A promising future for endometriosis diagnosis and therapy: extracellular vesicles - a systematic review

Endometriosis is a chronic, inflammatory gynaecological disease that can have severe negative impacts on quality of life and fertility, placing burden on patients and the healthcare system. Due to the heterogeneous nature of endometriosis, and the lack of correlation between symptom and surgical disease severity, diagnosis and treatment remain a significant clinical challenge. Extracellular vesicles (EVs) are biologically active particles containing molecular cargo involved in intercellular communication, that can be exploited for diagnostic and therapeutic purposes. We systematically reviewed studies exploring EVs and their role in endometriosis, specifically addressing diagnostic and therapeutic potential and current understanding of pathophysiology. Five databases (Pubmed, Embase, Medline, Web of Science, Google Scholar) were searched for keywords ‘endometriosis’ and either ‘extracellular vesicles’ or ‘exosomes’. There were 28 studies included in the review. Endometrium derived EVs contribute to the development of endometriosis. EVs derived from endometriosis lesions contribute to angiogenesis, immunomodulation and fibrosis. Such EVs can be detected in blood, with early data demonstrating utility in diagnosis and recurrence detection. EV isolation techniques varied between studies and only eight of twenty-eight studies fully characterised EVs according to current recommended standards. Reporting/type of endometriosis was limited across studies. Varied patient population, type of sample and isolation techniques created bias and difficulty in comparing studies. EVs hold promise for improving care for symptomatic patients who have never had surgery, as well as those with recurrent symptoms after previous surgery. We encourage further EV research in endometriosis with the inclusion of rigorous reporting of both the patient population and technical methodology used, with the ultimate goal of achieving clinical utility for diagnosis, prognosis and eventually treatment. Supplementary Information The online version contains supplementary material available at 10.1186/s12958-022-01040-y.

