The reproductive microbiome is an emerging topic in the field of obstetrics and gynecology [1–6]. In particular, the microbiome of the uterine cavity, specifically the endometrium, an anatomical niche where low-biomass microorganisms can modulate the local immune environment, is of particular interest [7]. Interest is primarily driven by the influence of the microbiota on embryo implantation and placenta formation, potentially affecting fertility and the development of obstetric complications [2][4][8][9]. Recent studies showed that in the context of in vitro fertilization (IVF), an endometrial microbiota, which was not dominated by lactobacilli (defined as <90% Lactobacillus spp.), was associated with a significant decrease in implantation, pregnancy, and live birth rates [10–12].

Sequencing-based bacterial detection methods are currently the cornerstone of microbiome assessment in low-biomass anatomical sites. These metabarcoding methods are based on the amplification and sequencing of the bacterial 16S ribosomal RNA (rRNA) gene containing nine hypervariable regions (V1-V9), which allows the differentiation and quantification of different microbial species present in a particular sample [13–15]. A significant limitation in the study of the endometrial microbiome is the currently used methods of sampling for its analysis, which have not received sufficient attention so far. At the same time, these procedures have one common disadvantage, namely, the possibility of contamination of endometrial samples with cervical or vaginal microorganisms, i.e., contamination from anatomical areas with high biomass and density of microorganisms (10⁷ and 10⁵ and higher than in the endometrium), which can completely negate the results of studies [4][10][16]. Thus, studies investigating the endometrial microbiome using a transcervical sampling catheter and studies of a histosample of endometrium obtained at hysterectomy and reaching the endometrial cavity showed radically different results both in terms of microbial density and colonizing species [1][17–20].

In the present study, the authors proposed to use a proprietary designed two-lumen uterine probe with collet, commonly used for embryo transfer, to obtain endometrial samples and assess the microbiota. This uterine catheter system is expected to reduce possible microbial contamination of the cervix and vagina. To investigate the reliability of this method of sample collection, the authors compared endometrial and vaginal microbiota in infertile women scheduled for IVF and frozen embryo transfer. As a secondary objective, the authors also assessed the association of the resulting microbiota with subsequent pregnancy probability.

The aim of the study was to evaluate the microbiome of the vagina and uterine cavity using a collet-guided uterine catheter in infertile patients before IVF.

The study included 73 women (mean age – 34.1±3.7 years) with infertility before preparation for the IVF cycle or intracytoplasmic sperm injection, having blastocysts for cryotransfer in 5 fertility centers of Krasnodar (Clinic of Kuban State Medical University (KubGMU), Family Planning Center of the Regional Perinatal Center of the Children’s Regional Clinical Hospital (CRCH), Fertility Center of the clinic “Ekaterininskaya”, Male and Female Health Clinic “OXY-center”, Fertility Clinic “Embrio”) in the period from January 2021 to January 2023 (Table 1).

Exclusion criteria: 1) current diagnosis of pelvic inflammatory disease; 2) presence of hydrosalpinx; 3) clinically significant endometrial cavity abnormalities including fibroids, endometrial polyps and uterine septum, Asherman’s syndrome; 4) antibiotic therapy within the last month; 5) hormonal therapy (progestins, estroprogestins, or gonadotropins) within the last month; 6) abnormal uterine bleeding; 7) embryo transfer scheduled in the same menstrual cycle. Women who agreed to participate were informed about the aim of the study, possible discomfort of the procedure, and possible risks, and signed informed consent.

Samples from the uterine cavity for microbiologic examination were taken from women in the lithotomic position on a gynecologic chair with inspection in speculum (Cusco type) between days 15 and 25 of the natural menstrual cycle. Sampling was performed with transabdominal ultrasound imaging. First, a sample of vaginal microflora was taken from the posterior vaginal vault (sterile swab). Then, after the removal of cervical mucus with a sterile swab, avoiding contact with the vaginal walls and non-sterile surfaces as much as possible, a uterine two-lumen collet probe was inserted, and endometrial material was sampled. After that, the collet was closed, and the probe was removed (Fig. 1).

The contents of the probe were suspended in 150 μL of sterile saline solution previously prepared in a 1 mL sterile Eppendorf tube. Then, the distal part of the catheter (2–3 mm) was broken off and lowered inside the tube along with the saline solution (the tube was stored at ‑80 °C).

