Here, we applied the original CIBERSORT gene signature file LM22 which defined 22 immune cell subtypes and analyzed datasets from ectopic endometrial tissue (EC) and eutopic endometrial tissue (EU) in patients with EMS and normal endometrial tissues (NO).

All microarray gene expression data were obtained from the public GEO database (https://www.ncbi.nlm.nih.gov/gds/) under the following conditions: “endometriosis, series, expression profiling by array, homo.” June 2019 was used as the deadline for searching.

According to the research purpose and research scope of this project, a total of 7 series chips were included: GSE120103, GSE51981, GSE25628, GSE37837, GSE7846, GSE6364, and GSE7305.

A total of 309 cases of gene transcriptome data were downloaded, including 116 cases of NO, 158 cases of EU EMS patients and 35 cases of EC EMS patients, and the corresponding r-AFS classification information for patients was downloaded.

Raw data from Affymetrix Human Genome U133 Plus 2.0 cel files and Affymetrix Human Genome U133A 2.0 cel files were processed by using robust multiarray analysis (RMA) method, normalized according to quantiles method and subsequently log-transformed in RMAExpress software (version 1.0.5, http://rmaexpress.bmbolstad.com/). Raw data from Agilent Technologies text files were processed by using the “normexp” function with an offset of 50 for background adjustment, normalized according to quantiles method and subsequently log-transformed in R software. Heterogeneity and latent variables are widely recognized as primary sources of variability and bias in high-throughput experiments, and the most well-known sources of latent variation in genomic experiments are batch effects when samples are performed on different days, by different people or in different groups. Surrogate variable analysis (SVA, http://www.bioconductor.org/packages/release/bioc/html/sva.html) package is used to identify and remove batch effects and other unwanted sources of variation. Data were further normalized using the combat package in SVA to correct for batch effect. The gene expression of each sample before batch normalization is shown in Supplementary Table 2. Principal component analysis (PCA) was performed on the data before (Supplementary Figure 1A) and after (Supplementary Figure 1B) batch normalization. All samples were analyzed for immune cell profiles by CIBERSORT, and the number of permutations was set to 100. Permutation represents times of the algorithm run using the default signature matrix. Theoretically, the accuracy of the results increases with the increase of running times (Newman et al., 2015). Twenty-two immune cell types, together with CIBERSORT metrics, such as the Pearson correlation coefficient, CIBERSORT p-value and root mean squared error (RMSE), were quantified for each sample. The CIBERSORT p-value reflects the statistical significance of the deconvolution results across all cell subsets and is useful for filtering out deconvolution with less significant fitting accuracy (https://cibersort.stanford.edu). From all samples analyzed, we selected 68/24/112 NO/EC/EU samples (Supplementary Table 3) that met the requirements of CIBERSORT p-value < 0.05. The histogram of 68/24/112 NO/EC/EU samples based on immune cell proportions was drawn by “barplot” function in R software (Supplementary Figure 2).

According to the r-AFS classification downloaded from the GEO database, 30 samples with missing staging information were removed, and the remaining 82 EU cases of EMS patients were divided into 20 cases of stage I~II endometrial tissue and 62 cases of stage III~IV endometrial tissue (Table 1). The list of selected samples and corresponding GEO accessions is shown in Table 1. The immune cell profile was calculated for each sample, and the upper quartile (P25), median value and lower quartile (P75) for each tissue type (NO, EC and EU) were calculated. Kruskal-Wallis tests were applied to analyze differences among EC, EU and NO. For macrophages, Pearson correlation coefficients with other immune cell types were calculated using SPSS 24.0 software.

The total macrophage fraction was calculated as the sum of M0, M1, and M2 macrophage fractions. Total T cells were calculated as the sum of CD8 T cells, naive CD4 T cells, resting CD4 memory T cells, activated CD4 memory T cells, follicular helper T cells, regulatory T cells (Tregs) and gamma delta T cells.

