Surgical removal of endometriotic lesions alters local and systemic proinflammatory cytokines in endometriosis patients

To evaluate the impact of lesion removal on the systemic circulating levels of cytokines, we compared plasma samples obtained from endometriosis patients before surgery, 2 weeks after surgery, and 3 months after surgery. We found that the levels of GM-CSF, IL-2, IL-8, and IL-10 decreased significantly after removal of lesions ( Fig. 1A ). However, post hoc analysis revealed that this difference was significant only when comparing presurgery versus 2-week postsurgery samples. At 3 months after surgery, the levels of these cytokines were found to increase close to the levels measured before surgery. An overall trend was noted where cytokine levels decreased after surgery in all except four cytokines, although the changes did not achieve statistical significance. The exceptions were IL-12 and IL-13, which remained the same after surgery and were then increased slightly at 3 months after surgery. IP-10 and RANTES seemed to have a delayed decrease: They showed a slight increase 2 weeks after surgery and then a decrease at 3 months after surgery. When the data were stratified by disease stage, GM-CSF levels were found to decrease significantly after surgery in advanced endometriosis patients (stages III–IV), whereas the decrease was not statistically significant in early endometriosis patients (stages I–II). However, post hoc analysis revealed that the differences seen in the advanced stage patients were not significant for any particular pairwise comparison, despite having an overall P value ( P =.039) that was significant ( Fig. 1B ). These results indicate that the decrease in GM-CSF levels was more pronounced in advanced-stage patients than in early-stage patients. No differences were detected for any other cytokine. When data were stratified by disease stage and time point, RANTES levels were found to be lower in presurgery samples from early-stage patients (stages I–II) compared with presurgery samples from advanced-stage patients (stage III–IV of endometriosis). This means that before surgery, early-stage patients had lower blood levels of RANTES than advanced-stage patients. After surgery, these differences disappeared. No differences were found in 2-week or 3-month postsurgery samples. Also, no differences were observed for any other cytokine ( Fig. 1C ). We performed several comparisons on plasma samples to determine if there were differences between endometriosis patients and normal women and to evaluate the potential influence of such differences in the pathogenesis of the condition. We found 19 different cytokines to be differentially expressed in endometriosis patients compared with normal subjects ( Table 1 ). Interestingly, the levels of 18 of these cytokines were higher in endometriosis patients compared with normal subjects. MCP-1 was the only cytokine expressed at lower levels in endometriosis patients compared with normal subjects, but only when comparing postsurgery samples. When comparing presurgery levels of MCP-1, no significant difference was observed. When endometriosis data were stratified by disease stage, we found that the levels of 11 cytokines were higher than in normal women ( Supplemental Table 1 , available online at www.fertstert.org ). For all 11 cytokines, the levels were higher only in advanced-stage presurgery samples compared with normal samples, meaning that before surgery, advanced-stage patients had significantly higher levels of these cytokines than normal women, and after surgery the differences disappeared. These differences were not observed in early-stage patients. However, within these 11 cytokines, G-CSF and IL-7 were found to be more highly expressed in early-stage patients at all time points compared with normal women. This means that endometriosis patients had higher levels of G-CSF and IL-7 than normal women regardless of the surgery. Finally, we compared endometriosis samples with normal samples stratified by phase of menstrual cycle (secretory