The Dynamics of the Vaginal Micro-Ecology During in Vitro Fertilization and Embryo Transfer (IVF-ET) Cycles and its Impact on Pregnancy Outcomes


 BackgroundAbnormal reproductive tract flora may cause infertility, and it may play a key role in the success of assisted reproductive technologies (ART). The obvious short-term changes in estrogen caused by clinical protocols with IVF-ET provide a unique perspective for us to assess the vaginal flora, shifting hormonal condition and investigate the potential associations of the vaginal micro-ecology with cycle outcome of pregnancy. The Vaginal Micro-ecology Evaluation System (VMES) as a tool to analyze the vaginal microbiomes in most areas of China. This study aims to apply the VMES to evaluate the dynamics of vaginal micro-ecology during IVF-ET, and investigate the correlations between vaginal micro-ecology with pregnancy outcome.Methods　150 patients were enrolled who underwent early follicular phase prolonged protocol IVF-ET due to tubal factors. The VMES is used to evaluate vaginal microbiology indicators of vaginal swabs obtained in different hormonal milieu during the IVF-ET cycle. The pregnancy outcomes were observed, if pregnant.Results　In our data, the prevalence of bacterial vaginitis (BV) accounts for 3.3%. During IVF procedure, the vaginal microbiome varied across hormonal milieu in some but not all patients. The proportion of BV, and unidentified dysbiosis were increased significantly on the day of human chorionic gonadotropin (HCG) administration. The vaginal micro-ecology on the day of HCG administration correlated with outcome (live birth / no live birth). The multivariable logistic regression model showed that the average age, the duration of infertility, and the vaginal micro-ecology after controlled ovarian hyperstimulation (COH) were associated with the live birth rate.ConclusionOur retrospective cohort study suggests that the VEMS has enabled discovery of unidentified dysbiosis shift in the vaginal micro-ecology during IVF-ET therapy. More importantly, the vaginal micro-ecology on the day of HCG administration was significantly associated with the live birth rate.