The diagnosis of endometriosis at present is reliant on clinical history and examination, ultrasound, and ultimately laparoscopic surgery and biopsy [3,4]. Whilst endometriomas and deep infiltrating endometriosis can be diagnosed with increasing reliability by ultrasound [3], the more common manifestation of superficial peritoneal disease is poorly predicted without surgery [5].
Around half of women who have clinically suspected endometriosis undergo surgery to confirm the diagnosis [6]. Clinical suspicion of endometriosis is an inconsistent predictor of disease; of women undergoing laparoscopy for suspected endometriosis, 18-77% have surgically confirmed disease [7], although recent evidence suggests that the rates of microscopic endometriosis in women with "negative laparoscopy" may be up to 39% [8]. There is a strong need for a reliable non-invasive diagnostic tool to reduce the need for laparoscopy to diagnose or exclude endometriosis.
Management of endometriosis is generally divided into conservative, hormonal and surgical treatments. There is need for non-hormonal, non-surgical treatment options for endometriosis, especially in those actively trying to conceive. Surgical management is frequently utilised; however, due to vast heterogeneity of the literature, the benefits of surgery are debated and appear to be dependent on surgeon, operating technique, symptoms, disease stage and multiple other factors [3,9]. Recurrence is common after surgery, with increasing risk of recurrence over time. Treatment after surgery to prevent recurrence at present is largely reliant on suppression of menstruation, with limited efficacy [3].
Extracellular vesicles (EVs) are non-replicating particles delimited by a lipid bilayer that are released by all cells. EVs carry a variety of cargo involved in extracellular communication, altering the recipient cell phenotype [10]. EV is an umbrella term for different secreted particles secreted that differ by size and biogenesis, generally between 40 and 4000 nm in size, including apoptotic bodies, microvesicles and exosomes [11]. EVs contain bioactive cargo such as selective packaging of proteins, mRNA and miRNA species, which are reflective of the cell of origin and can provide insight into the physiological state of the tissue. EVs are abundant in bodily fluids making them prime candidates for use as "liquid biopsy", having been successfully isolated from blood, urine and saliva [12]. There is hope that EVs may replace tissue diagnosis as a reliable and non-invasive diagnostic tool for many pathological processes [13].
This diagnostic and prognostic potential has led to an escalation in EV research in the past years [13]. As such, the International Society for Extracellular Vesicles (ISEV) was formed in 2012, providing reporting guidelines to ensure scientific rigour, reproducibility and validity of EV research. The minimal information for studies of extracellular vesicles (MISEV2018) guidelines proposed by the ISEV recommend all EV studies report detailed sample collection, EV isolation methodologies and EV characterization [14]. Knowledge of the role of EVs in endometriosis has rapidly expanded in recent years. These recent advances explore both diagnostic and therapeutic potential, as well as giving an ever-increasing understanding of the underlying pathophysiology. With longstanding multiple theories of endometriosis pathophysiology including Sampson's theory of retrograde menstruation [15], metaplastic theory and others [16], EVs have provided a plausible connecting link between these hypotheses [16,17]. It has been established that EVs released from the endometrium of people with endometriosis differ from those without endometriosis [18][19][20]; and retrograde flow of such EVs has been confirmed by their identification in peritoneal fluid [17]. It has been hypothesised that these EVs potentiate migration and implantation of endometrial cells shed during retrograde menstruation, as well as manipulating the immune response to avoid their destruction, bearing many similarities to their role in cancer [17]. Furthermore, EVs released from newly implanted ectopic endometrial cells (i.e. endometriosis lesions) have been identified, and hypothesised to drive angiogenesis and neurogenesis, creating vascular, proliferative and potentially painful lesions [17]. These processes are most plausible and relevant for peritoneal deposits, but may also play a role in endometriomas, deep infiltrating endometriosis lesions and even rarely reported distant lesions (e.g. lung, brain) [16]. As peritoneal endometriosis is the most challenging to diagnose without surgery, EVs hold potential as a non-surgical diagnostic tool for these patients. For EVs to be of diagnostic value, they need to be both unique to the process of endometriosis and easily detectable in non-invasive samples (such as blood, urine, cervical secretions or potentially even stool samples [21]). If both criteria could be met, especially for patients with disease not detectable on ultrasound, the need for laparoscopy to diagnose or exclude endometriosis could be avoided, and it may be possible to gain information about disease severity preoperatively for those needing surgery. In addition, such EVs could provide therapeutic targets for non-hormonal and non-surgical treatment; potentially modifying disease progression and recurrence.
Here we present a systematic review of studies exploring EVs and their role in endometriosis, specifically addressing diagnostic and therapeutic potential and the contribution to the current understanding of pathophysiology. We explore which bodily fluids these EVs have been reported in, how the EVs were detected and their proposed function.

Inclusion and exclusion criteria
Articles were included if they were in English, peer reviewed, and assessed samples of extracellular vesicles in subjects with endometriosis (with or without controls). Human and animal studies were included. Reviews and conference abstracts were excluded.

Data extraction
Studies were assessed for the following information: human or animal studies; source of bodily fluid to obtain EVs; isolation technique to identify EVs; target within EVs being studied; implications for diagnosis, treatment and pathogenesis of endometriosis; and limitations. Data from reports were extracted according to these headings (as shown in Table 1), with all three authors contributing independently to the data extraction and collation. No automation tools were required, and no further data was requested from included study authors.

Search results
Twenty eight studies were included in the review. The initial search identified 274 articles of which 139 were unique. When applying inclusion criteria, 105 of these studies were excluded. A further six studies were excluded as they either did not specifically state that EVs were assessed, or did not specifically state that subjects had endometriosis. The flow of study identification is shown in Fig. 1.

Terminology
For the purposes of this review we use the term "endometrium" to imply eutopic endometrium; "endometriosis lesions" to describe any lesions outside the endometrium (commonly described as "ectopic endometrium" in the cited studies) and "endometrioma" to describe any endometriotic cysts on the ovary. Much of the current EV literature uses inconsistent nomenclature, naming a subtype of EVs such as exosomes without demonstrating the biogenic origins of the EVs in the fractions isolated. Thus, this review will use the term 'EV' throughout.

