Microbiome analysis in women with endometriosis: Does a microbiome exist in peritoneal fluid and ovarian cystic fluid?

Abstract Purpose To investigate the relationship between the microbiome of the female genital tract and endometriosis. Methods This prospective cohort study included 36 women who underwent laparoscopic surgery for ovarian tumor from July 2019 to April 2020. Of them, 18 had endometriosis, and 18 did not have endometriosis. Vaginal secretions, endometrial fluid, peritoneal fluid, and ovarian cystic fluid were collected during surgery. Next‐generation sequencing of bacterial 16S rRNA was performed to characterize the microbiome. Results Specific microbiomes were not detected in either peritoneal fluid or ovarian cystic fluid regardless of the presence or absence of endometriosis and the type of cyst. When the cutoff value of infectious bacterial abundance in the vagina was set as 64.3%, there were many cases more than a cutoff value in the endometriosis group significantly (p = 0.01). When the cutoff value of infectious bacterial abundance in the endometrium was set as 18.6%, there were many cases more than a cutoff level in the endometriosis cases significantly (p = 0.02). Conclusion Peritoneal fluid and ovarian cystic fluid are almost sterile, although dysbiosis may occur in the vaginal and endometrial microbiome in women with endometriosis.

considering the modern and advanced lifestyle of women at this age. 2 It is common for women with endometriosis to experience iatrogenic pelvic inflammatory disease and the development of a tubo-ovarian abscess following endometrium biopsy, hysterosalpingography, and oocyte retrieval. 3 These cases may follow iterative and intractable progression and often are difficult to treat.
Additionally, endometriosis is a cause of sterility because of implantation failure. 4 The association between endometriosis and infection has been assessed, and recently, some studies have reported the microbiome of the female genital tract. 5,6 According to these reports, the microbiome is affected by age, reproductive condition, ethnicity, and other factors, as well as by highly dynamic changes throughout life.
Some microorganisms may increase the risk of genital tract infection. 7 Therefore, we assumed that a microbiome exists in patients with endometriosis, which we could not detect using normal bacteriological culture methods, and that it is related to chronic inflammatory conditions with endometriosis.
Recently, a small number of bacteria were noted to be present in the peritoneal fluid (PF) and endometrium, which was thought to be sterile. 8,9 However, it is unclear how the microbiome of the PF and endometrium affects human health. Because of the difficulty in measuring small numbers of microorganisms, the abdominal cavity and uterus were thought to be sterile. Using 16S rRNA sequence analysis, we amplified the 16S rRNA domain of bacteria and analyzed the microbiome with a little sample more precisely by analysis of sequence arrangement using next-generation sequencing in large quantities. Thus, we identified the bacteria present and characterized the bacterial community without culturing. 10 In this study, we clarified the relationship between the microbiome of the female genital tract and endometriosis by confirming the existence of the microbiome in the vagina, endometrium, PF, and cystic fluid in women with endometriosis.

| Study design/patients/purpose
This was a prospective cohort study of 36 women with or without endometriosis who had ovarian tumors and underwent laparoscopic surgery at Ryukyu University Hospital and Mie University Hospital from July 2019 to April 2020. The inclusion criteria were patients aged ≥20 years who provided consent with a preoperative diagnosis of unilateral or bilateral benign ovarian tumors. Among them, 18 patients had endometriosis (Endo group), and 18 did not have endometriosis (Non-Endo group). We excluded postmenopausal patients, patients with uterine anomalies, and patients who used antibiotics.
Patients without endometriosis who were found to have endometriosis lesions in the abdominal cavity at surgery were also excluded.
The outcome was to evaluate differences in the microbiome of the PF, ovarian cystic fluid, endometrium, and vagina in patients with and without endometriosis.