A total of 60 anonymized patients with ovarian endometriomas and 30 anonymized patients without EMS were enrolled and pathologically confirmed by the Pathology Department of the First Affiliated Hospital of Sun Yat-sen University. All patients with EMS of childbearing age without hormone or drug treatment in the last 6 months underwent laparoscopic ovarian cyst removal from December 2017 to December 2019. Ectopic endometrial tissue and eutopic endometrial tissue were collected from the same patient with ovarian endometriomas and classified according to the r-AFS system. Paraffin sections of endometrial tissues as the control group were taken from patients without EMS who underwent laparoscopic myomectomy. Normal endometrial tissue and eutopic endometrial tissue were obtained by endometrial biopsy. Ectopic endometrial tissue was obtained by laparoscopic ovarian endometriomas resection. The pathology of endometrial tissue specimens was proliferative endometrium. Endometrial hyperplasia or malignant transformation, previous history of EMS surgery or amenorrhea, acute inflammation and autoimmune disease were excluded from the study. All specimens were obtained with the informed consent of patients, and this study was approved by the Medical Research Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University. Clinical data were obtained in a well-designed questionnaire, including age, body mass index (BMI), menstrual cycle length in days, number of days of menstrual bleeding, dysmenorrhea visual analog scale (VAS) score and r-AFS score (Tables 4, 5). The VAS consists of a line usually 100 mm in length, with anchor descriptors such as “no pain” and “worst pain imaginable.” BMI is a statistical index using a person's weight and height to provide an estimate of body fat. It is calculated by taking a person's weight, in kilograms, divided by their height, in meters squared, or BMI = weight (in kg)/ height² (in m²).

To explore the distribution of macrophages in eutopic and ectopic endometrium of patients with EMS, immunohistochemical experiments were conducted to determine the integral optic density (IOD) values of CD68 and CD206 in endometrial tissues of the control group and eutopic and ectopic endometrium of patients with EMS to calculate the ratio of CD206/CD68.

Total macrophages and M2 macrophages were evaluated immunohistochemically using staining for CD68 and CD206, respectively. All endometrial tissues were fixed with 10% neutral buffered formalin fixative, dehydrated and embedded in paraffin. Wax blocks were sectioned into 4 μm continuous sections, and slides were randomly selected for immunohistochemical staining.

For IHC, 4-μm paraffin sections were baked at 60°C for 1 h, deparaffinized in xylene and rehydrated via graded ethanol. Then slides were microwaved in citric buffer (10 mM citric acid, pH 6.0) and cooled at room temperature to retrieve antigens. Endogenous peroxidase activity was blocked by 3% H₂O₂, and non-specific binding sites were blocked by 10% goat serum.

The following primary antibodies were used: mouse anti-CD68 antibody (ab31630; Abcam, USA; 1:200 dilution, cytoplasmic or/and membrane staining) and rabbit anti-CD206 antibody (ab64693; Abcam, USA; 1:1,000 dilution, cytoplasmic or/and membrane staining). Next, the primary antibody against CD68 or CD206 were added to the specimen sections at 4°C and incubated overnight. The primary antibody was removed, and the secondary antibody was incubated with the samples (1:1,000) at 37°C for 30 min. The immunohistochemical MaxVision kit solution was incubated with tissue samples at room temperature for 20 min, and thereafter, 3, 3′-diaminobenzidine (DAB) chromogenic reagent was added. Finally, hematoxylin staining was performed. The primary antibody was replaced with phosphate buffered saline, which was used as the negative control. IHC staining of positive samples was repeated twice.

