Associations between microbial presence in follicular fluid with IVF outcomes: a systematic review and meta-analysis

Abstract

Purpose

The aim of the study was to synthesize disparate studies to investigate potential impact of microbial presence in FF of infertile women on IVF outcomes.

Methods

Following preliminary searches to find medical subject heading (MeSH) terms plus free terms, a systematic search was performed in the PubMed, Cochrane Library, Embase, Web of Science, and Clinicaltrials.gov databases from January 10, 2022, to July 5, 2023. Data collected for each study were analyzed using RevMan 5.4 software available on the Cochrane website.

Results

After correcting for contamination from the vagina, the FFs of 289 women were detected positively by microbial culture and identification, ELISA, and IPA. The pregnancy rate of the FF-positive group was significantly lower than the FF-negative group (19.7% vs. 32.2%) and (OR: 0.57, 95% CI: 0.28–1.14, P=0.11; I²=56%) while the fertilization rate was almost equal (60.0% vs. 62.0%) and (OR: 1.03, 95% CI: 0.88–1.20, P=0.72; I²=0%). Evidence quality was very low.

Conclusions

The different species of microorganisms in FF of infertile women may have different effects on IVF outcomes. The Lactobacillus spp. may have a positive effect, while other microorganisms may have the opposite effect.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-023-02912-x.

Introduction

Microbiota living on and inside the human body has attracted considerable attention in the medical field over the past decades [1, 2]. Especially with the completion of the human microbiome project (HMP), momentum has been gained in this area of science [3]. Studies have revealed that the microbiota of the female genital tract (FGT) accounts for 9% of the total bacteria in the body [4]. Currently, there is growing evidence that the microbiota of FGT, especially in the lower genital tract (LGT), differs in infertile patients compared with healthy and fertile women across different populations [5–8]. Simultaneously, the difference between them can have a certain impact on reproductive outcomes. For example, a recent review [9] showed that bacterial vaginosis (BV) was associated with some adverse reproductive outcomes, such as miscarriage, preterm birth (PTB), preterm prelabor rupture of membranes (PPROM), and infectious diseases. A meta-analysis study [10] also showed that BV had a significant association with early spontaneous abortion. Some studies have demonstrated that Chlamydia trachomatis (Ct) infection was associated with a higher risk of several pregnancy- and fertility-related adverse outcomes [11, 12], including miscarriage, PTB, stillbirth [13–15], PPROM, low-birthweight babies, and babies small for gestational age [16]. In addition, another review [17] of genital mycoplasmas also proved that there were significant differences in M. genitalium with PTB, in M. hominis with spontaneous abortion (SA), in stillbirth and premature rupture of membranes, and in the coinfections of M. hominis and ureaplasma with SA and stillbirth. Although most of these data were obtained from studies of microbiota in LGT, currently, an increasing number of researchers have paid attention to microbiota in the female upper genital tract (UGT). They were concerned and speculated whether the presence of microorganisms in UGT could also affect IVF treatment outcomes in infertile women. This prompted us to investigate the literature to obtain more evidence.

Human follicular fluid (FF) is a low-coagulation, semiviscous liquid composed of a variety of biologically active molecules. It surrounds oocytes during folliculogenesis in vivo and is the microenvironment during oocyte development and maturation [18–20]. Moreover, multiple studies have shown that FF collected during vaginal egg retrieval was not sterile and could isolate a variety of microorganisms [18, 21–24]. Some researchers have suggested that the source of these microbes may be “contaminated by vaginal and cervical strains during the aspiration of vaginal oocytes”, but a growing body of research suggests that FF is colonized by bacteria that may be part of the body’s natural microbial community. For example, they came from the vagina (Lactobacillus spp.), gastrointestinal tract (Bifidobacterium spp., Enteric Bacteria, S. agalactiae), Skin (Staphylococcus spp.) and oral malt (Streptococcus spp.) [22]. Among them, C. trachomatis (Ct) infection in the UGT is related to infertility caused by tubal occlusion. The nucleic acid test method for C. trachomatis is the most common detection method. However, some studies [25–28] have shown that the results are negative after antibiotic treatment, but anti-Ct antibodies still exist in serum and FF. Importantly, they also found that the presence of antibodies in FF seemed to be associated with adverse IVF results. So according to the existing studies, the microbial detection methods in FF are divided into microbial culture and identification [22–24, 29], the immunoperoxidase assay (IPA) and enzyme-linked immunosorbent assay (ELISA), which are the antigen-antibody detection method for Ct [25–28, 30].