Pathophysiology
Proposed roles of EV cargo in the pathogenesis of endometriosis are outlined in Fig. 2. Biopsies from the endometrium showed differences in EV cargo between people with and without endometriosis in five studies [18-20, 30, 42]. These included cargo involved in immunomodulation [30] and angiogenesis [19], and multiple miRNAs with as yet unclear function [18,20]. Network analysis showed different expressions of miRNA, lncR-NAs and mRNAs between groups [42]. EVs isolated from peritoneal fluid also showed differences between people with and without endometriosis in three studies [23,33,46]. Specifically, cargo were identified that play a role in cell migration [33], immunomodulation [23,33,46], cell proliferation [23] and angiogenesis [23,33].
EVs isolated from plasma or serum showed differences between people with and without endometriosis in eight studies [18, 28, 34-36, 43, 45, 47]. Specifically, cargo were identified that play a role in cell migration    [35], immunomodulation [18,39,43,45], angiogenesis [18,28,34], fibrosis [47] and several lncRNAs with unclear function [36]. In addition, one study showed that EVs derived from mesenchymal stem cells may enhance migration of endometrial cells and therefore play a part in endometriosis [24]. Finally, one study showed differences in EVs derived from microbes in peritoneal fluid between women with endometriosis and controls, suggesting a role for microbiome in endometriosis aetiology [27]. The microbiome has been previously postulated to have an important role of the pathophysiology of endometriosis [21,49]; EVs may provide a powerful tool to explore this further.

Diagnostic markers
Three studies reported on sensitivity and specificity of specific EV cargo as biomarkers [28,36,45]. One study of 48 patients and 21 controls reported serum EV derived VEGF-C as a biomarker with sensitivity 81.3% and specificity 71.4% for endometriosis [28]. Another study of 85 patients and 85 controls reported serum EV derived lncRNA RP3-399L15.2 as a biomarker with sensitivity 67% and specificity 98%; or using a combination of lncR-NAs RP3-399L15.2 and CH507-513H4.6 a sensitivity of 80% and specificity of 85% [36]. A third study of 20 subjects and 20 controls reported a combination of serum EV derived miRNAs 320a and 22-3p as having sensitivity of 80% and specificity of 80% [45].
Additionally, a study of 52 patients and 21 controls reported serum derived pseudogene LGMNP1 having sensitivity 93% and specificity 76% in predicting recurrence within the four year study period, with a significant difference between subjects with primary and recurrent disease [39].

Extracellular vesicle derived therapeutic targets
Two studies utilised EVs derived from peritoneal macrophages as a potential therapeutic agent; demonstrating an ability to inhibit angiogenesis, migration and invasion of endometriosis in mouse models of endometriosis [29,30]. One study utilised EV derived miR-214 to downregulate fibrosis in a mouse model of endometriosis [41]. Another study showed that EV derived miR-301a-3p was overexpressed in endometriosis lesions compared with serum from healthy controls; and that downregulation of this miRNA in a mouse model influenced macrophage activity [26].

Limitations
Specific limitations for each study are listed in Table 1.
Other common limitations included small sample sizes, use of fetal bovine serum (FBS) for culture media supplementation without prior EV depletion [34,39,46] and unclear number of participants or samples used for assays.

Discussion
Whilst in its infancy, the use of EVs as diagnostic biomarkers or therapeutic targets in endometriosis has significant potential to improve the outcomes of people affected by this disease; additionally, several insights into the pathogenesis of the disease have been obtained. The recent excitement around EVs in cancer research has resulted in the development of strict reporting guidelines for researchers, with the next iteration of the MISEV guidelines to be released this year. As such the groundwork for investigating EVs in endometriosis is well established.
Using a systematic search protocol, we have identified and reviewed 28 studies that investigate EV cargo in endometriosis. Studies varied in their patient population, type of endometriosis, type of sample and EV isolation which created bias and difficulty in comparing studies; however, several insights into pathogenesis have been gained; diagnostic utility has been demonstrated; and therapeutic targets have been identified.

