Infertility is one of the conditions of the reproductive system that generates disability due to impairment of function [1]. This condition is estimated to affect 186 million individuals and 48 million couples worldwide [2]. In vitro fertilization (IVF) is an important technology that can assist women and couples to achieve pregnancy. Currently, over 5 million children have been born through IVF technology [3], and its use is constantly increasing. In light of the increasing use of IVF, the potential for IVF-induced health problems in these infants has been evaluated in recent studies. The main findings were that IVF was associated with increased risks of malformation and functional disorders, and with worse peripartum outcome [4]. The specific causes remains unknown, but may be related to parental factors or to the IVF technology utilized.

In recent years, the human microbiota has received considerable research interest because of its central role in numerous aspects of health and disease. These include protection from pathogens, nutrient metabolism, promotion of immune system development, and its influence on various mental and neurological functions [5, 6, 7]. The vaginal microbiota (VM) composition is believed to be associated with the reproductive health of women. Emerging evidence indicates that the maternal microbiota composition prior to delivery influences the development of infant immunity [8]. Maternal vaginal microbes can translocate to the placenta and amniotic fluid, thereby influencing preterm birth [9]. Current evidence suggests that preterm birth is more common in IVF pregnancies than in spontaneous pregnancies. The VM could therefore be a causative factor in the link between IVF pregnancy and risk of preterm birth [10].

The female vagina is colonized by an array of commensal microorganisms, among which Lactobacillus is the most dominant species in healthy women of reproductive age. The VM is particularly dynamic, with fluctuations occurring in response to sexual development and sexual intercourse, pregnancy, hormone levels, as well as personal hygiene. Pregnancy is a special physiologic period for women and is accompanied by hormonal and immune changes that can modulate the structure and function of microbiota, thereby making it different to that of non-pregnant women. During pregnancy, the VM is less diverse and more stable, with predominantly Lactobacillus species [11]. Women who undergo IVF may experience routine clinical interventions, including medication to stimulate ovulation and embryo implantation. All of these factors can affect the structural adjustment of female vaginal microbes. Amplicon sequencing was recently used to analyze the VM of females undergoing IVF. The results suggest that VM may be useful in predicting the outcome of IVF, since women with low percentages of Lactobacillus were less likely to have successful embryo implantation [12]. The VM composition during the first trimester (T1) of gestation could also explain the association between IVF pregnancy and preterm birth risk [10]. Longitudinal analysis of the VM in 45 preterm births and 90 term births conducted by the integrative Human Microbiome Project found that preterm women had significantly lower levels of Lactobacillus crispatus, but higher levels of BVAB1, Sneathia amnii, TM7-HI, and a group of prevotella species [13]. So far, however, very few longitudinal studies have been carried out on the VM of women who undergo IVF. In the present work, we performed the first longitudinal cohort study to analyze the dynamic changes in VM during all three trimesters (T1-T3) of gestation in women undergoing IVF.

The VM is a complex, dynamic community that changes constantly in response to hormonal fluctuations. Its composition is known to play important roles in vaginal and reproductive health, and has been associated with the risks of miscarriage, preterm birth, and sexually transmitted disease [14]. Currently, IVF is an effective method for conception in women who fail to conceive naturally. Dynamic changes in the VM of women who become pregnant through IVF are still poorly understood. Using high-throughput sequencing, we found that VM is dominated by Lactobacillus species during pregnancy. The alpha-diversity in women who undergo IVF decreased with increasing gestation time. The variation in microbiota diversity was the same in the IVF and control groups. The greatest difference between the two groups was observed during T1, and this gradually decreased during gestation. Women in IVF group received hormonal treatment to stimulate ovulation and followed a progesterone supplementation treatment until 10 weeks of gestation. The first trimester vagina samples were collected from 10 weeks to 13 + 6 weeks. During this period, although the clinical practice and hormone supplementation were no longer performed, the effect of IVF intervention on vaginal microbiota was still existed, which may be the reason for the difference in microbiota between the two groups. The vaginal microbiota difference between the two groups gradually decreased with the increase of gestation days. At T3 of gestation, beta-diversity analysis revealed that the two groups clustered in the same area, thus suggesting a high degree of similarity. These results indicate that the influence of IVF on the VM occurred mainly during T1 of gestation and then gradually decreased as gestation progressed.