The findings on the relationship between microbial presence in FF and IVF outcomes have not always been consistent after investigation. Some studies have suggested that the presence of microorganisms in FF could lead to a reduction in the pregnancy rates of infertile women [22, 23, 25, 28]. The outcome could be attributed to damaging oocyte and embryo quality [22, 23, 28] and decreasing embryo transfer rates [22, 27]. Finally, it affects the live birth rate [22]. However, other studies have argued that these microorganisms were not associated with pregnancy rates [24, 29], which was attributed to the fertilization rates being statistically insignificant [24, 27, 30]. Controversy among these studies makes the relationship between microbial presence in FF and IVF outcomes very equivocal. Simultaneously, no studies have systematically analyzed these results. Therefore, we intend to conduct a systematic review and meta-analysis of the included studies according to Pelzer et al.’s definition of FF “colonizers” and “contaminants” to comprehensively explore the potential relationship between FF “colonized” microorganisms and IVF outcomes.

Methods

The MOOSE Checklist for Meta-analyses of Observational Studies has been followed meticulously [31]. The study was exempt from institutional review board approval as a systematic review and meta-analysis of previously published information. The objectives of the present systematic review were prespecified in a review protocol and registered in advance in PROSPERO, registration number: CRD42022340278.

Literature search strategy

The PubMed, Cochrane Library, Embase, Web of Science, and Clinicaltrials.gov databases were systematically searched without time of publication and ethnicity restrictions in the Microsoft Edge browser (version 103.0.1264.71, 64-bit). In addition, medical subject heading (MeSH) terms plus free terms were used to search. The search strategy was “(follicular fluid) AND (fungi)”, “(follicular fluid) AND (bacteria)”, “(follicular fluid) AND (microbe)”, “(follicular fluid) AND (microorganism)”, “(follicular fluid) AND (microbiota)”, “(follicular fluid) AND (microbiome)”, “(follicular fluid) AND (microbiology)”. No language restrictions were applied too. Translation software (Google Translate, Mountain View, CA, USA) was used to determine eligibility for studies published in languages other than English. In addition, any method was attempted to search for unpublished studies. The additional eligible studies were identified by the manual search of reference lists from relevant original and review articles. This study began searching on January 10, 2022, and the last search was on July 5, 2023.

Eligibility criteria and study selection

An eligible study that was included in the meta-analysis met the following criteria: (1) subjects: infertile women undergoing IVF treatment; (2) research types: randomized clinical trials, case–control studies, or cohort trials; (3) exposure: the microbiota of FF; (4) primary outcomes: clinical pregnancy rate and fertilization rate; (5) secondary outcomes: immature oocytes (GV), mature oocytes (MII), high fertilization rate (>50%), grade I embryos, embryo discard rate, embryo transfer rate, and live birth rate.

The exclusion criteria were as follows: (1) the research subjects were animals and the same study populations; (2) the exposure involved viruses and parasites; (3) the research was a review, case report; (4) there were insufficient raw data and no related outcomes reported; and (5) the full text could not be obtained.

Data extraction

Two authors (OSS and LM) implemented a database search, data extraction and study quality assessment separately. If disagreements occurred, they were resolved through discussion or consultation with a third party. The title of the article was first read during literature screening, and the abstract and full text were further read to determine whether to include them or not after excluding obviously irrelevant literature. If necessary, the authors of the original studies were contacted by email or telephone for undetermined but important information. For each included study, the following information was extracted: year of publication, first author’s name, study design, country, study population, age, detection method, sample size of the positive group, number of control groups, outcome, effect estimate, and main finding.