At the time of evaluation, samples from the vagina and uterine cavity were thawed, and bacterial DNA was isolated for its examination by real-time PCR (RT-PCR) using the Femoflor-16 reagent kit (NPO DNA-Technology, LLC, Russia): DNA was isolated from 100 μl of the sample using the reagent kit “Proba-GS”. This test allows for the simultaneous amplification and detection of target nucleic acids of microflora associated with bacterial vaginosis: Gardnerella vaginalis, Atopobium vaginae, Megasphaera, Mobiluncus spp., Bacteroides fragilis, as well as Lactobacillus spp. (L. crispatus, L. gasseri, L. jensenii). The assay provides automatic interpretation of the microflora using quantitative analysis.

Sequencing of 16S rRNA of bacterial genes was performed using the Microbiota B solution kit for hypervariable regions V3-V4-V6 by using degenerate primers including 2 assays: EMMA® (Endometrial Microbiome Metagenomic Analysis – analysis of endometrial microflora to assess reproductive prognosis) and ALICE® (Analysis of Infectious Chronic Endometritis – analysis of pathogenic endometrial microflora (Igenomix, Spain)) [14][15]. The resulting amplification of hypervariable regions V3-V4-V6 allows for the identification of most of the bacterial populations present in the microbiota. Sequencing data were analyzed automatically using the MicrobAT software (Microbiota analysis Tool, Igenomix, Spain), which provides a report for each sample with the assignment of operational taxonomic units.

The data were analyzed using the Statistical Package for Social Sciences software (SPSS 23.0, IL, USA). The data were presented as number (%), mean ± SD or median (interquartile range – IQR). Comparisons were performed using Student’s T-test and Fisher’s nonparametric test (p<0.05 values were considered statistically significant). The relative proportion of different bacterial genera was calculated using only the total number of informative reads as the denominator. The endometrial microbiome was considered lactobacillus-dominant if the relative proportion of Lactobacillus was greater than 90%. The Shannon index (SH) and Shannon equivalence index were calculated by the MicrobAT software.

The study was approved by the Ethical Committee of KubSMU (protocol No. 17 dated January 12, 2021) and was conducted in accordance with the Declaration of Helsinki (revision 2013, Brazil), the rules of Good Clinical Practice (GCP; 2016, Astana) and clinical practice in the Russian Federation (Decree of the Ministry of Health of the Russian Federation No. 200n, 2016). Patients were examined according to the Decrees of the Ministry of Health of the Russian Federation No. 1130n dated October 20, 2020 “Approval of the procedure for medical care in obstetrics and gynecology”, No. 803n dated July 31, 2020 “The procedure for the use of assisted reproductive technologies, contraindications and restrictions to their use”, and clinical recommendations “Female infertility” (2021).

The interquartile range (IQR), the total number of sequence readings for endometrial and vaginal specimens, was 2583 (385–6083) and 3822 (910–6920), respectively. The Shannon IQR indices for them were 2.73 (2.49–3.17) and 1.70 (1.44–2.22) (p<0.001), the Shannon Equitability IQR indices were 0.57 (0.52–0.74) and 0.44 (0.40–0.53) (p<0.001). Lactobacillary microflora predominance was noted in 30.1% (22/73) of endometrial and 53.4% (39/73) of vaginal samples; 86.3% (19/22) of women who had Lactobacillus spp. as the most common species in endometrial samples also had Lactobacillus as the most common species in vaginal samples. Microflora that could be classified as Lactobacillus-dominantus (Lactobacillus >90%) was found in 8.2% (6/73) of endometrial and 41.1% (30/73) of vaginal samples (the Spearman correlation index between the proportion of Lactobacillus spp. in endometrial and vaginal samples was Rho=0.53, p<0.001 – weak correlation) (Table 2).

The dominant bacterial genera of the endometrial and vaginal microbiota coincided in only 9.6% (7/73) of women. For the endometrium, the most common bacterial genera were Lactobacillus, Pelomonas, Propionibacterium, Pseudomonas, Streptococcus, and Escherichia coli (Tables 2–3). For the vagina, these were Lactobacillus, Gardnerella, and Bifidobacterium.

At the same time, the data of multiplex RT-PCR in endometrial samples did not significantly coincide with the results of 16srRNA sequencing, which seems logical, since the PCR detection system is primarily aimed at detecting the microflora responsible for the development of bacterial vaginosis and is not tropic to the endometrium (Table 3).

Further, 89.0% (65/73) of the patients underwent the transfer of unfrozen embryos of 3–5 days (in 8 women, the transfer was canceled because the optimal endometrial thickness was not reached). Only clinical pregnancy (visualization of cardiac activity at week 6) was registered, which was observed in 31.5% (23/73) of women. Live birth was registered in 24.7% (18/73) of patients (5.5% (4/73) had spontaneous miscarriage, 6.8% (5/73) had preterm delivery (22–36 weeks of gestation), and the remaining 19.2% (14/73) delivered at term).