or proliferative phase), to determine if menstrual cycle fluctuations would influence the comparison ( Supplemental Table 2 , available online at www.fertstert.org ). The results mostly agree with the findings presented in Table 1 . We found that 12 out of 18 cytokines were more highly expressed in endometriosis patients at all time points compared with all normal patients. The remaining six cytokines had varying results. Levels of IL-1β and IL-13 were higher in presurgery and 3-month postsurgery samples compared with normal samples from women in the proliferative phase, but not the secretory phase. Eotaxin levels were higher in 2-week postsurgery samples compared with normal proliferative-phase samples only. IL-2 levels were higher in presurgery samples compared with normal proliferative samples. To evaluate the differences between eutopic and ectopic endometrium, we compared cytokine levels between matched eutopic and ectopic tissue obtained from the same patient before surgery ( Fig. 2 ). Differential expression of nine cytokines was detected in eutopic compared with ectopic tissue. Levels of bFGF, IP-10, IL-1Ra, G-CSF, MIP-1β, IL-7, and IL-5 were significantly higher in eutopic than in ectopic tissue. Conversely, levels of GM-CSF and IL-15 were lower in eutopic compared with ectopic tissue. When samples were stratified by disease stage, we detected differential expression of IL-1Ra, IL-5, IL-7, IL-9, IL-15, bFGF, G-CSF, GM-CSF, IP-10, MIP-10, and TNF-α when comparing eutopic versus ectopic tissue in early stage patients. When comparing eutopic and ectopic tissues in advanced-stage patients, only IL-2 was found to be differentially expressed in eutopic versus eutopic tissue ( Fig. 3 ). No differences were observed for the other cytokines. When tissue samples were stratified by disease stage and tissue type, levels of PDGF-BB, IL-4, IFN-γ, IL-9, bFGF, and TNF-α were found to be higher in the early-stage eutopic samples than in advanced-stage eutopic samples ( Supplemental Fig. 1 , available online at www.fertstert.org ). When comparing early- versus advanced-stage ectopic samples, GM-CSF levels were higher in early-stage ectopic samples than in advanced-stage ectopic samples. Conversely, ectopic samples had lower G-CSF levels in the early-stage samples than in the advanced-stage samples, but those differences were not seen in the eutopic samples. No differences were observed for the other cytokines. Peritoneal fluid samples from endometriosis patients were compared to evaluate the impact of disease stage on local cytokine levels ( Supplemental Fig. 2 , available online at www.fertstert.org ). IL-5 levels changed significantly when comparing disease stages I, II, and III, but post hoc tests revealed that differences were not significant despite a significant overall P value. IL-12 was also found to change significantly, but only when comparing stage III versus stage I, which could indicate that a significant change occurs in the transition from early to advanced stages. The cytokine profile of PF in these patients was significantly proinflammatory, containing high levels of inflammatory cytokines ( Supplemental Fig. 2 ). To determine whether there was an association between cytokine levels in different tissues, we performed a correlation analysis to compare cytokine levels in plasma, tissue, and PF ( Supplemental Table 3 , available online at www.fertstert.org ). We found significant correlations between plasma and ectopic tissue. We found a positive correlation when comparing VEGF levels in presurgery plasma versus ectopic tissue (0.536; P =.0219). When comparing 2-week postsurgery plasma versus ectopic tissue, we found a positive correlation for IL-5 (0.52; P =.0269). When comparing 3-month postsurgery plasma versus ectopic tissue, we found a positive correlation for IP-10 (0.478; P =.0451) and G-CSF (0.499; P =.0352) and a negative correlation for GM-CSF (−0.529; P =.024). Finally, when comparing eutopic versus ectopic tissue, we found a significant negative correlation for IL-4 (−0.658; P =.00301).