Quality of evidence

To assess the quality of nonrandomized/cohort studies, the Newcastle–Ottawa Scale (NOS) study was used for each article. The NOS evaluation scale consists of 3 blocks (population selection, comparability, exposure/outcome evaluation) and 8 items, including selection with a total score of 4, comparability with a total score of 2, and exposure (case–control studies)/outcome (cohort studies) with a total score of 3. The high-quality research can earn up to 9 points [17]. Furthermore, we applied GRADE (GRADEpro GDT software [McMaster University, Hamilton, Ontario, Canada]) for data quality analysis of clinical pregnancy rate and fertilization rate.

Data synthesis

Meta-analysis was performed using RevMan 5.4 software available from the Cochrane website. For dichotomous data, the odds ratio (OR) was used as an effect analysis statistic. The point estimation and 95% confidence interval (CI) of each effect quantity were provided. P < 0.05 indicated that the difference was statistically significant. Meanwhile, the heterogeneity between the included research results was quantitatively judged using I². When I²≤50%, it indicated no significant heterogeneity among data, and a fixed effect mode was adopted; when I²>50%, it indicated heterogeneity among data, and a random effect mode could be used. Sensitivity analyses were performed by sequentially excluding any of the included studies to clarify the effect of a study on the overall results [32]. A funnel plot was also constructed and used to assess publication bias [33].

Results

Study selection

The details of the database search are shown in Fig. 1. A total of 1726 records were identified through the preliminary screening of the database. The database yielded 679 studies after excluding duplicate articles. Next, 606 studies were further excluded because they were not related to the purpose of the meta-analysis based on titles and abstracts. Then, for the remaining 73 studies assessed by full-text reading, a total of 64 studies were excluded for the following reasons: duplicate study population; no full text; ongoing experiments; insufficient raw data; no related outcomes reported; duplicate studies; case reports and reviews; exposure to the virus and animal studies (Table S1). In total, 9 studies were ultimately included in the present meta-analysis.

Fig. 1.

PRISMA flow diagram of study selection

Study characteristics

The characteristics of the included studies are shown in Table 1. Overall, 9 studies, i.e., 3 case–control studies [22, 27, 30] and 6 cross-sectional studies [23–26, 28, 29], with 898 infertile women, were analyzed in the meta-analysis. The studies were published between 1990 and 2021. The three methods of microbial culture and identification, ELISA, and IPA were used for the determination of FF microorganisms. Among them, the FFs of 289 women were determined to be microbial “colonizers” or to contain anti-Ct antibodies. Simultaneously, the scores of quality assessment for all eligible studies ranged from 7 to 9, indicating high quality (Table S2 to S3).

Table 1.

Characteristics of the included studies

References	Country	Study population	Age	Study design	Diagnostic criteria	Main cause of infertility	Main findings	Outcomec	Whether to use antibiotics
Lunenfeld (1990) [30]	Israel	55 women undergoing IVF	32 (23–34)	CCa	A positive serum reaction	Study group: unknown. Comparison group: 298 women from community.	No statistical association was found between the presence of Ct IgG and IgA antibodies in FF to oocyte fertilization.	①	No
Cottell (1996) [29]	Ireland	28 couples undergoing IVF-ET	Unknown	CSa	Colony size	Unexplained infertility: 12 (42.9%); male factor: 5 (17.9%)	Fertilization, cleavage, and pregnancy rates were independent of microbial presence.	①②	Yes
Neuer (1997) [25]	Germany	149 women undergoing a cycle of IVF	Unknown	CSa	Instructions of manufacturer	Tubal factor: 64(43%); male factor: 41(27.4%)	IgA antibodies to rLPS of Ct were associated with a failure to become pregnant after embryo transfer.	①	Yes
Cortiñas (2004) [26]	Venezuela	41 women undergoing IVF	34 (±5)	CSa	Positive samples of manufacturer	Tubal factor: 15(37%); male factor: 10(24%)	A high prevalence of Ct infection was found among the infertile population.	①	No
Jakus (2007) [27]	USA	253 women who underwent 323 cycles of IVF	21–43	CCa	OD above the equivalence range	The positive or negative groups of anti-HSP60 antibodies: male factor: 100(40%); tubal factor: 58 (25%),	Anti-HSP60 antibody detection of Ct was unrelated to the number of oocytes collected or the percentage of oocytes fertilized.	⑤	No
Pacchiarotti (2009) [28]	USA	235 women undergoing IVF	Unknown	CSa	Positive samples of manufacturer	Unknown	Pregnancy and implantation rates were significantly reduced in the presence of IgA anti-chlamydia antibodies.	①③④⑥⑧	Yes
Pelzer (2013) [22]	Australia	263 couples for fully stimulated IVF cycles	37 (±4)	CCa	‘Colonizers’b	Infertile group: Idiopathic infertility:66 (32.7%); endometriosis: 49 (24.3%) and polycystic ovary syndrome: 48 (23.8%). The ‘fertile’ group: male factor: 60 (23%)	Decreased pregnancy, embryo transfer, live birth rates and quality embryos.	①②⑦⑧⑨	No
Kim (2018) [23]	Korea	49 infertile female undergoing IVF/ ICSI cycles	Unknown	CSa	‘Colonizers’b	Unknown	Follicular fluid is not sterile but follicular fluid micro-organisms have no significant negative impacts on clinical IVF outcomes.	①②④⑥	Unknown
Usman (2021) [24]	USA	90 women for IVF-ET	35(±3.5)	CSa	‘Colonizers’b	Tubal factor: 27 (31.4%); ovarian factor: 21 (24.4%)	There was no statistically significant difference in the fertilization and pregnancy rates.	①⑤	No