Introduction
Worldwide, approximately 10-15% of couples have di culty conceiving spontaneously [1]. ART is the cornerstone of contemporary infertility treatment. Commonly used ART procedures are invasive, costly and do not guarantee a pregnancy. As the success rate of ART is still low over the years, it is increasingly important to understand the hidden causes and to improve the e ciency of IVF techniques. Female reproductive micro-ecological system has been suggested to affect infertility, and it may play a key role in the success of ART, such as embryo implantation and pregnancy [2].
Infections are one of the disrupting problems in this arena. Multiple evidence indicated that infertile patients harbor a differential lower reproductive tract microbiota compared to fertile patients [3,4].
Therefore, pathological shifts of lower reproductive tract microbiota may be the cause or consequence of conditions in women's infertility. Research has shown that beside the known factors (the woman's age, duration of subfertility, antral follicle count, and the percentage of motile sperm used) in prediction models, outcome of assisted reproduction might be predicted by the composition of the vaginal microbiota [5].
To perform an IVF-ET cycle, pituitary down-regulation with Gonadotrophin releasing hormone agonist (GnRH-a) is widely used to prevent endogenous luteinizing hormone (LH) surge and spontaneous ovulation, and recruit more follicles in ART. As we all know,during prolonged pituitary down-regulation with full-dose of GnRH-a and controlled COH, the estrogen level in the body rises rapidly from extremely low level in the short term, even exceeding the level during pregnancy. This uctuation may directly affect the proliferation of vaginal mucosal epithelial cells, vaginal pH, cleanliness and lactobacillus ratio, resulting in vaginal dysbiosis [6]. However, there are few reports on how this short-term estrogen change affects the vaginal ora.
Our study utilized VMES to develop an in-depth and systematic understanding of the composition and ecology of the vagina microbial ecosystem in an IVF population. We aimed to answer the following questions: rst, is the change of certain vaginal micro-ecology associated with IVF or intracytoplasmic sperm injection (ICSI) treatment? Second, can the vaginal micro-ecology be used as an independent predictor for IVF or ICSI outcome? Furthermore, the VMES might provide the clinician with valuable information that could increase their understanding of the vaginal micro-ecological status, propensity for infection and treatment regimens for vaginal infectious diseases.
Herein, 150 patients who underwent GnRH-a prolonged protocol IVF-ET were enrolled in the study. The changes of vaginal micro-ecology before and after COH and its correlation with pregnancy outcome were observed and compared,providing new ideas for improving the pregnancy outcome of ART. Ovarian stimulation protocol and IVF-ET All patients in this study was performed GnRH-a prolonged protocol IVF-ET. It is composed of ovarian suppression with a GnRH-a, followed by COH with recombinant and urinary-derived gonadotropins, followed by HCG administration to induce oocyte maturation. Pituitary down-regulation was achieved with a single full-dose injection of 3.75mg GnRH-a (Leuprolide Acetate; Shanghai Livzon Pharmaceutical co., LTD) on day 2 of the menstrual cycle. Successful pituitary down-regulation was confirmed with follicle diameter<8mm, serum estradiol (E 2 ) <50pg/mL, serum LH<5IU/L, and endometrium thickness<5mm.
Then, COH with recombinant human follicle stimulating hormone (rFSH) ranging from 75 to 300 IU per day would start 32-38 days later. The dosage was determined according to patients' BMI, antral follicular count (AFC) and basal follicle-stimulating hormone (FSH) level. When two leading follicles reached a mean diameter of 18 mm, 6000-8000 IU HCG (Livzon Pharmaceutical Group Inc., China) was used for a trigger.
Transvaginal oocyte retrieval was performed 36-37h after HCG administration. Fertilization was achieved using standard IVF or ICSI. Gametes and embryos were handled separately according to standard laboratory procedures. Embryo quality was analyzed according to the Istanbul consensus workshop on embryo assessment. Fresh embryo transfer was performed on day 3 or day 5 after fertilization. Embryo transfer was determined according to the embryo quality, the thickness and state of endometrium. All embryos were at least good quality (Grade B) with 7-9 cells and less than 10% fragmentation and even symmetry, or high-quality blastocyst.
Luteal phase support was sustained from the oocyte retrieval day and was continued until the day of serum HCG testing. For women with positive HCG data, luteal phase support was continued until 10 weeks of gestation.

Serum E 2 concentration measurements
Blood was drawn, and serum prepared at three time points during each treatment cycle: the baseline; the rst day of Gn; the trigger day.
De nition of clinical outcomes The primary outcome was the live birth rate after fresh embryo transfer. The secondary outcomes include the clinical pregnancy rate, the implantation rate, the biochemical pregnancy rate, the early miscarriage rate. The secretion on the posterior fornix and upper 1/3 segment of the vagina was collected by rotating two sterile long cotton swabs to evaluate the vaginal micro-ecology. The swab 1 was taken on the day of pituitary down-regulation and the swab 2 was taken on the day of HCG administration, respectively. It also required menstruation to be clean for at least 3 days. Patients with LRTI were treated accordingly.

Evaluation of vaginal micro-ecology
Vaginal micro-ecology evaluation system (VMES) includes microscopic detection of ora density, ora diversity, dominant bacterial ora, pathogen, aerobic vaginitis (AV) score for AV, Nugent score for BV and the functional indicators.
(1) Low power microscope was used to observe the presence or absence of trichomonas in wet physiological saline, and AV score was performed at the same time.
(2) After Gram staining of secretion smear, 10×100 times oil microscope was used to check the ora density, diversity and predominant ora, and whether there were budding spores or pseudo hyphae, and Nugent score was carried out. Diagnostic criteria Flora density: "II" and "III" were de ned as normal, while "I" and "IV" were de ned as abnormal.
Flora diversity: "II" and "III" were de ned as normal, while "I" and "IV" were de ned as abnormal.
Predominant ora: when the dominant bacterial ora was Gram-positive rods, the predominant ora was normal; and other cases were de ned as abnormal.
Functional indicators: When H 2 O 2 was positive, it was determined as normal function of Lactobacillus; when the other four items were negative, they were de ned as normal.