The results of slides were identified by a double-blind method, and immunohistochemical staining results of all slides were independently judged by two senior pathologists. Using semi-quantitative integration method, six visual fields were randomly selected under a high-power microscope, and immunohistochemical results are expressed by the average IOD of six random visual fields: IOD = density (mean) ^* area. Density reflects the concentration or intensity of positive protein, and IOD reflects the total protein expression in the selected area. For measurement IOD, the image system comprised a Leica CCD camera DFC420 connected to a Leica DM IRE2 microscope (Leica Microsystems Imaging Solutions Ltd, Cambridge, United Kingdom). Photographs of representative fields were captured under high-power magnification (×400) using Leica QWin Plus v3 software. The IODs of in each image were measured using Image-Pro Plus 6.0 software (Media Cybernetics Inc., Bethesda, MD, USA).

The median value (P25-P75) was used to express the fraction of three groups of different immune cells obtained from the GEO database, and the Kruskal-Wallis test was used to compare them because the data didn't conform to normal distribution and homogeneity of variance. The pearson correlation test or spearman rank correlation test was used to analyze the correlation between the two groups of quantitative data. SPSS 24.0 statistical software was used for data analysis. Counting data were expressed by percentage. Qualitative data were described by frequency, and measurement data were measured by nonparametric mean test (X² test and t-test). Results were considered statically significant when p < 0.05.

The fraction of follicular helper T cells was lower in EC than in EU (p = 0.041) (Figure 1C, Table 2). The fractions of CD8 T cells, resting CD4 memory T cells and plasma cells in EC were higher than that in EU and NO (Figures 1D–F, Table 2). There was no significant difference in other T cell subsets and total T cells among the three groups (Figure 1A, Table 2). There was no significant difference in the fractions of total B cells (Figure 1B, Table 2) and B memory cells among the three tissues (Table 2). The fraction of naive B cells in all three tissues was low (Table 2).

The fraction of total macrophages was significantly higher in EC than in EU and NO (p = 2.7e⁻⁶, p = 0.0015), but there was no significant difference between NO and EU (p = 0.052) (Figure 2A). Compared to EU and EC, total dendritic cells were increased in NO (p = 0.011 and p = 0.051, respectively), but there was no significant difference between EC and EU (p = 0.7) (Figure 2D). In contrast, total NK cells were significantly decreased in EC (p = 3.9e⁻⁷, p = 4.1e⁻⁵) (Figure 2C). The fractions of total mast cells, neutrophils, eosinophils and monocytes were not significantly altered among the tissue types (Figures 2B,E–G).

M2 macrophages were increased in EC compared to NO and EU (p = 5.1e⁻⁵ and p = 9.4e⁻⁷, respectively) (Figure 3C). M2 macrophages/ (M1 macrophages+M2 macrophages) in EC were higher than that in NO or EU (p = 0.36 and p = 0.46, respectively) (Figure 3D), but there was no significance difference, indicating that M2 macrophage polarization existed in EC. In contrast, neither M0 macrophages nor M1 macrophages were significantly different among tissues (Figures 3A,B).

The fraction of activated mast cells in endometrial tissues was low (Figure 4B). Resting NK cells were significantly decreased in EC (p = 0.0077, p = 0.011), but there was no significant difference between EU and NO tissues (p = 0.81), and activated NK cells were also significantly decreased in the EC group (p = 0.0097, p = 0.00055) (Figures 4C,D). Activated dendritic cells were decreased in EC compared to NO and EU (p = 0.012 and p = 0.13, respectively) (Figure 3F). The fractions of resting mast cells and resting dendritic cells were not significantly altered among the tissue types (Figures 3A,E).

The r-AFS classification is used to predict the recurrence potential of EMS after surgery, reflecting severity of EMS to a certain extent. Generally speaking, stage III ~ IV exhibits early recurrence and poor prognosis. Therefore, we investigated differences in immune cell patterns of eutopic endometrium between r-AFS stage I ~ II and stage III ~ IV. Fraction of resting NK cells (p = 0.003) were decreased in eutopic endometrium from people with stage III ~ IV compared to stage I ~ II (Figure 5, Table 3). In contrast, fractions of activated NK cells (p = 0.003) and M2 macrophages (p = 0.026) were increased in eutopic endometrium from people with stage III ~ IV (Figure 5, Table 3). M2 macrophages/ (M1 macrophages+M2 macrophages) were higher in eutopic endometrium from people with stage III ~ IV compared to stage I ~ II (p = 0.414) (Table 3), but there was no significance difference. The primary immune cells in tissues are follicular helper T cells, NK cells, M2 macrophages and resting mast cells (Figure 5, Table 3). Thus, different EMS r-AFS classifications were associated with distinct immune phenotypes.