Limitations
There are multiple limitations in the included studies, which we broadly divide into clinical population limitations and laboratory processing limitations.
The clinical populations included in these studies were heterogeneous; with few studies reporting subject age and symptoms, and 65% reporting laparoscopic diagnosis and disease severity using accepted scoring systems. The limitation of current endometriosis scoring systems, especially with regards to clinical application, is well established; however in a research setting accepted scoring systems give some insight into the pathological extent of disease [50]. Nine studies only included subjects with endometriomas, and it is unclear at present whether EVs in this setting would differ from other manifestations of endometriosis.
With regards to laboratory processing bias, adherence to established MISEV standards was inconsistently Page 15 of 18 Scheck et al. Reproductive Biology and Endocrinology (2022) 20:174 achieved, limiting the reproducibility and reliability of many of the included studies. Complete EV characterisation is highly recommended to provide evidence of EV involvement when proposing a role in pathogenesis or EV cargo as biomarkers. The challenge of EV research is the heterogeneity of EV isolation methodologies creates difficulty in making meaningful comparisons across studies. Isolation methodology impacts on the composition of the EV fraction, with different size and density EVs and co-isolated contaminants favoured [51]. The lack of overlapping targets identified in any of the studies reviewed is likely a reflection of the diverse combinations of EV source, isolation techniques and methodologies used. Study reproducibility and the ability to validate study results is hindered by the absence of standardised sample processing and EV isolation methods. The majority (57%) of the listed studies isolated EVs from ex vivo cell culture of primary stromal, mesenchymal or endometriosis samples. Cell culture purity or validation of cell type in these studies were not performed. The digestion of tissue and 2D culturing of primary cells in a variety of media (EV depleted or not) has a significant impact on the EV and EV cargo that is released [52]. Therefore, it is difficult to compare EV isolation from primary cell culture to those found directly in patient samples.

Diagnostic potential
There is a critical need for non-invasive diagnostic biomarkers in endometriosis. The lack of specific symptoms, limited understanding of the pathology of the disease and delay from symptom onset to diagnosis via invasive laparoscopy creates a serious clinical challenge in treating endometriosis. As cell signalling mediators which carry tissue/disease specific cargo, EVs hold the potential to be exploited as diagnostic biomarkers. EVs in peritoneal fluid and serum probably represent a combination of both diagnostic and prognostic categories; however, sampling peritoneal fluid is not particularly useful as a diagnostic tool as laparoscopy or other invasive procedures are required. Identifying the EV cargo derived from a combination of endometrium and endometriosis lesions in serum likely holds the greatest pragmatic potential as a non-invasive diagnostic biomarker and may provide insight into both "potential for disease" and "progression of disease".
Four studies demonstrated that in samples taken from the endometrium, differences could be identified between people with and without endometriosis [18][19][20]30]. As shown in Fig. 3, these differences may imply a susceptibility to developing endometriosis as molecules within these EVs appear to facilitate the development of endometriosis lesions by influencing cell migration and proliferation and immunomodulation. These EVs may contribute to a phenotype susceptible to developing endometriosis, and are unlikely to be affected by surgical resection of disease, however this is speculative and yet to be investigated.
In contrast to this, EVs obtained from the peritoneal fluid and serum are likely to contain EVs derived from the endometrium in addition to EVs derived directly from endometriosis lesions, as well as numerous other sources. The diagnostic potential is therefore increased; however, reliable identification of the relevant EVs poses a more pronounced challenge. It could be assumed that analysis of EVs in these samples, in contrast to endometrial samples, would be markedly different after complete surgical excision of endometriosis; and therefore may bear promise as markers of current disease state. This could be exploited to detect recurrence, as was demonstrated in one small study [39]. Only 8 of the 28 studies systematically reviewed identified or discussed potential diagnostic markers, and only 4 of these measured AUC and/or sensitivity and specificity. These were serum RP3-399L15.2 [36] (Sn 67% and Sp 98%), serum VEGF [28] (Sn 81% Sp 71%), serum miR-22-3p and miR-320a [45] (AUC 0.883) and LGMNP1 [39] (as a marker for recurrence Sn 93% Sp 75%). Whilst promising, these studies have limited sample sizes (ranging from n = 20 to n = 86) and therefore further investigation and validation is required before clinical utility can be addressed.
Validation of these candidate diagnostic markers using clinically relevant means of isolation and quantification are also required. miR-22-3p and miR-320a were measured as biomarkers in EVs isolated using UC, while SEC was utilised when investigating EV-VEGF. While UC is the still the most widely used EV isolation methodology [53] and SEC generally yields higher purity EV fractions (depending on biofluid) [54], both have limited clinically utility due to specialist equipment requirements, typically time-consuming protocols and lack of scalability. Commercial kits tend to have short, straightforward protocols that are appealing for clinical use, however are plagued with high levels of non-EV contamination [55]. Therefore, the ideal means of EV isolation for clinical use may not yet have been developed; candidate biomarkers identified thus far may require future validation with a clinically implementable test.