^aCC Case-control study, CS cohort study. ^b“colonizers” if microorganisms were detected within the FF but not within the vaginal swab (at the time of oocyte retrieval), or “contaminants” if microorganisms detected in the vagina at the time of oocyte retrieval were also detected within the FF. ^cOutcomes: ①pregnancy rate; ②fertilization rate; ③immature oocytes (GV); ④mature oocytes (MII); ⑤high fertilization rate (>50%); ⑥grade I embryos; ⑦embryo discard rate; ⑧embryo transfer rate; ⑨live birth rate

According to the positive definition of FFs in the original studies, they could be divided into the following two. The first was the culture and identification of microorganisms. (1) The definition of “colonizers” [22–24]. Microorganisms isolated from FFs were classified as “colonizers” if microorganisms were detected within the FF but not within the vaginal swab (at the time of oocyte retrieval) or “contaminants” if microorganisms detected in the vagina at the time of oocyte retrieval were also detected within the FF. (2) Colony size [29]. Aerobic and anaerobic cultures were considered as positive when they contained 88<³ organisms/mL, or the samples of M. hominins and Ureaplasma urealyticum were reported as positive when iti¹ colony–forming units/mL were observed. The second was to detect the presence of anti-Ct antibodies in FF by ELISA and IPA. The following were detected by ELISA assays: (1) IgA antibodies, samples were evaluated according to the provided positive samples [26, 28] or instructions (rLPS IgA antibody titer was 1:200) [25] of the manufacturer; (2) anti-HSP60 antibodies, the optical density value (OD) that was at least 2 SD above the mean value of known negative samples [25], or OD was at least twice the value of the OD of the corresponding nonantigen well and higher than cutoff value (average OD of seven negative samples +2 SD) [26], or OD was higher than the equivalence range (the mean OD value of the negative control +0.350), which is ±10% [27]. The judgment standard of the IPA test [30] was that a positive serum reaction was indicated by a deep-blue precipitate in the cytoplasm of the infected cells, which could be seen with an ordinary light microscope. Then, in this article, our positive group included the women whose FFs were “colonized” and only considered positive in the original studies.

Different bacterial species were detected in colonized FFs according to the 4 studies using microbial culture and identification [22–24, 29]. Lactobacillus spp., Staphylococcus spp., and Streptococcus spp. were all isolated in the 3 studies [22, 24, 29]. Enterococcus was isolated in 2 studies [22, 24]. Peptostreptococcus spp. and Escherichia coli were isolated in 2 studies [22, 23]. M. hominis, U. urealyticum, Gardnerella vaginalis, and diphtheroids were only isolated in the study [29]. Kocuria kristinae was only isolated in the study [23]. Actinomyces spp., Propionibacterium spp., and Bifidobacterium spp. were isolated in the study [22]. Candida albicans was also only isolated in the study [24]. In addition, in the study [22], there were also subtle differences in the most prevalent microbial species isolated in different FFs of the detected infertile populations. Lactobacillus spp. was the most common species isolated from the FFs of all infertile women. Bifidobacterium spp. was isolated in the infertile women with male factor, idiopathic and genital tract infections. Actinomyces spp. was isolated in the infertile women with male factors and genital tract infections. Staphylococcus spp. was isolated in the infertile women with idiopathic and genital tract infections. Propionibacterium spp. was only isolated in the infertile women with idiopathic disease.