Statistical analysis
Data were analyzed using IBM SPSS 25.0 statistical software. Quantitative data were described as mean ± standard deviation (Mean ± SD) or median. Results were analyzed using t tests for comparison between the study groups and the control group. Pearson χ2 or Fisher exact test were used for the comparison of proportion, as appropriate. Then, multivariate logistic regression was performed. P <0.05 was considered to indicate statistical signi cance.

Ethics statement
This study was a retrospective analysis of clinical outcomes, and the Institutional Review Board of the A liated Hospital of Qingdao University approved our analysis of the data.

Baseline Characteristics
The swab 1 was taken on the day of pituitary down-regulation and the swab 2 was taken on the day of HCG administration, respectively. For swabs 1, there were no signi cant differences in age, BMI, type of infertility, duration of infertility, basal FSH, total dosage of Gn used, and duration of stimulation among patients with AVM or NVM (P>0.05). (Table 4) The proportion of BV, and abnormal ora density/diversity patients evidently increased (P<0.05), while increasing the proportion of patients with intermediate BV, and VVC, but the difference in the latter is not statistically signi cant (P >0.05).

Discussion
Several studies indicated that abnormal reproductive tract ora may cause infertility, also may be related to miscarriage, premature rupture of membranes, and premature delivery [10][11][12]. Since IVF-ET involves transfer of embryos by a catheter through the cervix into the uterus, vaginal and cervical micro ora and pathogens and microbial contamination of the catheter tip have been suggested to affect implantation rates and pregnancy outcomes [13][14][15][16]. Thus, it is worth considering whether IVF outcomes be in uenced by the microbial ecosystem present during infertility treatment.
Vaginal microbiota undergoes important composition uctuations during women's life, sex hormones playing a key role in this scenario. The vaginal micro-ecological system composed of microbiome, endocrine regulation system, vaginal anatomy and local immune system has been well known [17]. The vaginal microbiome is an intricate and dynamic system [18]. At present, the application of qPCR and 16S rRNA sequencing technology can obtain 107-109 copies of vaginal micro ora genes from 1g of vaginal secretions [19]. However, culture-based technologies and 16S rRNA sequencing technology are limited by high cost and low throughput, hence only small numbers of samples have been analyzed, and the depth of sample analysis was not suitable for clinical applications. should be taken as reference indicators [13]. VMES could provide a new viewpoint for the comprehensive management of vaginal dysbiosis when infections occur.
Consistent with suppose, the proportion of BV, unidenti ed dysbiosis (abnormal ora density/diversity, and intermediate BV), and VVC were increased after COH. We were particularly interested in unidenti ed dysbiosis. Besides BV, we found that abnormal ora density/diversity increased signi cantly.
Indeed, BV is the most common form of vaginal dysbiosis [23]. BV may be asymptomatic in up to 50% of cases. The incidence of BV is signi cantly higher in patients with tubal infertility compared with patients with non-tubal infertility [24]. A large heterogeneity was observed among the studies in infertility patients, as the lowest reported prevalence was 4% and the highest prevalence 38% [25,26]. The early spontaneous miscarriage rate was 28% in BV patients compared with 17% in normal vaginal microbiota patients [27]. Several studies have shown that BV is associated with poorer result, and patients with BV are more likely to experience pregnancy loss after receiving ART [28,29]. Although BV has received the most attention, abnormal vaginal microbiota is not always BV and other conditions have separate effects on pregnancy outcome. In our data, the prevalence of BV accounts for 3.3%. The ve BV patients for swab 1 enrolled in this study are presented in S1. After COH, ve patients are also AVM. Moreover, two patients demonstrated non-pregnancy, and three patients had miscarriage. There were fourteen BV patients for swab 2, where twelve patients became from AVM, including BV, intermediate BV, and abnormal ora diversity, etc. Only three patients (21.4%) had a live birth. In addition, two patients became pregnant, but had a miscarriage (S2). Its mechanism has not been fully elucidated so far, and the biologically plausible explanation could be related to intrauterine infection caused by BV-associated bacteria ascending, resulting in chorioamnionitis and inducing uterine contractions [30].
