{"paper_id":"ff868c43-2f4c-478b-95c4-11e60b9b57b3","body_text":"In the 20th century, Ivo Brosens focused on the crucial issue of bleeding, which he\nconsidered to be at the core of the pathogenesis of endometriosis, with a simple,\ndefinitive, and catchy title ( Brosens, 1997 ).\nIvo had the rare gift of scientific synthesis, as when he defined the endometrioma as an\nextraovarian pseudocyst ( Brosens  et\nal. , 1996 ) or an extraovarian haematoma ( Brosens  et al. , 1994 ), or when he first drew\nattention to the then understudied, but highly risky, haemorrhagic consequence of deep\ninfiltrating endometriosis (DIE), i.e. spontaneous haemoperitoneum in pregnancy ( Brosens  et al. , 2009 ).\nBleeding now returns as a  leitmotif  of the endometriosis trajectory and\nderailment, in the role of a potential precursor of DIE. In a study published in this issue\nof  Human Reproduction Open ,  Chaggar\n et al.  (2024)  prospectively recruited and followed 51\npremenopausal non-pregnant women who were consecutively evaluated for acute lower abdominal\npain. Within 6 months of the acute episode, 7/15 (47%) women who presented with\nhaemoperitoneum, i.e. the presence of blood clots and echogenic fluid in the peritoneal\ncavity, developed sonographic evidence of DIE, compared with 0/36 of those without\nhaemoperitoneum. Interestingly, a haemorrhagic corpus luteum was identified as the origin of\nthe intra-abdominal haemorrhage in 13/15 (87%) cases. As the authors are also well known for\ntheir exceptional expertise in ultrasonography, the possibility of a diagnostic bias can be\nreasonably ruled out.\nNone of the women in the haemoperitoneum group were using combined oral contraceptives or\nprogestogen-only pills compared with 10/36 (28%) in the non-haemoperitoneum group. Three\nwomen in the first group were using a levonorgestrel-releasing intrauterine device\n(LNG-IUD). However, the LNG-IUD does not inhibit ovulation for more than a few months after\ninsertion ( Guillebaud, 2003 ). Thus, it is\nreasonable to assume that ovulation occurred in all 15 patients presenting with\nhaemoperitoneum.\nThis cohort study follows the publication of a previous prospective pilot study by the same\nresearch group.  Bean  et al. \n(2019)  followed 35 women with severe acute lower abdominal pain and detected\n de novo  DIE at ultrasound follow-up in 4/6 (67%) of patients with\nevidence of haemoperitoneum compared with 1/29 (3%) of those without haemoperitoneum at\nbaseline assessment. Seven of the eight women who presented with intra-abdominal bleeding\nhad a haemorrhagic functional ovarian cyst. Two of these eight patients did not complete the\nfollow-up.\nPooling data from the above studies, 11/21 (52%; 95% CI, 32–72%) patients presenting with\nhaemoperitoneum-associated lower abdominal pain developed DIE within a few months of the\nacute episode compared with 1/65 (2%; 95% CI, 0.1–9%) of those without haemoperitoneum\n(relative risk, 34; 95% CI, 5–248). In addition, a bleeding functional ovarian cyst was the\ncause of haemoperitoneum in 19/23 (82.6%; 95% CI, 62–94%) patients with sonographic\nevaluation at presentation. To conclude, spontaneous haemoperitoneum not in pregnancy\n(SHniP) is very often associated with a haemorrhagic corpus luteum. Moreover, the\nassociation between SHniP and subsequent development of DIE is so strong that it can\nreasonably be considered causal.\nThe demonstration that substantial intraperitoneal bleeding is a precursor of infiltrating,\nfibrotic endometriotic lesions is extremely interesting and may open up a completely new\nview of the natural history of endometriosis. However, we should not mistake the finger for\nthe moon here, because the most important information seems to be the central pathogenic\nrole of haemorrhagic corpora lutea in the progression of endometriosis to advanced forms. In\nother words, the real precursor of DIE may not be haemoperitoneum  per se ,\nbut what caused the haemoperitoneum, i.e. a haemorrhagic corpus luteum. If this is true,\nthen haemoperitoneum acts as a common thread between the source of bleeding and the\ndevelopment of DIE, and what remains to be clarified are the biomolecular mechanisms along\nthis pathway.