The differences in VM between IVF and control groups were greatest in T1. We therefore analyzed the microbiota in greater detail during this period. The IVF group showed a higher abundance of Alloscardovia, Sporosarcina, unidentified_Enterobacteriaceae, Pseudomonas and Vibrio was compared to controls, and a lower abundance of unidentified_clostridiales, Elstera and Sulfurospirillum. Alloscardovia is a recently reported microorganism with unknown pathogenic implications. Alloscardovia omnicolens is present in premature rupture of membranes (PPROM) and positively related to PPROM [15]. Pseudomonas belonging to the phylum $\gamma$-Proteobacteria are a well-known opportunistic pathogen in the human urogenital system. Previous studies indicated that Pseudomonas aeruginosa is involved in development of cervical lesions [16]. Several Vibrio spp (Vibrio cholera, Vibrio parahaemolyticus and Vibrio alginolyticus) can cause infectious diseases in humans [17]. The IVF group showed a higher abundance of these opportunistic pathogens than controls, indicating the VM was dysbiotic due to the use of drugs and frequent vaginal flushing. Commensal Clostridiales strains have been reported to modulate immune functions, including the immune response and immune homeostasis [18]. Studies on the co-existence of VM from pregnant women and intestinal bacteria from their babies have shown that Clostridium may be the strongest bacterium for vertical transfer from mother to child [19]. Sulfurospirillum were reported to be hydrogen-producing bacteria that were important for the environment, as hydrogen is the main electron donor for prokaryotes in anoxic ecosystems [20]. The IVF group in the present study had lower abundance of protective commensal bacteria than the controls, indicating the VM was dysbiotic during T1 in the IVF group.

Previous studies have showed that pretem birth is more common in IVF pregnancies as compared to naturally coinceived pregnancies [21, 22]. Furthermore, VM dysbiosis was associated with preterm birth and pregnancy loss [23, 24]. In the current study, three spontaneous preterm births occurred in the IVF group. Comparison of structure differences between women who underwent preterm birth or term delivery found that the greatest VM differences occurred during T1. This included higher species complexity in the vaginal ecosystem of preterm birth women, a greater abundance of pathogenic bacteria (Escherichia_Shigella and Alloscardovia), and lower abundance of commensal bacteria (Prevotella and Lactobacillus). We also found higher abundance of Lactobacillus_iners and lower abundance of Lactobacillus_jensenii, Lactobacillus_gasseri and Lactobacillus_equicursoris in VM of preterm women compared to term delivery women. The presence of a Lactobacillus_iners-dominated VM during gestation has been linked to abortion, spontaneous preterm delivery with intact membranes, and premature membrane rupture [25]. Lactobacillus_iners has therefore been suggested as a marker of vaginal cleanliness and leukocyte esterase [26]. In the present study we observed an increase in Lactobacillus_iners-dominated VM and pathogenic microorganisms, which may be a cause of spontaneous premature delivery. Preterm parturition is a syndrome induced by various factors. The prediction and prevention of preterm birth is a global challenge. Our data suggested that specific microbial taxa may be useful in defining the risk of preterm birth. Future studies may use probiotics or prebiotic to manipulate vaginal microbiota and prevent preterm birth.

The main limitation of this study is the small sample size. The small number of patients conceived through IVF due to male infertility together with the exclusion of patients with pregnancy complications further reduced the sample size. Thus, it should be noted that our conclusion should be seen in the context of this limitation. To our knowledge, this is the first longitudinal retrospective study to assess variability of vaginal microbiota in woman who undergoing IVF. For future studies, it will be desirable to increase the cohort size and to consider baseline samples prior to IVF or pregnancy in order to validate our findings.

In summary, the dynamic alterations in VM of IVF women during pregnancy were found to be similar to those of normal pregnant women. Furthermore, the diversity of VM was found to decrease during the gestation period. Lactobacillus was the predominant vaginal bacteria observed in all pregnant women in the present study. The major differences in VM between IVF and control groups were observed to occur during T1 of gestation, where IVF women had more abundant potential pathogenic bacteria and less abundant commensal bacteria. More abundant Lactobacillus_iners, Escherichia_coli and Alloscardovia_omnicolens was also linked to preterm birth in women undergoing IVF. Further investigations into the causal relationships between VM and pregnancy outcomes are required, in addition to studies of the underlying mechanism.

We uploaded the raw 16S rRNA gene sequencing data to the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA728871, accession no. PRJNA728871).

ZW contributed to conception and design of the study. YT performed the experiments and analyzed the data. QS and XS contributed to acquisition of data and revise the manuscript. YT and ZW wrote the manuscript. All authors read and agreed to the final version and are accountable for what is published.

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Dalian Women’s and Children’s Medical Center Group (approval number 20160021). All participants gave written informed consent prior to entering the study.

The authors extend their sincere gratitude to all medical workers for their enthusiastic help to assist in samples collection in our hospital.

This work was supported by National Key R&D Program of China (No. 2022YFC2704504), and Dalian Science and Technology Innovation Fund (No. 2019J13SN83).

The authors declare no conflict of interest.