but may act differently in the endometrium. The potential regulatory mechanisms of each microbial species on implantation are still not clear, and currently, we can only speculate from the previous findings regarding chronic endometritis or bacterial vaginosis, etc. It may be speculated that NLDM may trigger an inflammatory response in the endometrium that affects embryo implantation, as inflammatory mediators are tightly regulated during the adhesion of the blastocyst to the endometrial epitherium [
3].
Streptococcus agalactiae is regarded as one of the major pathogens of CE [
19] and is also well-known as one of the leading causes of neonatal infections by vertical transmission from colonized mothers. CE has been suggested to contribute to diminished success rates of both spontaneous and IVF conceptions as well as obstetrical/neonatal complications [
19].
Atopobium vaginae and
Gardnerella vaginalis are known to be the major bacterial vaginosis-associated bacteria; they stimulate an innate immune response from vaginal epithelial cells and possibly contribute to the pathogenesis of bacterial vaginosis [
21].
Bifidobacterium is the dominant member of some vaginal microbiomes and is suggested to have the potential to be as protective as lactobacilli, contributing to a healthy vaginal microbiota in reproductive aged women [
22].
Lactobacillus spp.-dominated human vagina has been known to be protective against cervico-vaginal infections; however, the level of protection against infection varies by species or strain of
Lactobacillus, and some species, although dominant, are not always optimal [
23]. This may arise from the fact that antimicrobial factor (e.g., lactic acid) producing ability differs with each
Lactobacillus spp. [
24,
25]. In that sense, analyzing microbiota at the species-level resolution may be necessary for identifying the true pathogenic bacteria of the endometrium and avoiding over-intervention against non-
Lactobacillus microbiota; further studies are necessary to analyze the mechanism of how the pathogenic bacteria affect embryo implantation. Furthermore, not only the existence of pathogens but also the immune status of the host itself may be more critical for presenting clinical manifestations [
26,
27].
This study is different from the previous study [
3] in terms of background (ethnicity, age, etc.), transferred embryos, definition of eubiosis/dysbiosis, and the timing of sample collection; those factors may have contributed to the different research outcome of this present study. As for definition, Moreno et al. defined the bacterial status of the endometrium as
Lactobacillus-dominated microbiota (LDM, > 90%
Lactobacillus spp.) or non-
Lactobacillus-dominated microbiota (NLDM, < 90%
Lactobacillus spp. with > 10% of other bacteria), based on the composition of the microbiota in the endometrial fluid [
3], and reported that the presence of NLDM was associated with significant decrease in implantation, pregnancy, ongoing pregnancy, and live birth rates [
3]. Meanwhile, in our previous study, the pregnancy rate of IVF was comparable between LDM and NLDM; however, it tended to be higher in patients with ≥ 80%
Lactobacillus-dominated endometrial microbiota compared to those with < 80%
Lactobacillus [
4]; furthermore,
Bifidobacterium-dominated endometrium was also suspected to be an acceptable environment for implantation [
4]. Thus, in the current study, we defined a eubiotic endometrium as ≥ 80%
Lactobacillus +
Bifidobacterium spp. (eubiosis) and a dysbiotic endometrium as < 80%
Lactobacillus +
Bifidobacterium spp. with ≥ 20% of other bacteria (dysbiosis). Actually, we have analyzed our current data according to the previous criterion of LDM vs NLDM [
3] and found comparable implantation/pregnancy/miscarriage rates between LDM vs NLDM (Supplementary tables
1 and
2).
Considering the background of this study, the participants included in this study were relatively young and were not necessarily limited to recurrent implantation failure (RIF). Bacterial profiles and the pregnancy outcomes may have been different if the subjects had been limited to RIF patients. This study included only Japanese infertile patients and the results may have been different with people of other ethnicities. Preimplantation genetic testing for aneuploidy (PGT-A) or oocyte donations are prohibited in Japan at present and were therefore not performed in this study. If PGT-A had been used in this study, patient recruitment and outcomes may have been different, because PGT-A would have influenced embryo selection; but the patient profiles and the quality of transferred blastocysts were comparable between eubiosis and dysbiosis (Tables
1 and
2).
As for the timing of sample collection, previous studies had microbial analysis in cycles before embryo transfer [
3,
4]; but there are possibilities that microbiomes change over time, so we consider sampling timing as the critical factor of the study. Franasiak et al. [
1] analyzed endometrial microbiome from the transfer catheter of 33 patients undergoing euploid single embryo transfer and compared the bacterial profiles between patients with ongoing pregnancy vs those without ongoing pregnancy;
Lactobacillus was the top species call for both outcomes, and there were major species which appeared to vary by outcome, but the differences were not statistically significant [
1]. In our present study, the EF specimens were collected immediately before embryo transfer; thus, the microbial results were assumed to reflect the environment of the uterine cavity during the window of implantation among Japanese infertile patients. There may be a concern that EF specimens may have contained some endocervical fluid at the time of sample collection, or the transvaginal examination may have influenced the endometrial microbial results, thus limiting this study. Using a double-lumen catheter for EF aspiration [
3] may avoid the contamination of endocervical fluid; however, the cost for such a device was much more expensive compared to an IUI catheter as used in our study. Considering that the uterine cavity and the cervical canal is a continuum and that vaginal and endometrial bacterial communities are reported to be closely related in most of the subjects tested [
3], it can be assumed that the microbial results obtained in this study mostly reflected the microbial environment of the uterine cavity. EF can be aspirated in the same cycle as embryo transfer is performed without negatively affecting implantation [
28,
29]; thus, EF aspiration prior to embryo transfer in our present study was suspected to be harmless for implantation. Meanwhile, microbiota composition in EF may not fully reflect that in endometrial tissue [
30]; however, endometrial tissue sampling is harmful when performed immediately before embryo transfer. The exact amount of each aspirated EF was not measured in our study, but the quality of the samples was sufficient for microbiome analysis with a library concentration > 10 ng/μL.
Other limitations of this present study are the short follow-up period, limited study numbers, absence of analysis of other gynecological histories (e.g., bacterial vaginosis), lifestyle habits (such as sexual contact and sanitary conditions), past oral contraceptive usage, etc. All the EF samples in this study were collected in an HRT cycle. There may be a concern that hormonal therapy may have influenced on endometrial microbiome, but the hormonal regimen in this study was mostly uniform, and it is reported that endometrial microbiota is independent of hormonal regulation [
3]. To the best of our knowledge, there are no reports analyzing the relationship between endometrial microbiome and hormonal therapy; there are also no reports comparing the route of estrogen delivery and degree of change in the vaginal microbial community [
31].
As there were a small number of patients who had a miscarriage during this study period, the correlation between endometrial microbiota and miscarriage was not analyzed. There might be a correlation between miscarriage or preterm birth and dysbiotic endometrium at the time of embryo transfer, but this remains to be elucidated.
Antibiotic resistance is a growing problem and there is a risk of disturbing normal bacterial flora with the blind, cumulative usage of antibiotics. Not every single microbe other than Lactobacillus spp. may necessarily be eradicated from the endometrium. In order to avoid over-intervention with antibiotics, further studies are necessary to properly diagnose a “true dysbiosis” of the endometrium resulting in implantation failure.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.