Endometriosis is a chronic disease characterized by the growth of endometrial-like tissue outside of the uterus. It affects 5 to 10% of women globally (Giudice, 2010a). This condition is defined clinically by its potential to cause pelvic pain, dysmenorrhea, and infertility (Giudice, 2010a). Annually endometriosis costs in the United States have been estimated to be as high as $50 billion due to medical expenses, loss of worker productivity, and both social and personal burdens (Simoens et al., 2012). One of the most challenging problems of endometriosis is the prolonged time between disease onset and definitive diagnosis by laparoscopic surgery, which is approximately 7 to 10 years (Hadfield et al., 1996). This delay in diagnosis and early treatment is caused, in part, by the lack of non-invasive diagnostic tests for endometriosis (Ballard et al., 2006; Giudice, 2010b).

Although the causes of endometriosis are not completely understood, one commonly accepted theory suggests that “retrograde menstruation” (the flow of menstrual effluent (ME) through the fallopian tubes into the peritoneal cavity) is a major contributing factor. Retrograde menstruation occurs in almost all women (Halme et al., 1984) and therefore other factors must contribute to the development of endometriosis. In this context, a key characteristic of ME is the presence of mesenchymal stem cells (MSCs) and stromal fibroblast cells (SFCs), which are phenotypically similar (Denu et al., 2016; Haniffa et al., 2009; Hematti, 2012; Ulrich et al., 2012), and are found within endometriosis lesions (Figueira et al., 2011) and eutopic endometrium (Barragan et al., 2016; Gargett, 2006; Masuda et al., 2010; Gargett, 2007; Fazleabas et al., 2002). MSCs and SFCs can be readily grown in culture and exhibit immunomodulatory and angiogenic properties, making them a potential therapeutic or regenerative resource for a variety of conditions, such as cardiac ischemia, limb ischemia, stroke, and lung injury (Zhang et al., 2013; Meng et al., 2007; Vu et al., 2015; Borlongan et al., 2010; Rodrigues et al., 2016).

Previous studies report that SFCs isolated and cultured from endometrial biopsies have the capacity to decidualize (Barragan et al., 2016). Decidualization is a normal process within the mid-secretory phase during which endometrial stromal cells differentiate into specialized secretory decidual cells required for successful embryo implantation and placental development (Gellersen & Brosens, 2014; Dunn et al., 2003; Ramathal et al., 2010). Giudice and colleagues have clearly shown that SFCs derived from endometrial biopsies obtained from patients with endometriosis display impaired decidualization ex vivo when compared to endometrial biopsy-derived SFCs obtained from control subjects (Barragan et al., 2016). Thus, we postulated that ME-derived SFCs may offer a non-invasive resource for the early diagnosis of endometriosis based on their phenotypic differences and decidualization capacity. In addition, we explored whether other phenotypic characteristics of ME-derived cells could be informative for diagnosing and better understanding the pathobiology of endometriosis, as well as other reproductive disorders. To enable these goals we established a protocol for the collection and transport of ME for investigative studies.

Almost all women experience retrograde menstruation, a process by which menstrual effluent (ME) shed from the endometrium is effluxed through the fallopian tubes into the peritoneal cavity, where it implants on the serosal surface (Halme et al., 1984). Evidence that effluxed ME is likely to be a major source of cells responsible for the formation of peritoneal endometriosis lesions in patients is supported by studies in baboons showing that consecutive intraperitoneal injections of shed menses results in endometriosis lesions in the peritoneal cavity and by the presence of spontaneous endometriosis in menstruating non-human primates, along with the complete absence of endometriosis in non-menstruating mammals (Fazleabas et al., 2002; D'Hooghe et al., 1994; D'Hooghe et al., 2009). Our findings suggest that analysis of ME shed from the endometrium can be harnessed to investigate the pathobiology of endometriosis and to develop early, non-invasive diagnostic methods for endometriosis.

Individual differences in the cellular content of ME offer a potential tool for investigating why only 5–10% of females actually develop endometriosis, given the ubiquity of retrograde menstruation. A relatively small number of published studies have analyzed the composition of ME, comparing it to that of peripheral blood; these studies primarily focused on phenotyping the CD45⁺ cell populations (van der Molen et al., 2014; Hosseini et al., 2015). Consistent with previous reports we found that ME is mainly comprised of CD66b + granulocytes, followed by CD3+ lymphocytes, CD56^bright uNK cells, and then CD14+ monocytes and CD20+ B cells (van der Molen et al., 2014; Hosseini et al., 2015). The variability that we observed among the various subpopulations of CD45+ cells (Fig. 1a  and c) was similar to that previously reported (van der Molen et al., 2014; Hosseini et al., 2015). Neither of these previous studies compared ME obtained from endometriosis subjects to controls. Despite the variability in the cellular composition of ME, we found that endometriosis subjects had significantly lower percentages of uNK cells when compared to controls (Fig. 1c). uNK cells are CD56^bright/CD16-, unlike peripheral blood NK cells which are CD56^dim/CD16+. uNK cells are proposed to play an important role in endometrial vascularization and decidualization, as well as the development of decidualization-related infertility observed in women with endometriosis (Gellersen & Brosens, 2014).