The female genital tract commonly dominated by Lactobacillus spp. [31]. The presence of Lactobacillusdominated microbiome pro le has been associated with higher LBR after ART [32,33], and positively correlated with other reproductive success, namely higher CPR and IR [4,24,34]. Pelzer ES et al. [11] have been found that the non-Lactobacillus species,including Escherichia coli, Staphylococcus, Streptococcus, Enterobacteriaceae, and G. vaginalis species associated with poor reproductive outcomes. Different studies have shown that the correlation between AVM and ART results is contradictory. Studies using sequence-based technologies found that AVM had a negative effect on ART. However, studies using culture-based technologies found that AVM was not associated with ART outcome [35]. Hyma et al. [6] showed that the vaginal ora of infertile patients before IVF treatment was not signi cantly different from that of general gynecological outpatients, but the vaginal ora of some patients changed after receiving different COH treatments, and the abundance of vaginal ora on the day of embryo transfer is related to whether there is a live birth (P = 0.034). Singer et al. [28] concluded that women with AVM were approximately 1.4 times less likely to have a successful early pregnancy after an IVF compared with women with normal microbiota.
It is the rst study in which the data of VMES used as predictor for IVF outcome. In our current study, the live birth rate (LBR) was declined signi cantly with AVM on the day of HCG administration. The early miscarriage rate (EMR) of the AVM group were also signi cantly higher than that of the NVM group (P = 0.011). There were no signi cant differences in the biochemical pregnancy rate, implantation rate, and clinical pregnancy rate between the two groups (P > 0.05). As expected, multivariate logistic regression analysis and adjusted marginal means (95% CIs) of the live birth rate also contributed to eliminating the confounding factors. The average age, known as the best predictor of embryo quality, the duration of infertility, and the vaginal micro-ecology after COH were associated with the live birth rate in our regression analysis result. Thus, the data suggests that the vaginal microbiome on the trigger day affects pregnancy outcome.
In the case of the AVM pro le, women do not generally suffer from a clinically evident infection. However, our results indicate that women with AVM on the day of HCG administration have a limited chance of success after a fresh embryo transfer. A deep understanding of VM in ART may lead to personalized therapeutic options, such as vaginal administration of antibiotics, pre-or probiotics, aiming at speci cally modulation of AVM towards a more normal pro le [36]. These women could postpone the fresh ET, freeze the resulting embryos and transfer them later with a NVM pro le. Delaying a fresh embryo transfer when couples have a strong wish to have children could lead to discontent. However, Patients may get many bene ts, including reducing physical side effects, reducing the economic burden of treatment and mental burden. Our future research will elucidate whether modulation of the vaginal microbiota is possible and whether this may indeed improve outcomes of IVF and IVF-ICSI in patients with an abnormal vaginal micro-ecology pro le.
Our research has certain limitations. Firstly, we were not collected the data of vaginal micro-ecology on the day of fresh embryo transfer. Nevertheless, we could still bene t a lot from our data. Secondly, this study was a retrospective design from a single medical center. However, the large sample size, use of a multivariate regression model for a wide array of possible confounding factors, and marginal means of LBR adjusted by protocols rendered the conclusion relatively reliable. Our conclusions warrant further con rmation by larger, multicenter, prospective studies.

Conclusion
Our retrospective cohort study suggests that the VEMS has enabled discovery of unidenti ed dysbiosis shift in the vaginal micro-ecology during IVF-ET therapy. More importantly, the vaginal micro-ecology on the day of HCG administration was signi cantly associated with the live birth rate.