Clinical pregnancy rate was reported in 8 studies. Four used the method of microbial culture and identification [22–24, 29]. Of the 4 studies, 3 defined pregnancy as follows: (1) the presence of a gestational sac at 5 weeks after ET (29); (2) 3/5 days after ET, details not available [23]; (3) the World Health Organization definition, which was ultrasonographic evidence of a gestational sac containing one or more fetuses with a heartbeat [22]; (4) the presence of a gestational sac at 4 weeks after ET [24]. Among the remaining studies, 3 studies used ELISA text [25, 26, 28], and 1 study used IPA text [30]. They also did not propose a definition of pregnancy. In addition, the 2 studies reporting fertilization rates were all detected by microbial culture and identification. One of them defined how the fertilization rate was calculated: fertilized oocytes/total oocytes collected for all women [22]. Another article defined normal fertilization: 8 to 20 h after insemination, and oocytes were examined for the presence of two pronuclei [29].

Positive rate

The positive rates reported in the included studies ranged from 8.2 to 54.1%, and great heterogeneity was observed. The reasons for the heterogeneity could be explained by the difference in FF assay methods, design types, and the characteristics of the study samples.

Three of the 4 studies that used microbial culture and identification were cohort studies. The mean age of the study samples in one study was 35 (±3.5) years [24], while it was unknown in two other studies [23, 29]. In the study samples, the main causes of infertility were respectively unexplained infertility (42.9%) [29] and tubal infertility (31.4%) [24]. The secondary causes of infertility were respectively male infertility was secondary (17.9%) [29], and ovarian infertility was secondary (24.4%) [24]. But one was unknown [23]. The merge positive rate of 3 was 19%. Moreover, the mean age of the study samples of the case-control study [22] was 37 (±4) years. The study samples were dominated by idiopathic infertility (25%) and male factor infertility (23%), followed by endometriosis (18.7%), polycystic ovary syndrome (18.3%), and genital infection (14.9%). Its positive rate of the study was 28.6%. Finally, the positive rate of the cohort study was 9.6% lower than that of the case–control study.

Three of 4 studies using ELISA assays were cohort studies. The mean age of the study samples in one study was 34 (±5) years [26], while it was unknown in two other studies [25, 28]. Meanwhile, it was dominated by tubal infertility (37% and 43%), and the secondary was male factor infertility (24% and 48%) [25, 26]. But both were unknown in another study [28]. The merge positive rate of 3 was 40.9%. Then, one of 4 studies was a case–control study [27]. The age range of the study samples was 21–43 years. The infertile composition was that male factor infertility was the primary (40%), and tubal infertility was secondary (25%). Its positive rate was 54.2%. In the ELISA test group, the positive rate of cohort studies was 13.2% lower than that of case-control studies. In addition, the only study using the IPA text was a case-control study [30]. The mean age of the study samples was 32 years (23–34 years). The main cause of infertility was mechanical infertility (69%), and unexplained infertility was secondary (25%). Then, the positive rate for these women was 21.8%. Finally, the positive rate of cohort studies was 7.5% lower than that of case-control studies.

Although most studies did not report ethnicity, a total of seven studies had study samples from Western European countries (Ireland, Australia, the USA, Germany, and Venezuela) [22, 24–29], and 2 studies were from a Middle Eastern country (Korea, Israel) [23, 30].