\nBut there is a question positioned at an even higher pathogenic level: why is there such a\nhigh incidence of haemorrhagic corpora lutea in women who go on to develop DIE? The\nassociation between cystic corpora lutea and DIE was recently reported by the same research\ngroup also in a large prospective cohort study of more than 1000 women attending a general\ngynaecology clinic ( Chaggar  et al. ,\n2023 ).\nThe question may prove to be crucial, especially considering that haemorrhagic corpora\nlutea are the precursor not only of DIE, but also of many ovarian endometriomas. We followed\n109 patients who did not use postoperative hormonal treatments for 2 years after surgical\nremoval of endometriomas. A total of 27 (25%) participants developed a cyst recurrence,\nwhich in 11 (41%) cases was preceded by a haemorrhagic corpus luteum ( Vercellini  et al. , 2009 ). The central pathogenic\nrole of ovulation in endometrioma formation is confirmed by the impressive protective effect\nof postoperative ovulation suppression after laparoscopic cyst removal ( Zakhari  et al. , 2021 ;  Chiu  et al. , 2022 ).\n\nThe answer to the question of what causes such a high frequency of haemorrhagic corpora\nlutea, leading to the development of DIE and endometriomas through heavy bleeding, may lie\nin the nature of endometrioma formation.\nSince the original study by  Hughesdon\n(1957) , it has been accepted that endometriomas are atypical cysts, in that they are\n‘on’ the ovary rather than ‘in’ it, like most other cysts (e.g. serous, mucinous, and\ndermoid cysts). In other words, the ovary invaginates and duplicates so that the cortex\nitself constitutes the so-called pseudocapsule, which surrounds and contains the typical\nchocolate-like fluid consisting of old blood and siderophages ( Brosens  et al. , 1994 ,  1996 ). The first step in this process is the adhesion of the\nlateral gonadal aspect to the pelvic sidewall. Such adhesion is caused by inflammation\noriginating from superficial endometriotic implants located on the peritoneum of the ovarian\nfossa and the posterior leaf of the broad ligament ( Vercellini  et al. , 2009 ). Indeed, according to  Hughesdon (1957)  and  Brosens  et al.  (1994 ,  1996 ), superficial endometriotic implants are usually observed in\ncorrespondence with the site of ovarian inversion, whereas they are only irregularly found\nlining the inner wall of the pseudocyst.\nThe next question might be: ‘Where does the tarry, thick fluid content of endometriomas\ncome from?’ The most plausible answer is not from the accumulation of menstrual debris\nsecondary to the shedding and bleeding of endometriotic implants, but more likely from acute\nand heavy bleeding associated with ovulation. The strong relationship with haemorrhagic\ncorpora lutea and the protective effect of ovulation suppression would be difficult to\nexplain otherwise. When the ovary adheres to the pelvic side wall, blood from a corpus\nluteum is trapped and acts as an ‘invagination head’ causing ovarian duplication ( Vercellini  et al. , 2009 ).\nIf the establishment of peritoneal endometriosis on the pelvic sidewall is a prerequisite\nfor endometrioma formation, a further question would be whether the superficial implants\nsimply behave as innocent bystanders or may play an active role in ‘guiding’ the site of\novulation and also determining the haemorrhagic nature of what would otherwise be limited\novulatory bleeding.\n\nOvulation involves inflammation and cytolysis. From a biomolecular perspective, superficial\nperitoneal implants adhering to the ovarian cortex may influence the site of ovulation\nthrough the surrounding secondary inflammatory microenvironment. Of relevance, cytokines\nsuch as interleukin-1 and -6, and tumour necrosis factor play a critical role in both\nendometriosis-associated inflammation and the ovulatory process ( Vercellini  et al. , 2014 ;  Zondervan  et al. , 2018 ;  Kotlyar  et al. , 2019 ). Thus, follicular rupture\nmay occur causally, and non-casually, in correspondence with the site of superficial\nendometriotic implants.\nMoreover, both endometrial and endometriotic cells release potent fibrinolytic molecules.