This study is the first to report that the excision of endometriotic lesions produces a measurable systemic shift in the inflammatory cytokine profile of patients with endometriosis. Inflammation is one of the defining features of endometriosis and has been shown to be a major driving factor for the condition. Our findings provide striking evidence that endometriotic lesions are not simply a result of the condition, but rather act as major regulators and promoters of inflammation, thus aiding the progress of endometriosis at a local and systemic level. Lesion removal resulted in a measurable shift in cytokine levels that extended beyond the local pelvic environment. All except two of the plasma cytokines exhibited a trend of decrease after surgery. Although only four were statistically significant, this trend further supports the idea that endometriotic lesions are the major drivers of inflammation in these patients. Importantly, at 3 months after surgery, cytokine levels seemed to begin to return to their initial levels. The fact that the change was transient suggests that either surgery was unable to remove all traces of the lesions or that additional factors are involved in the onset of endometriosis. In the pooled plasma analysis, we detected a significant decrease in circulating levels of GM-CSF, IL-2, IL-8, and IL-10 after removal of lesions, which suggests that in these patients, systemic inflammation is driven locally by the ectopic lesions through several mechanisms. IL-10 is mostly known as a key regulatory cytokine essential for the dampening and resolving of inflammation during the immune response. Our findings support previous reports that IL-10 is indeed overexpressed in the serum endometriosis patients ( 20 ) and that it is capable of promoting lesion growth by locally suppressing the immune cells that would help clear the endometrial fragments ( 13 , 20 ). Because IL-10 can inhibit IL-2 and GM-CSF, the increased levels of IL-10 could reflect an attempt to inhibit the high levels of IL-2 and GM-CSF observed in these patients. IL-2 is normally secreted by activated T cells and natural killer cells and is an important regulator of self tolerance ( 21 ). The increased expression of IL-2 in plasma may point to the inability of these patients to regulate inflammation and could indicate a predisposition to autoimmune characteristics. GM-CSF is usually secreted in response to other inflammatory cytokines, such as IL-1, and seems to act locally, mediating the recruitment and activation of macrophage-lineage cells ( 22 ). We speculate that the high circulating levels of GM-CSF observed in these patients simply reflect an inability to regulate the inflammation at the local level, as supported by the immediate postsurgery decrease in circulating GM-CSF levels. Furthermore, the decrease in GM-CSF levels after surgery was significant in advanced-stage patients and not in early-stage patients. Given that GM-CSF is mostly locally acting, lesions that are larger and more established would be necessary for circulating levels to be influenced. Additionally, ectopic tissue was found to express higher levels of GM-CSF than eutopic tissue, further indicating that ectopic lesions are driving the aberrant expression of this cytokine. Meanwhile, IL-8 is a chemotactic and angiogenic factor previously shown to contribute to neutrophil recruitment and blood vessel formation in endometriotic lesions ( 23 , 24 ). High levels of IL-8 have previously been detected in the PF and serum of women with endometriosis ( 25 , 26 ), which supports our findings. However, the fact that IL-8 levels decreased immediately after surgery suggest that the ectopic lesions were the main drivers of these up-regulated levels. IL-8 is also a potent chemotactic factor for basophils and GM-CSF–primed eosinophils ( 27 ). Given the high expression of GM-CSF in both plasma and ectopic tissue, IL-8 would aggravate immune infiltration at the lesion sites. Endometrial cells have been shown to produce IL-8 and to have increased expression of IL-8 and cyclooxygenase (COX) 2 under stimulation of IL-17 in vitro ( 28 ), meaning they can stimulate local blood vessel formation. Our group recently confirmed that IL-17A is expressed in the eutopic and ectopic endometrium of women with endometriosis, and that IL-17A stimulation of cultured endometrial cells induces the production of VEGF, G-CSF, PDGF-AA, and stromal cell–derived factor 1 ( 23 ), which could be a mechanism by which ectopic lesions increase IL-8 production. To further support the impact of ectopic lesions on systemic inflammation, we detected lower levels of RANTES in the plasma of early-stage patients compared with advanced-stage patients. RANTES is a chemokine involved in the activation and recruitment of immune cells to sites of allergic reaction ( 29 ). Given that RANTES is a locally acting chemokine that is also involved in late-phase inflammatory response, we speculate that it may play a more active role in advanced stages of