Therapeutic potential
Many of the studies in this review used ex vivo cell culture techniques to investigate EV function in mediating macrophage polarisation, angiogenesis or proliferation. Therefore, the EV cargo identified in these studies were deemed potential therapeutic targets as they had some impact on endometriosis growth and inflammation.
Four studies using ex vivo cell culture techniques and mouse models were able to demonstrate disruption to the pathogenesis of endometriosis by targeting EVs therapeutically [26,29,30,41]. As discussed above, the effect of ex vivo cell culture on EVs limits the reliability of such research; nonetheless there is promise and a myriad of potential targets for ongoing research to develop therapeutic agents. Significant advances need to be made before use in the clinical setting becomes conceivable.
If there is indeed a specific EV phenotype that facilitates development of endometriosis, it is plausible that therapeutic agents aimed at cargo of these specific EVs may be able to prevent disease or recurrence; however, the utility in treating already established lesions may be limited.

Recommendations for future research
As the diagnostic and therapeutic potential for endometriosis from EV research increases; we expect to see a larger focus on clinical aspects in future studies. Endometriosis is a vastly heterogeneous disease, with clinical symptoms ranging from asymptomatic through to severely debilitating; and surgical disease ranging from minimal to major pathological anatomical changes, with limited correlation between the two [56]. Detailed clinical data on study participants, their symptomatology and surgical data would add further understanding into how EVs differ across this incongruous disease spectrum.
While several studies have shown promising diagnostic potential for the use of EVs as a non-invasive diagnostic biomarker; validation of these studies in large populations is warranted; additionally, more detailed comparison between subjects with differing clinical and surgical severity would add further insight into the clinical utility of such tests. Future studies into diagnostic biomarkers should focus on using clinically accessible EV isolation methods, to maintain the relevance and translatability of the study.
The effect of surgical excision on endometriosis associated EVs is yet to be established. As outlined above, we hypothesise that some endometrium derived EVs may predispose to endometriosis, and that these would not be affected by surgical intervention; whilst EVs derived from endometriosis lesions would diminish significantly after surgery and therefore potentially act as a marker of recurrence. Only one study looked at recurrence [39] and would support this hypothesis, however, studies to specifically address this issue would provide further insight.
As shown in Fig. 2 above, there was very little crossover between identified EV targets across the 28 studies included in this review; and currently there are hundreds if not thousands of specific identified EV cargo from multiple bodily fluids that play a role in endometriosis. With such a wide number of comparisons across studies and relatively small participant numbers, it is likely that some false positives exist. There may be merit in future studies further validating currently identified targets rather than searching for new targets, such that meta-analyses may be performed in the future to bolster subject numbers. Identifying similar targets across multiple different bodily fluids may also provide further insight into their pathological relevance and clinical utility.