Decidualization is a process in which endometrial stromal fibroblast-like cells differentiate into specialized secretory decidual cells during the mid-secretory phase to prepare the uterus for implantation and placental development (Gellersen & Brosens, 2014; Dunn et al., 2003; Ramathal et al., 2010). A very recent study reveals that uNK cells target and eliminate senescent endometrial stromal fibroblast-like cells, and this may be an important process for successful decidualization (Brighton et al., 2017). A reduction in uNK cells found in ME collected from subjects with endometriosis may be linked to the impaired elimination of senescent uterine cells that compromises the decidualization process – which is implicated in the infertility observed in patients with endometriosis. Future studies will focus on more detailed analyses of the NK cell population found in ME collected from controls and subjects with endometriosis.

Several previous studies have highlighted the regenerative capacity of ME-derived CD45- mesenchymal and stromal cells (Zhang et al., 2013; Meng et al., 2007; Vu et al., 2015; Borlongan et al., 2010; Rodrigues et al., 2016). The CD45- cell fraction contributes to approximately 1.5–2% of the total non-RBC portion of ME. The majority of these cells were not endothelial or epithelial cells, but rather CD73/CD90/CD105 expressing cells. We observed no significant differences in these subpopulations of CD45- cells found in ME collected from endometriosis subjects vs. controls (Fig. 1d). However, we did not include a distinction between the endometrial MSC and SFC populations in the ME; this could be addressed in future studies using CD146/CD140b markers, as described by studies of endometrial biopsies (Barragan et al., 2016). Among cultured SFCs, we did observe a lower level of expression of CD105 (endoglin) in those isolated from ME collection from endometriosis patients. CD105 is an auxiliary receptor for the TGFβ receptor complex and regulates the binding of various ligands, including TGFβ1 and TGFβ3, as well as BMP-2 and -7 (Guerrero-Esteo et al., 2002). Differences in CD105 expression between endometriosis patients and controls have been reported by others (Hayrabedyan et al., 2005), but these studies focused on CD105 expression by the microvasculature of eutopic endometrium and ovarian endometriosis and not non-vascular CD105 expression. This finding will need to be explored further in future analyses.

A major innovation of this study over prior studies profiling the cellular composition of ME is that we included functional assessment of the SFCs (of the CD45- fraction). Taking advantage of the fact that defects in stromal cell decidualization (obtained from invasive uterine biopsies) have been reported in the setting of endometriosis (Klemmt et al., 2006; Aghajanova et al., 2011; Yin et al., 2012), we compared ME-derived SFCs obtained from controls and endometriosis subjects for their decidualization response using cAMP. We demonstrate that a reduced decidualization response by ME-derived SFCs is clearly associated with endometriosis, as shown in Fig. 3. In fact, ME-SFCs obtained from women with endometriosis exhibit significantly reduced IGFBP-1 expression in the absence of decidualization stimuli (Fig. 3). These differences persist over time in culture and likely reflect a quantitative impairment in decidualization capacity. This difference is observed with cAMP alone, as well as when decidualization is induced in the presence of cAMP + MPA + E2 (Additional file 2: Figure S1).

Interestingly, we have occasionally observed impaired decidualization capacity in control subjects’ SFCs (Fig. 3). This may reflect the presence of asymptomatic endometriosis in these individuals. Alternatively, genetic variation may influence the decidualization response in the absence of endometriosis. Genetic differences play a significant role in risk for endometriosis; a recent meta-analysis of over 170,000 endometriosis patients and almost 200,000 controls revealed 14 genetic regions of interest, including WNT4, GREB1, IL1A, and CDKN2B (Sapkota et al., 2017). The examination of phenotypic diversity of ME may also shed light on the mechanisms of genetic predisposition (Rahmioglu et al., 2015). In this context it will be of interest to examine decidualization capacity in control subjects who carry genetic risk alleles for endometriosis (Rahmioglu et al., 2015). For example, WNT4 plays a direct role in the decidualization response, and WNT4 risk haplotypes for endometriosis have been reported to contain an estrogen response element that may regulate its expression (Zhang et al., 2017). Over the last decade, we have established a large registry of normal controls (GaP) with GWAS genotype data and subjects can be recalled for phenotypic studies based on a genotype of interest (Gregersen, 2009; Gregersen et al., 2015). This offers the possibility of examining a variety of endometriosis risk haplotypes with regard to their impact on decidualization capacity in various populations. The use of ME as a source of SFCs makes such studies possible without the need for invasive procedures. Thus, we believe that one of the major benefits of this study is to open up the possibility of carrying out population-based genotype-phenotype studies that are relevant to better understanding the underlying pathogenesis of endometriosis.