disease, where it would be produced by the larger ectopic lesions and promote immune cell infiltration. Elevated levels of RANTES have been previously detected in the PF and eutopic and ectopic tissues of women with endometriosis ( 30 , 31 ), and its production can be induced by IL-1β ( 32 ). Our analysis of eutopic and eutopic tissue indicates that IL-1 expression is dysregulated at the lesion sites, further supporting previous reports. The combined effect of these factors is a local proangiogenic microenvironment aggravated by unresolved inflammation that leads to a systemic inflammatory response. We also found these results to be supported when we compared plasma from endometriosis patients with that from normal women. Of the 19 cytokines that were statistically significant, 18 were higher in endometriosis patients compared with normal women, including IL-2, IL-8, IL-10, and RANTES. These results indicate that these cytokines are not only dysregulated in endometriosis patients, but also are expressed aberrantly compared with normal subjects. The remaining cytokines have been shown in previous reports to be dysregulated in women with endometriosis, as extensively reviewed elsewhere ( 33 , 34 ). In our analysis of matched eutopic and ectopic tissue, we found differential expression of nine cytokines, bFGF, IP-10, IL-1Ra, G-CSF, MIP-1β, IL-7, IL-5, GM-CSF, and IL 15, which further supports the idea that this local aberrant cytokine profile is likely driving the systemic inflammation via a positive feedback cycle. Most of these cytokines were more highly expressed in the eutopic than ectopic endometrium. The only exceptions were GM-CSF and IL-15, which were more highly expressed in ectopic tissue. As we previously mentioned, GM-CSF may contribute to the increased inflammatory response, particularly by inducing the differentiation of granulocytes and macrophages or their proliferation at lesion sites. Conversely, IL-1Ra was more highly expressed in eutopic than in ectopic endometrium. Furthermore, after data stratification, we found that this difference was most significant when comparing matched tissue in early-stage patients, not in advanced-stage patients, which may reflect local dysregulation of IL-1 signaling and an inability to control immunerelated inflammation at an early stage. The IL-1 family, a known key regulator of the inflammatory response, has been found to be aberrantly expressed in women with endometriosis. Active endometriotic lesions seem to have increased expression of the IL-1 type I receptor ( 35 ), whereas expression of the decoy IL-1 type II receptor is decreased in the normal endometrium of endometriosis patients compared with normal women ( 36 ). These results point to a severely impaired regulation of the proinflammatory signaling of IL-1 in women with endometriosis. We detected higher levels of IL-1β, IL-1Ra, and IL-6 in endometriosis patients than normal women, which supports the notion that there is dysregulation of IL-1 signaling in these patients. For example, IL-1 can induce secretion of IL-6 by fibroblasts, endothelial cells, and circulating monocytes ( 37 ) and can stimulate IL-8 secretion by endometriotic cells ( 24 ). The decreased expression of IL-1Ra in ectopic tissue would leave IL-1 unregulated and able to promote IL-8 production, as was found in the plasma of patients with endometriosis in this study. IL-6 has been shown to stimulate the secretion of VEGF in isolated peritoneal macrophages and neutrophils from mice ( 38 ) and human cell lines ( 39 ). Also, high levels of IL-6 have been found in the PF ( 40 ) and both eutopic and ectopic endometrium of endometriosis patients compared with normal women ( 41 ). The latter study ( 41 ) also reported higher levels of IL-1β in the endometrium of these patients, which likely further induces local IL-6 secretion. IL-1β has also been shown to induce neutrophil migration to the peritoneal cavity of rats, and such migration was dependent on macrophages and mast cells ( 11 ). Local inflammation is likely sustained by the immune infiltration observed in the pelvic environment of endometriosis patients. Inflammation can be further aggravated by local production of hormones promoted by aberrant expression of cytokines. For example, TNF-α was one of the cytokines we found to be more highly expressed in endometriosis patients compared with normal subjects. IL-1 and TNF-α are able to induce the expression of COX-2, which mainly produces prostaglandin E2 (PGE2). PGE2 is not only capable of self-induction via up-regulation of COX-2, but can induce aromatase P450 production, which is the key regulatory enzyme in the biosynthesis of estrogens ( 8 ). COX-2 has been shown to be overexpressed in the ectopic tissue of patients with endometriosis ( 42 , 43 ). Furthermore, IL-1Ra, which blocks the IL-1–induced production of COX-2 in endometrial cells ( 44 ), is down-regulated in the PF of women with endometriosis ( 45 ), which is consistent with our findings that IL-1Ra was