| Sample collection
We collected vaginal secretions (VS), endometrial fluid (EF), PF, and ovarian cystic fluid (OF) of patients with endometriosis and without endometriosis during surgery. VS samples were collected by swab before vaginal sterilization. After collecting VS samples, we sterilized the vagina with povidone-iodine three times, washed it with saline, and collected EF using a brush for cell collection (ASKA Pharmaceutical, Tokyo, Japan). PF samples were collected by suction using a sterile procedure during laparoscopic surgery. In a case where none of the PF were recognized, we washed the abdominal cavity with 15-20 ml saline and collected the sample. We collected OF samples by puncturing the ovarian tumor directly in a bag aseptically after salpingo-oophorectomy or via the abdominal wall using a sterile procedure during cystectomy.

| Microbiome analysis
The hypervariable regions of the variable regions 1-2 (V1-2) of the bacterial 16S rRNA gene were amplified and analyzed using nextgeneration sequencing to identify the bacteria. to taxa using the database reported in a previous study. 11 Bacteria that were frequently observed in the negative control were grouped as background bacterial contamination (Supplemental Data 1), and after screening of the samples, the background-contaminated bacteria were excluded from the microbiome profile.

| Sample screening and clustering
Nonhierarchical clustering of microbiome profiles in VS, EF, PF, and OF samples and negative controls were conducted using weighted UniFrac distance. Samples clustered with negative controls were not used for subsequent analyses. Hierarchical clustering of VS and EF samples with microbiome profiles excluding backgroundcontaminated bacteria was conducted using Bray-Curtis distance matrix, and heatmaps were generated.
After rarefaction analysis, including Shannon index, Chao1 richness, and PD whole tree using microbiome profiles, alpha diversity indexes were compared between the Endo and Non-Endo groups in VS and EF samples. Beta diversity was analyzed using principal coordinate analysis. Multivariate analysis based on weighted UniFrac distance was conducted to compare differences in the microbiome between the Endo and Non-Endo groups in VS and EF samples. Beta diversity was analyzed using permutational multivariate analysis of variance (PERMANOVA) test.
The relationship of the rate of Lactic MB in VS and EF samples with endometriosis was analyzed using receiver operating characteristic (ROC) analysis, and the best criteria of the combination rate were investigated. ROC analysis was also conducted for Infect MB using the abovementioned method.

| Statistical analysis
All analyses were conducted using R software version 3.6.2. The normality and homoscedasticity of continuous data were analyzed using the Shapiro-Wilk test and Bartlett's test, respectively. When the data had normality and homoscedasticity, Student's t test was used. When the data had only normality, Welch's t test was used.
When the data did not have normality, Wilcoxon's rank-sum test was used. For discrete data, Fisher's exact test was used. A p-value of <0.05 was considered statistically significant.

| Background
The patients' backgrounds are shown in Table 2. The mean ages of the Endo and Non-Endo groups were 37.9 and 35.2 years, respectively, and no significant difference was noted (p = 0.29). Bilateral ovarian lesions (n = 10) were more common than unilateral lesions (n = 8) in the Endo group compared with that in the Non-Endo group, in which only one patient had bilateral ovarian lesions. Five cases were stage Ⅲ and 13 cases were stage IV based on the re-

| The microbiome of PF and OF
We found 121 and 94 species in PF and OF, respectively. However, almost all PF and OF samples were clustered, similar to negative controls, and the correlation coefficient between PF samples and the negative control was approximately "1.00." In addition, the number of sequence reads after filtering was extremely low. Therefore, we concluded that almost all PF and OF samples had no specific microbiome both in the Endo and Non-Endo groups. Thus, the PF and OF samples were excluded from subsequent analyses.
We could not detect any specific microbiome in PF and OF, though we detected very small numbers of bacteria, such as Paracoccus yeei, which were not present in the negative control. This implies that the possible existence of small quantities of distinctive bacteria cannot be denied.

| The microbiome of the vagina and endometrium
We  There were significantly few cases with Lactic MB more than cutoff levels and were significantly many cases with Infect MB more than cutoff levels at the vagina and endometrium in the Endo group.