Another potential benefit of studying ME, which can be sampled throughout menstruation and monthly, is that it may lead to the development of early, non-invasive diagnostic methods. The long delay between symptom onset and diagnosis in this disease is well established (Hadfield et al., 1996; Arruda et al., 2003), and no laboratory diagnostic makers have emerged to replace, or even guide, the performance of the gold standard of laparoscopic surgery and histologic confirmation. The expression of BCL6 (Evans-Hoeker et al., 2016) has been reported to be altered in secretory phase endometrial biopsies and may be useful diagnostically, but this approach will require invasive biopsy procedures. Other possible diagnostic markers for endometriosis include CA-125 (Hirsch et al., 2017), VEGF, microRNAs, immunologic markers, and soluble ICAM levels (Acimovic et al., 2016; Vodolazkaia et al., 2016; Gagné et al., 2003; Matalliotakis et al., 2001; Kuessel et al., 2017; Mosbah et al., 2016; Bohler et al., 2007; Drosdzol-Cop et al., 2012; Wu et al., 2015); however, none of these can identify endometriosis patients with both high sensitivity and specificity. It is clear that the need for a non-invasive diagnostic marker or method remains, and it would appear that the analysis of ME may offer a path to develop such tests. As discussed above, our findings of a reduced uNK cell population in the ME of endometriosis subjects may not be specific for endometriosis, and may simply reflect the relative lack of decidualization in these subjects, as decidualization is accompanied by extensive uNK infiltration during the mid-luteal phase under normal conditions (Gellersen & Brosens, 2014). Future studies are warranted to investigate the mechanism underlying this difference.

Although we observed significant decidualization by ME-derived SFCs (based on IGFBP-1 protein release) within 6 h of cAMP simulation, a diagnostic based on culturing ME-derived stromal cell populations to assess decidualization capacity is not practical for routine clinical use. We believe that a direct assessment of freshly isolated stromal cells from ME is likely to be the most promising approach, based on gene expression, epigenetic analysis or other phenotypic changes in these cells. The fact that gene expression differences, such as ALDH1A1, can be observed in unstimulated cultured SFCs is encouraging. However, our sample size is small and a much larger study of patients and controls is required to establish which, if any, gene expression patterns by ME-derived SFCs will have diagnostic utility. Ideally, these studies should be done on selected, uncultured ME cells in order be consistent with a practical diagnostic test.

This study is the first to demonstrate that uNK cells are significantly reduced in ME obtained from endometriosis subjects and that ME-derived SFCs from subjects with endometriosis exhibit significantly impaired decidualization, along with lower expression of ALDH1A1 than controls. However, there are several limitations that need to be considered. First, the population sizes for our experiments are small, given the prevalence of endometriosis. In order to be confident of our results, we collected multiple ME samples during the first 3 days of the cycle (days 0–2 only) and over multiple cycles for several subjects for the phenotyping studies. We also restricted our functional assays on decidualization capacity using SFCs collected from donors who reported no hormone use, in case hormone use affects decidualization outcomes. In addition, given the small number of samples analyzed by RNA-Seq, we verified our results on ALDH1A1 using qPCR with SFCs using a larger group of controls and affected subjects. We did not stage our endometriosis subjects based on disease severity, in part because we did not have the confirmed, documented data and because of our restricted sample size. However, all endometriosis diagnoses were confirmed by laparoscopy. It is also important to note that the stage of endometriosis does not necessarily reflect the severity of clinical symptoms a patient presents with (Johnson et al., 2017). It is our goal to develop a diagnostic tool that identifies endometriosis across all stages of endometriosis (particularly, during the early stages); however, having information on the stage of endometriosis would be useful to identify population subsets. Finally, as discussed above, our control group was not laparoscopically confirmed and this may contribute to the small number of subjects that did not ‘classify’ with the result of the controls.

Our studies demonstrate that the ready availability of ME will permit larger studies in the future to develop non-invasive diagnostic methods for endometriosis. These studies will also enable investigation of disease heterogeneity and mechanisms of pathogenesis. Decidualization defects may arise through many different genetic and environmental mechanisms, and these may suggest different approaches to therapy. Indeed it is possible that decidualization defects and gene expression changes in eutopic endometrium may in some cases arise as a result of the presence of endometriosis lesions in the peritoneum, as suggested by studies in the baboon (Fazleabas et al., 2002). In this context it will be of interest to study ME phenotypes before and after surgical or medical treatments and to correlate these findings with the presence or absence of persistent infertility. ME is particularly useful for these types of studies because it can be collected in a non-invasive manner repeatedly over several months. Finally, ME analysis may be very useful in the general evaluation of female infertility, particularly related to decidualization, outside of endometriosis (Gellersen & Brosens, 2014).

In conclusion, we propose that the routine analysis of ME phenotypes holds promise for understanding and diagnosing endometriosis, as well as potentially other reproductive disorders. This may include the use of proteomic and metabolomic methods, as well as emerging technologies of single cell analysis of freshly isolated cells. Collection using the menstrual cup has proved remarkably useful and should be able to support a variety of such studies, although we believe there is room for development of collection devices that may be tailored to particular analytic questions. We hope that the studies reported here will provoke a wider interest in ME as a clinical and research tool.