down-regulated in ectopic compared with eutopic tissue. PGE2 thus provides a mechanism for lesions to locally sustain the production of prostanoids and estrogens. PGE2 is also capable of altering the phagocytic ability of immune cells and of promoting angiogenesis via VEGF expression ( 8 ), further contributing to the survival of lesions in the peritoneal cavity. Meanwhile, IL-5 was one of the cytokines more highly expressed in eutopic versus ectopic tissue, particularly in early-stage patients after data stratification. IL-5 plays a major role in the late-phase allergic response, being produced by mast cells within allergen-challenged tissues ( 46 ). Based on our results, we could speculate that IL-5 plays a major role in promoting the local inflammation seen in endometriosis patients, and may play a more important role at earlier stages of disease when lesions are in the process of establishing. MIP-1β is an inflammatory chemokine that is stimulated by IL-7. Both were more highly expressed in the eutopic endometrium compared with paired ectopic endometrium. IL-7 has been found to be an important immunostimulatory cytokine and to be overexpressed in inflamed tissues of patients with autoimmune diseases such as rheumatic disease ( 47 ). IL-7, RANTES, IL-2, and IL-10, all of which were up-regulated in plasma, are cytokines associated with autoimmune conditions and may provide further support for the concept of endometriosis being an autoimmune condition. Finally, the PF of endometriosis patients in this study expressed high levels of various cytokines, which mostly supports previous reports. The most highly expressed cytokines included Eotaxin, bFGF, G-CSF, GM-CSF, IFN-γ, IP-10, MCP-1, VEGF, IL-1Ra, IL-5, IL-10, and IL-15. Our stage comparisons of IL-5 and IL-12 were marginally significant, so further analyses would be required to understand the true significance of the cytokine profile in the PF of these patients.

We provide new insights on the causes of a heightened inflammatory response at local and systemic levels in endometriosis patients. Evidence presented in this manuscript clearly indicates that endometriotic lesions are probably the major drivers of inflammation, because the removal of lesions led to significant reduction in the systemic level of several potent inflammatory factors. Furthermore, most of the cytokines that were dysregulated in plasma or tissue of endometriosis patients were also aberrantly expressed compared with normal subjects. Another important piece of evidence emerging from this study is the striking difference in the immune microenvironment of eutopic and ectopic tissues, because measurable differences were detected when comparing ectopic versus eutopic tissue, as well as the influence of different stages of disease. Although patients recruited in this study were free of any form of therapy for 3 months before surgery to avoid any treatment bias on the cytokine profiles, patients were not tracked specifically for any nonsteroidal antiinflammatory drug (NSAID) use, which should be taken into consideration in future studies. Nevertheless, our results strongly suggest that in endometriosis, an impaired ability to dampen the inflammatory response triggered by the endometrial fragments helps the condition to evolve into a chronic condition. Although our study does not conclusively establish whether inflammation is initiated by the lesions or vice versa, it provides a solid basis that surgical intervention might provide a window of opportunity to dampen inflammation, which would allow for therapeutic interventions, such as reducing subfertility and modulating conditions that permit lesion recurrence.

Ethics approval for this study was obtained from the Greenville Health System (South Carolina) and the University of North Carolina (Chapel Hill), and written informed consent from every subject was obtained before sample collection and storage (IRB protocol no. Pro00000993). Samples were obtained from 43 patients with visually diagnosed endometriosis undergoing resection of lesions because of infertility or pain at Greenville Hospital. Patients from whom we obtained plasma samples (n = 19) were further subgrouped into early (stages I and II; n = 14) and advanced (stages III and IV; n = 5) disease categories. Subjects with endometriosis were recruited and consented at the preoperative visit. Subjects were recruited if they were undergoing diagnostic laparoscopy and included if endometriosis was clinically identified at the time of surgery. All of the endometriosis patients included in this study underwent surgery. Stage of endometriosis was assigned by the surgeon based on the revised American Society for Reproductive Medicine criteria ( 19 ). Blood was drawn before surgery for collection of plasma in an EDTA-containing vacuum tube. Endometrial biopsy was performed after uterine lavage and cervical brushings. Laparoscopy was performed as part of the diagnostic evaluation for infertility and/or pelvic pain. Before resection