| DISCUSS ION
In this study, PF and OF appear to be sterile regardless of the possible presence of endometriosis and the type of cyst, although we were able to show that dysbiosis may occur in women with endometriosis in the vaginal and endometrial microbiome.
The etiology of endometriosis is unclear, although various methods of onset have been hypothesized. One hypothesis, "implantation of endometrium," states that the endometrium refluxes into the peritoneal cavity with menstrual blood from the fallopian tubes. 23 According to this hypothesis, the risk factor of onset is said to be related to the presence of inflammation. The mechanism of onset of chronic pelvic inflammation is unknown, although some recent reports suggest a relationship with the microbiome. Khan et al. reported that the lower genital tract in humans is constantly exposed to microorganisms, which could infect the upper genital tract through direct migration. They suggested that bacterial infection after migration from the vagina to contaminating menstrual blood results in the accumulation of endotoxin in the PF and initiation of pelvic inflammation. 24 Additionally, they reported that the levels of cytokines or growth factors increase in the PF of patients with endometriosis, which may lead to the progression of endometriosis lesions. 24 Although it was thought that the abdominal cavity was sterile, a microbiome was found to be present in PF. 25 The microbiome of the endometriotic cyst was also supposed to be present, and two courses of migration, that is, the ascending course from the vagina and from the bowels via adhesion in the abdominal cavity, were assumed to take place. 26  there were many cases with a high abundance rate of bacteria relating to infection in the Endo group in both the vagina and endometrium. In our study, we analyzed infectious bacteria by grouping as Infect MB across the distinction between phyla. Thus, it is a new viewpoint which is not seen before. Because the outcome was to evaluate differences in the microbiome between cases with and without endometriosis, we could not use the existing cutoff values for pregnancy outcome. 9,28 Therefore, we thought it was necessary to establish a new cutoff value as an indicator to be applied in clinical practice and to determine the presence of endometriosis.
In this analysis, we tested multiple cutoff values and set the value with the lowest p-value and highest accuracy as the cutoff value (Supplemental Data 5). It is well understood that Lactobacillus produces lactic acid and hydrogen peroxide to prevent inflammation, and Bifidobacterium produces acetic acid and aggravates the barrier function of the mucous membrane. 29 We found a low abundance of bacteria working with such mechanisms in the Endo group with tendency in both the vagina and endometrium. However, increased abundance of Bifidobacterium has been reported in mouse models of endometriosis, 30 and thus far, the presence of endometriosis is not necessarily associated with a decrease in the abundance of bacteria such as Lactobacillus and Bifidobacterium.
The association between bacterial inflammation and endometriosis has been reported until now, 24  In recent studies, it was discovered that microbiome directly con- broad-spectrum antimicrobials in women with endometriosis. This is expected to improve infertility outcomes by ameliorating dysbiosis.
Further investigation will be needed to determine which treatments improve clinical outcomes such as infertility and pain of the endometriosis by correcting dysbiosis.
To the best of our knowledge, this is the first study to simultaneously analyze the microbiome of the vagina, endometrium, PF, and OF. The limitations of this study include the small number F I G U R E 3 Analysis of Lactic and Infect MBs in vagina and endometrium using cutoff value by ROC curve. There were significantly few cases with Lactic MB more than cutoff levels and were many cases with Infect MB more than cutoff levels in Endo Group at the vagina and endometrium of patients and lack of information about previous treatment of bacterial vaginosis. The existence of cases with hormone therapy before surgery and the lack of information on the menstrual cycle that may affect the microbiome are also serious limitations. We found that PF and OF were approximately sterile regardless of the presence of endometriosis and the type of cyst, implying that we were able to eliminate bacterial contamination in the sample collection and inspection process. This study suggests that dysbiosis may occur in women with endometriosis because there were many patients with a high abundance of infectious microorganisms and fewer patients with a high abundance of Lactobacillus and Bifidobacterium spp. in the Endo group in both the vagina and endometrium. Further research is needed to clarify whether dysbiosis really exists in endometriosis.

ACK N OWLED G M ENT
The authors would like to thank Enago for the English language review.

H U M A N R I G HT S S TATE M E NT S A N D I N FO R M E D CO N S E NT
All procedures in this study were conducted in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the principles of the Helsinki Declaration of 1964 and its later amendments. Informed consent was obtained from all patients. This article does not contain any studies with animal subjects.

CLI N I C A L TR I A L R EG I S TRY
Not applicable.