Clinical pregnancy rate

In the microbial culture and identification group, the FF-positive group had a lower clinical pregnancy rate, albeit non-significantly, than those with the FF-negative group (22.5% vs. 36.6%) and (OR: 0.56, 95% CI: 0.24–1.32, P=0.18; I²=34%; Fig. 2). The pregnancy rate of positive FF was significantly lower than that of the negative FF in the ELISA test group (18.8% vs. 30.2%) and (OR: 0.41, 95% CI: 0.21–0.80, P=0.009; I²=26%; Fig. 2). However, evidence based on only 1 study using IPA tests [30] showed that the pregnancy rate in the FF-positive group was higher than that in the FF-negative group (33.3% vs. 0%) and (OR: 42.88, 95% CI: 2.11–873.34, P=0.01; Fig. 2).

Fig. 2.

Meta-analysis results of clinical pregnancy rate

Finally, the pooled result of 3 groups with a random-effects model showed that the overall pregnancy rate in the FF-positive group was lower than that in the FF-negative group (19.7% vs. 32.2%) and (OR: 0.57, 95% CI: 0.28–1.14, P=0.11; I²=56%; Fig. 2). Furthermore, a sensitivity analysis found that the difference among them was statistically significant, and the heterogeneity was reduced (OR: 0.46, 95% CI: 0.29–0.73, P=0.001; I²=19%) after excluding the studies Lunenfeld et al. [30] (OR: 42.88, 95% CI: 2.11–873.34). The heterogeneity of this study may be related to the publication time, testing method, definition of a positive diagnosis, race, and small sample size. Using the GRADE approach, the quality of evidence for the clinical pregnancy rate was very low (Table S4).

Fertilization rate

Meta-analysis using a fixed-effects model found that the pooled fertilization rate of the FF-positive group was slightly lower than the FF-negative group, which was detected by microbial culture and identification (60.0% vs. 62.0%) and (OR: 1.03, 95% CI: 0.88–1.20, P=0.72; I²=0%; Fig. 3). In the other 2 studies of the microbial culture and identification group, whose data could not be extracted, the average fertilization rate of one had a similar result (70.0% vs. 84.4%) [23], while the other had the opposite (81.0% vs. 64%) [24]. But both were not statistically significant. Using the GRADE approach, the quality of evidence for fertilization rate was very low (Table S4).

Fig. 3.

Meta-analysis results of fertilization rate

Other reproductive outcomes

In the microbial culture and identification group, Kim et al. [23] showed the mature oocyte rate and I embryo rate of the FF-positive group higher than the FF-negative group (50.0% vs. 67.4% and 0% vs. 31.3%). Pelzer et al. [22] also studied the outcomes of the embryo discard rate (43.3% vs. 38.1%), embryo transfer rate (53.3% vs. 77.0%) and live birth rate (18.0% vs. 27.1%). Among infertile patients with endometriosis and polycystic ovary syndrome, the embryo transfer rate in the FF-positive group (39.0% and 36.0%) was significantly lower than that in the FF-negative group (94.0% and 79.0%) and the difference was statistically significant (P<0.0001 and P=0.006). Meanwhile, among infertile patients with endometriosis, the FF-positive group had a significantly higher embryo discard rate than the FF-negative group (63.0% vs. 34.0%), which was also statistically significant (P<0.0001). In addition, according to the study by Rosen et al. [33], the concept of a high fertilization rate was a fertilization rate > 50%, and they aimed to study the proportion of women with a high fertilization rate in the population. The study [24] showed that a high fertilization rate was seen in 80.0% of women with positive FF and 87.0% of women with negative FF.

In the ELISA test group, Neuer et al. [25] showed that a high fertilization rate was seen in 60.0% of women with positive FF and 58.4% of women with negative FF. Simultaneously, Pacchiarotti et al. [28] reported the outcomes of immature oocyte rate (34.4% vs. 7.5%), mature oocyte rate (35.1% vs. 64.5%), grade I embryo rate (28.2% vs. 47.3%), and embryo transfer rate (13.3% vs. 27.0%), which were all statistically significant (all P<0.05).

Publication bias

Funnel plots representing the meta-analysis of clinical pregnancy rate are shown in Figure S1. The plots were symmetrical based on visual inspection, suggesting a low risk of publication bias. The publication biases underlying the meta-analysis of other reproductive outcomes were unable to be determined since only 1 or 2 studies were included for these two outcomes.