To further elucidate immue cell network in EC, we analyzed correlations of different immune cell populations by calculating r² Pearson correlation coefficients (Figure 6). Immune cells with larger correlation coefficients in EC included naive B cells and memory B cells (−0.52); naive B cells and CD4 memory activated T cells (0.7); naive B cells and resting dendritic cells (0.62); resting dendritic cells and CD4 memory activated T cells (0.59); resting dendritic cells and activated NK cells (−0.51); activated mast cells and resting mast cells (−0.57); activated mast cells and M0 macrophages (0.53); naive CD4 T cells and resting NK cells (0.78); naive CD4 T cells and monocytes (0.6); naive CD4 T cells and Tregs (0.78); monocytes and Tregs (0.54); activated dendritic cells and eosinophils (0.65) (Figure 6). Naive B cells correlated positively with CD4 memory activated T cells and resting dendritic cells in EC. However, they correlated negatively with memory B cells in EC. Furthermore, resting dendritic cells correlated positively with CD4 memory activated T cells, and correlated negatively with activated NK cells in EC. Naive CD4 T cells correlated positively with monocytes, resting NK cells and Tregs in EC, and monocytes correlated positively with Tregs. Additionally, activated mast cells correlated positively with M0 macrophages, and correlated negatively with resting mast cells in EC. Activated dendritic cells correlated positively with eosinophils.

To verify the exploratory data obtained for macrophages, we conducted immunohistochemical experiments to detect the IOD values of CD68 and CD206 in endometrial tissues of control, eutopic and ectopic endometrium of patients with ovarian endometriomas and calculated the ratio of CD206/CD68, using SPSS 24.0 statistical software to analyze the data.

Patients with ovarian endometriomas were divided into two groups according to r-AFS stage, including 30 cases of stage I ~ II and 30 cases of stage III ~ IV. Patients in the control group were 24–49 years old with an average age of 36.20 ± 5.40. Patients in stage I ~ II were 24–46 years old with an average age of 32.37 ± 6.19, and patients in stage III ~ IV were 24–48 years old with an average age of 33.93 ± 7.13 (p = 0.065). These differences were not statistically significant (Table 4).

Other basic data, including BMI, menstrual cycle length in days, number of days of menstrual bleeding, etc., showed no significant difference among the three groups (p > 0.05) (Table 4). There was no significant difference in VAS scores of dysmenorrhea between stage I ~ II and stage III ~ IV EMS patients (Table 5). The r-AFS scores of patients with stage I ~ II EMS were 2~15, with an average score of 10.03 ± 4.11, while those of patients with stage III ~ IV EMS were 30~132, with an average score of 78.27 ± 26.73 (p < 0.001) (Table 5).

Examples of CD68 and CD206 macrophage staining in EC tissues, along with a quantification summary, are shown in Figure 7 and Table 6. Immunohistochemical results demonstrated that CD68 and CD206 were expressed in normal endometrium, eutopic endometrium and ectopic endometrium, exhibited cytoplasmic/membrane staining. Staining was distributed in the periphery and stroma of glands, and its positive expression showed as brown yellow and its nucleus as blue.

In agreement with the CIBERSORT results, macrophage density was higher in EC compared to eutopic and normal endometrial tissue. Meanwhile, CD206/CD68 was increased in EC, which confirmed M2 macrophage polarization in ectopic endometrium. Compared to stage I ~ II, CD68 and CD206 density and CD68/CD206 in endometrial tissue of stage III ~ IV EMS were increased.