or any surgical ablation or hysteroscopy, PF was obtained with the use of a cannula and syringe. At the 2-week postoperative visit and at 3 months, venous blood sampling was repeated. Women on oral contraceptives or injectable progestins were excluded from the study. Only women of reproductive age without hormone treatment in the preceding 3 months were included. We also obtained plasma samples from 25 apparently healthy normally cycling women volunteers aged 20–36 years from the University of North Carolina School of Medicine. They were recruited by advertising in the local media and screened according to study guidelines. Clinical information was obtained with the use of a standardized questionnaire reviewed by the study coordinator. None of the healthy volunteers had signs or symptoms of endometriosis or infertility-related problems. Eutopic and ectopic samples obtained from endometriosis patients were used for total protein extraction. Briefly, tissue samples were placed in 1.5-mL microcentrifuge tubes containing 200 µL phosphate buffered saline solution. The protease inhibitor aprotonin was added to each tube (2 µL; Sigma Aldrich). The tissue was homogenized with the use of a rotorstator homogenizer on ice. The samples were then centrifuged at 4°C and the supernate collected. The samples were diluted to a protein concentration of 5 µg/µL with the use of DNAse/RNAse–free water and subsequently stored at −80°C. Protein concentrations were determined with the use of a protein assay (Bio-Rad). A single panel of a commercial multiplex assay (Luminex xMAP laser bead technology) available from Bio-Rad was used for analysis of selected cytokines/chemokines involved in the inflammatory and angiogenic pathways: platelet-derived growth factor (PDGF) BB, Eotaxin, basic fibroblast growth factor (bFGF), granulocyte colony–stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), interferon (IFN) γ, interferon-inducible protein (IP) 10, monocyte chemotactic protein (MCP) 1, macrophage inflammatory protein (MIP) 1α, MIP-1β, Regulated on activation, normal T expressed and secreted (RANTES), tumor necrosis factor (TNF) α, vascular endothelial growth factor (VEGF), interleukin (IL) 1β, IL-1 receptor antagonist (IL-1Ra), IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12, IL-13, and IL-15. Briefly, color-coded polystyrene beads were coupled with capture antibodies for each respective target cytokine. Plasma samples were aliquotted into 15 µL and diluted four times with the use of the provided sample diluent. Diluted plasma and protein samples were preincubated with 50 µL antibodies conjugated to magnetic beads in a 96-well plate. The plate was covered and incubated in the dark at room temperature on a shaker for 30 minutes. The plate was washed three times with the wash buffer provided with the kit. After diluting 25 µL detection antibody in detection antibody diluent, the plate was incubated in the dark at room temperature on a shaker for 30 minutes. The plate was washed three times in wash buffer, and 50 µL streptavidin-phycoerythrin fluorescent conjugate was added to the wells. The plate was incubated for 10 minutes before being washed, and then a Bio-Plex 200 System was used to read it. The standards for this assay were provided by the manufacturer as a lyophilized cocktail of proteins and were run for each individual target analyte. The cocktail was then serially diluted and standard curve dilutions were run within the assay. Increasing fluorescent intensity signals are read as fluorescence intensity values, which are in direct proportion to protein bound to the specific analyte bead population. Observed concentration for each target analyte was calculated against standard curve regression, and values were recorded in units of pg/mL. Cytokines were quantified through the detection of conjugate, which is in direct proportion to the amount of the target analyte. Statistical analysis was performed with the use of Sigma Stat 3.5. Plasma samples from endometriosis patients (n = 19) were pooled as well as further subgrouped into early (stages I and II; n = 14) and advanced (stages III and IV; n = 5) disease categories. Plasma samples from normal subjects (n = 25) were pooled as well as subgrouped by menstrual cycle phase into secretory phase (n = 9) and proliferative phase (n = 9). For the subgrouping, we excluded seven samples for which we did not know the menstrual cycle phase. Plasma data from endometriosis patients were analyzed with the use of one-way repeated-measures analysis of variance (ANOVA) and Tukey post hoc test, or with the use of Friedman ANOVA on ranks for nonparametric data. Paired t test or Wilcoxon signed rank test was used to analyze paired tissue data. Plasma data comparing endometriosis and normal subjects were analyzed with the use of Kruskal-Wallis one-way ANOVA and Dunn post hoc test. Peritoneal fluid data were analyzed with the use of one-way ANOVA or Friedman ANOVA on ranks. A P value of <.05 was considered to be significant.