Discussion

Main findings

In the systematic review and meta-analysis, a total of 9 studies, including 898 infertile women, were analyzed. To reduce bias, we have corrected contamination from the vagina. Finally, the FFs of 289 women were positive by nonspecific flora detection, including microbial culture and identification, and specific flora detection, including ELISA tests and IPA tests, which were for Ct. However, there was a large heterogeneity among the results attributed to but not limited to the different methods, the positive definitions of FF, sample size, and study types. Then, we found that microbial species in FF varied among studies and among types of infertility patients in the nonspecific flora detection group. The available data suggested that the pregnancy rate of the FF-positive group was significantly lower than the FF-negative group, while the fertilization rate was almost equal. But the results of GRADE were very low. Furthermore, we also found that the quality evaluation of oocytes and embryos in the FF-positive group was negative.

Strengths and limitations

The review was the first to systematically explore the relationship between the presence of microorganisms in the FF of infertile women and IVF outcomes, which has not been done before. At the same time, we corrected contamination from the vagina, which reduced the bias. As homogeneity analysis showed significant heterogeneity due to different test methods, we performed sub-group analysis. The method of meta-analysis was also used to study their relationship to reveal the influencing factors of female infertility more accurately and find new ways to treat infertility. However, the current evidence also has some limitations. First, unpublished findings were not included in this study due to language and search constraints that may have resulted in a lack of necessary data. Next, different study types, methods and positive definitions of FF, the number of studies and some old studies (1990/1996), and the size and geographical distribution of the study sample could result in an overall weak quality of evidence and increase uncertainty in estimates. Furthermore, no definitive conclusions could be drawn due to limited evidence or large heterogeneity between studies.

Interpretation

In the review, there was still significant heterogeneity in the detection methods. Traditionally, microbiome studies were conducted with culture-based methods that were used to identify bacterial species and optical magnification techniques that were used to identify bacteria based on phenotype or morphological details [34, 35]. It was also used in the nonspecific flora identification group. However, some studies have demonstrated that the culture-based methods were still informative, but they only detected a small proportion of organisms that were not representative of the ecological niche under investigation or even failed to identify potentially significant microbial organisms with regard to health and disease [36–38]. This may explain why the 3 studies using culture-based methods detected different microorganisms. In addition, the 3 recent studies using 16S rDNA sequencing [39–41] suggested that the FF was not sterile and the alteration of the microbiota or the presence of certain microbes in FF may have a negative impact on IVF therapy. It is a pity that they did not correct the contamination from the vagina. Therefore, we should use more accurate and appropriate detection methods to obtain more relevant information, such as high-throughput DNA sequencing, 16S rRNA or whole genome sequences, nucleic acid amplification tests (NAATs), and multiplex quantitative real-time PCR (qPCR).

Apart from the detected methods, it is also worth noting the different definitions of positive FF in different studies. For example, diphtheroids are considered “colonizers” in one study [29], while “contaminants” in another study [24]. These studies included were published over a relatively wide span of years, which may lead to inconsistencies in the microbiological diagnostic criteria. In addition, there were 3 included studies [25, 28, 29] using antibiotics, which also questioned the homogeneity of the exposure. A current review of intestinal microbiota [42] showed the specific changes that occur over time in the microbial composition during antibiotic-mediated dysbiosis and recovery. Thus, it is not known whether antibiotics will also have the same effect on the microorganisms in FF. In addition, some studies are very old (1990/1996), which was one of the reasons for heterogeneity. Because IVF had probably also different procedures and another pregnancy success rate at the time.