Out of 23 women with successful IVF outcome (parameter – registered clinical pregnancy), the predominance of lactobacillary microflora in endometrial and vaginal samples was noted in 11 patients (47.8%) – 2 (8.7%) in endometrium and 9 (39.1%) in the vagina (p=0.16). No significant differences were noted between IVF positive and negative groups: Shannon’s index for endometrial samples between pregnant and non-pregnant women was 2.41 (1.12–3.90) and 3.29 (2.84–3.82) (p=0.036) respectively, Shannon’s equitability index was 0.76 (0.57–0.87) and 0.55 (0.51–0.64) (p=0.002), which may indicate higher biodiversity in relation to the establishment of pregnancy, but, given the sampling size, is not significant (Figs. 1–2).

A statistically significant difference was found for both indices in endometrial microbiomes (p=0.036 and p=0.002, respectively), and conversely, no significant differences were observed for vaginal microbiomes.

The results of this study in the non-pregnant women that were obtained using the author’s method of obtaining endometrial samples revealed significant differences in the taxonomy between the composition of the endometrium and the vaginal microbiome, which confirms its validity. At the same time, no confirmation was obtained about the expressed positive influence of the dominance of lactobacillary endometrial microflora on IVF success, nor about the negative influence of opportunistic flora responsible for the development of local genital dysbiosis. Interestingly, higher Shannon indices were observed among women with a positive embryo transfer outcome (clinical pregnancy).

The vagina and cervix are areas with a high biomass content, which in most cases causes significant contamination of biomaterial during their passage (the concentration of microorganisms in the cervix is estimated to be 10⁵ higher than in the endometrium), which is a significant limitation in the assessment of endometrial microbiocenosis, which determines the relevance of the search for new methods of transcervical collection of biomaterial [10]. At the same time, the assessment of endometrial and vaginal microbiome composition in this study and, what is especially important, the assessment of its influence on the effectiveness of IVF programs, differs in many respects from the data of other researchers [1]. Both the authors of this study and most other authors do not question the similarity between the vaginal and endometrial microbiocenosis due to their interdependence and proximity. Still, in the authors’ opinion, the observation of differences is an argument in favor of the accuracy of our sampling technique.

There were attempts to overcome the contamination of endometrial specimens with cervicovaginal contents in clinical practice. Liu et al. (2019) used a double-sheath catheter to obtain an endometrial sample [21]. Their recorded results of more expressed endometrial contamination compared to Lactobacillus vaginal samples are inconsistent. Meanwhile, other researchers (Carosso et al., 2020) who used two-lumen catheters for embryo transfer, showed an average occurrence rate of Lactobacillus (27%), which correlates with the results of most studies [22]. Thus, Verstraelen et al. (2016) used a transcervical device with a sheathed silicone brush designed to collect endometrial histosamples and failed to show Lactobacillus dominance in the endometrial microbiome, with a rather “rough” impact of this cytobrush being reported to disrupt the integrity of the endometrium [23]. Qiu et al. (2021), who performed hysteroscopic sampling to avoid contamination, reported an even lower Lactobacillus detection rate of 7% (a significant disadvantage of this method is its complexity and invasiveness) [24]. Thus, the use of two-lumen catheters for embryo transfer has been the most successful so far [25][26]. However, with this method, there is still a risk of contamination when the internal catheter passes through the external catheter. The model used in the present study, by opening/closing the collet mechanism, helped to overcome this risk.

The authors of this study also attempted to draw parallels between the endometrial microbiome composition and IVF outcome, but the obtained results had some contrast with other studies [12]. It is important to mention that the occurrence rate of lactobacillus-dominant cases in the patients was scarce, and there was a lack of association between the composition of the endometrial microbiota and IVF outcomes. That said, the results of this study do not question the findings of Moreno et al. (2016) on the association of lactoflora dominance and positive outcome, but question the fact that these authors actually had data on the endometrial microbiome (it is more plausible that their conclusions were based on the evaluation of the cervical microflora). An interesting and unexpected result of the present study is the association between higher alpha diversity and pregnancy probability in IVF protocols: 2 measures of diversity (Shannon and equitability indices) revealed statistically significant differences between women with positive and negative embryo transfer outcomes. At the same time, the authors believe that it is necessary to conduct further studies to assess the biodiversity of the endometrial microbiota and its role in the realization of fertility potential.

Finally, the data obtained with multiplex RT-PCR (Femoflor 16) indicate that this method is not completely suitable for the analysis of endometrial samples, but it is able to confirm differences between vaginal and endometrial colonizing microorganisms.

The obtained data confirmed the efficacy of using two-lumen probes with collet to test the endometrial microbiome, which creates new perspectives for both studying its biodiversity and determining its role in endometrial receptivity and female fertility.