The different species of microorganisms in FF may have different effects on IVF outcomes. Some animal studies have shown that some bacteria, including E. coli and Streptococcus spp. in porcine FF, might inhibit follicle-stimulating hormone (FSH) from binding to its receptor on granulosa cells [43–45]. Simultaneously, Pelzer et al. [22] also showed that some opportunistic pathogens (including Propionibacterium spp., Streptococcus spp., Actinomyces spp., Staphylococcus spp., and Bifidobacterium spp.), which were detected similarly in another study [46], were related to adverse IVF outcomes, including decreased embryo transfer rates, which may result in poor-quality embryos due to damage by microorganisms or their metabolites and decreased pregnancy rates or no pregnancy (failed implantation), while Lactobacillus spp. within FF correlated with positive IVF outcomes, including an increase in embryo transfer rates. Some researchers believe that the positive effect of Lactobacillus spp. was attributed to the presence of hydrogen peroxide and lactic acid, which could be inhibitory substances to other microbial species [22, 47–49]. Some studies on the vaginal microbiota [9, 50] support this conclusion. Simultaneously, it was also believed that Lactobacillus spp. could reduce the expression of inflammatory factors and increase the expression of anti-inflammatory factors, thereby reducing vaginal inflammation [51, 52]. Perhaps they also affect FF through a similar mechanism. Ct, an intracellular bacterium, is one of the most frequent causes of genital tract infections and can lead to infertility by chronic infection with pelvic inflammatory disease and tubal lesions. Our review showed that the presence of anti-Ct antibodies in FF may affect the outcome of pregnancy in infertile women (OR: 0.41, 95% CI: 0.21–0.80, P=0.009; I²=26%; Fig. 2). It may be explained by several mechanisms, including ascending infection to the upper reproductive tract, chlamydia heat shock protein (cHSP60)–induced delayed hypersensitivity, or proinflammatory responses in the epithelium to either specific antigens or bacteria [16]. It is necessary to expand the sample size to further clarify the result.

The types of studies may be an important factor that affected heterogeneity. So we conducted a meta-analysis with a random-effect model based on it for clinical pregnancy rate. The new pooled result was the same as the original result. The clinical pregnancy rate of the FF-positive group was significantly lower than the FF-negative group in the cohort group and case-control group. The former had statistical significance (18.8% vs. 30.2%) and (OR: 0.52, 95% CI: 0.30–0.90, P=0.02; I²=18%; Figure S2) while the latter did not (23.1% vs. 36.4%) and (OR: 2.87, 95% CI: 0.02–440.04, P=0.68; I²=90%; Figure S2). The reason for the difference in results between the two groups may be due to the sample size (the former > 400, the latter < 400) and the study [30] with extremely low reliability was included in the case-control group. It is worth noting that the study [46] has shown that the women who had colonized FF tend to get an abortion in their first trimester more than women whose FF was categorized as contaminated. Future research can consider increasing the outcome. Interestingly, the results of other reproductive outcomes suggested that the presence of microorganisms and their metabolites in our FF may interfere with follicle growth, oocyte development, and maturation. This would result in poor embryo quality and may even trigger an immune response that can lead to inflammation of the reproductive tract or infection of the developing fetus, with significant consequences for both the fetus and the pregnant woman [22, 28, 53]. A study [54] on the bacterial influence on oocyte quality revealed that a higher total bacteria mass in follicles from oocytes that failed fertilization while Lactobacillus was not present. It requires more studies to explore their potential relationship.

Conclusion

Our review suggested that the different species of microorganisms in FF of infertile women may have different effects on IVF outcomes. The Lactobacillus spp. may have a positive effect, while other microorganisms, such as Staphylococcus spp., Streptococcus spp., and Actinomyces spp., may have the opposite effect. Furthermore, the available data suggested that the pregnancy rate of the FF-positive group was significantly lower than the FF-negative group, while the fertilization rate was almost equal after correcting for contamination from the vagina. But the results of GRADE were very low. Future research should continue to focus on their potential mechanisms and explore new ways to improve diagnosis and treatment strategies.

Author contribution

Shanshan Ou and Bo Liu contributed to the conception, planning, and carrying out the review. Shanshan Ou and Lanyu Cui performed the analyses and drafted the manuscript. Ming Liao and Yuehui Du assisted in the study selection and the updated search on September 30, 2022. Ming Liao, Ling Zhao, and Chuyu Peng were responsible for the revision of the article structure and content and the addition of reproductive knowledge. Li Jiang acted as the senior/consulting author. All authors critically reviewed the manuscript and approved the final version.

Funding

This work were supported by the Natural Science Foundation of Guangxi Province, grant numbers 2022GXNSFAA035504, and the Guangxi Medical University Training Program for Distinguished Young Scholars.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethical approval

Ethical approval was not required for secondary use of data in this systematic review and meta-analysis.

Competing interests

The authors declare no competing interests.