\nEndometrial fibrinolytic activity effectively prevents intrauterine clot formation during\nnormal menstrual flow. This also results in platelet deactivation and facilitates the\ntranscervical expulsion of sloughed endometrium and blood ( Davies and Kadir, 2012 ). Fibrinolysis also prevents scarring\nduring endometrial repair. Indeed, the repeated tissue injury and repair process that occurs\nduring the perimenstrual phase is uniquely characterized by a lack of scarring ( Critchley and Maybin, 2011 ;  Critchley  et al. , 2020 ).\nFibrinolysis interferes with thrombus formation via the conversion of plasminogen to\nplasmin induced by tissue plasminogen activator (tPA) and urokinase plasminogen activator\n(uPA) enzymes. The endometrium is a rich source of tPA and uPA ( Critchley and Maybin, 2011 ;  Critchley  et al. , 2020 ). Importantly,\nendometrial concentrations of plasminogen activators increase during the proliferative phase\nand peak at mid-cycle, i.e. just during the ovulatory phase ( Davies and Kadir, 2012 ). Thus, in addition to driving the site of\novulation, superficial endometriotic implants may create a pro-haemorrhagic environment\naround the ovulatory stigma.\nHigher levels of uPA have been observed in the endometrium of women with endometriosis.\nAccording to some researchers, this would increase the capacity of refluxed endometrial\nfragments to implant in the pelvis due to an increased potential to degrade the\nextracellular matrix ( Zorio  et al. ,\n2008 ). At the same time, this would also increase the local pro-haemorrhagic\npotential of endometrial fragments once ectopically implanted. Indeed, patients with\nendometriosis generally experience heavy menstrual bleeding (HMB) ( Parazzini  et al. , 2017 ;  Shafrir  et al. , 2018 ). The endometrium of women\nwith HMB is characterized by a high production of proteolytic enzymes, such as matrix\nmetalloproteinases, and fibrinolytic activity ( Ferenczy, 2003 ), and synthesizes more vasodilatory PGE 2  than\nvasoconstrictive PGF2α. In addition, the endometrium of women with HMB releases higher than\nnormal amounts of prostacyclin, known to inhibit platelet aggregation and generate\nvasodilatation, and shows reduced expression of the potent vasoconstrictor endothelin-1 and\nincreased expression of neural endopeptidase, its metabolizing enzyme ( Livingstone and Fraser, 2002 ;  Critchley and Maybin, 2011 ;  Davies and Kadir, 2012 ;  Critchley  et al. , 2020 ;  Jain  et al. , 2022 ). A local decrease in vascular impedance may\nincrease blood loss.\nThus, it can be further hypothesized that peri-ovarian superficial endometriotic implants,\nin addition to determining the site of ovulation, may also promote excessive bleeding,\nultimately leading to the formation of a haemorrhagic corpus luteum.\n\nChaggar and co-workers pointed out the possible presence of undiagnosed superficial\nperitoneal endometriosis as a limitation of their study, also because it is impossible to\nknow ‘ how this could have contributed to the development of deep\nendometriosis ’. They also interpreted their findings as the result of peritoneal\nhealing occurring over a blood clot, so that endometrial cells present in the peritoneal\nfluid, regardless of the presence of endometriosis, ‘ become trapped underneath the\nperitoneal surface and trigger the development of deep disease ’ ( Chaggar  et al. , 2024 ).\nAlthough this is certainly possible, it cannot be excluded that no deep endometriosis would\nhave developed in the absence of superficial implants, since according to the blood-based\npathogenic hypothesis described above, ovarian haemorrhage from corpora lutea is provoked\nexactly by the local pro-haemorrhagic environment induced by peri-ovarian peritoneal\nendometriosis.\nSupposedly, most women experience retrograde menstruation, some develop limited peritoneal\nimplants, while only a minority eventually progress to advanced endometriosis. If the\ntranstubal menstrual reflux theory is correct, the necessary condition for the onset of all\nendometriotic lesions is the co-presence of endometrial cells and blood. In light of the\nfindings of  Bean  et al. \n(2019) ,  Chaggar  et al. \n(2024) , and  Vercellini  et\nal.  (2009) , it appears that the amount of intra-pelvic blood makes the\ndifference between the establishment of only superficial peritoneal lesions or the\nprogression to deep lesions and endometriomas. In other words, superficial peritoneal\nendometriosis would be an early, micro-haemorrhagic form, because retrograde menstruation\nallows only a limited amount of blood to reach the pelvis, whereas deep lesions and\nendometriomas would be advanced, macro-haemorrhagic forms, because bleeding from a corpus\nluteum can lead to haemoperitoneum. If dense adhesions between the ovary and the pelvic\nsidewall prevent blood pouring from a corpus luteum from escaping, an endometrioma may form,\nwhereas if blood can flow freely towards the most dependent part of the pelvis, a deep\nlesion may form. If the blood distribution is mixed, both types of lesions may develop\nsimultaneously. In all cases, the blood, not the endometrium, would be the main pelvic\naggressor.\nIndeed, research efforts have focused more on the endometrial rather than on the blood\ncomponent of retrograde menstruation. Nevertheless, several studies, notably by the group of\nJacques Donnez, have demonstrated an iron overload in the pelvis of patients with\nendometriosis ( Lousse  et al. ,\n2009 ,  2012 ;  Defrère  et al. , 2011 ;  Donnez  et al. , 2016 ). In extreme synthesis,\npelvic macrophages internalize refluxed erythrocytes and participate in the process of iron\nrecycling. However, when an excess of blood overwhelms the scavenging capacity of pelvic\nmacrophages, non-protein-bound, catalytic iron is released into the peritoneal fluid. This\ngenerates reactive oxygen species (ROS), which in turn damage the fragile mesothelial\nlining, expose the sub-mesothelial connective tissue, facilitate implantation of\nregurgitated endometrial fragments, and promote neo-angiogenesis and endometrial cell\nproliferation through various molecular and cellular factors ( Ng  et al. , 2020 ;  Wyatt  et al. , 2023 ;  Ni and Li, 2024 ).\nThe above process may explain the development of superficial peritoneal endometriosis when\na limited amount of blood is present in the pelvis (micro-haemorrhagic endometriosis).  Chaggar  et al.  (2024)  rightly\nsuggest that the same mechanism is at work when haemoperitoneum develops because, in\naddition to causing oxidative stress-induced mesothelial lining breakdown of vast peritoneal\nareas, a large amount of blood leads to platelet activation and favours\nepithelial–mesenchymal transition and fibroblast–myofibroblast transdifferentiation. Indeed,\nthe presence of clots is evidence that fibrinolysis is inadequate and that fibrin bridges\nare formed between adjacent structures. This is the beginning of an inter-organ adhesion\nprocess that leads to the burial of what were once superficial endometriotic implants, and\nto fibrogenesis, which is the hallmark of DIE ( Guo\n2018 ;  Vigano  et al. ,\n2018 ;  Viganò  et al. ,\n2020 ;  Garcia Garcia  et\nal. , 2023 ). Thus, what  Chaggar\n et al.  (2024)  define as ‘ peritoneal healing over a blood\nclot ’ may instead be the result of adhesion between organs whose mesothelial\ncovering has been damaged by catalytic iron-induced oxidative stress. For example, Douglas\npouch DIE may be interpreted as ex-superficial peritoneal endometriosis buried by adhesions\nbetween the anterior rectal and posterior uterine aspects.\nThe mechanisms underlying mesothelial injury, post-traumatic peritoneal repair, the role of\nblood, ROS, fibrinolysis, platelet and macrophage activation, and fibroblast growth in\ndriving the intra-abdominal adhesion formation and fibrotic process have been well reviewed\nby  Koninckx  et al.  (2016) .\nThe fundamental role of intra-abdominal trauma in the pathogenesis of endometriosis has been\nrepeatedly suggested by  Canis  et al. \n(2017  and  2020 ).\n\nThe working hypothesis of the blood-based pathogenesis of endometriosis is shown in  Fig. 1 . In synthesis, intraperitoneal blood may\npave the way for the ectopic implantation of refluxed endometrial cells by inducing\noxidative stress-mediated damage to the mesothelium. In this micro-haemorrhagic phase, the\namount of blood is not sufficient to injure large areas of mesothelium, and indeed,\nextensive adhesions are not typically associated with limited superficial peritoneal\nendometriosis in the pelvis. However, because the lateral aspect of the ovary is juxtaposed\nto the pelvic sidewall exactly at the site of repeated monthly blood and endometrial efflux,\nthe development of superficial lesions on the peritoneum of the ovarian fossa with secondary\ngonadal adhesion is common. The perilesional inflammation directs the rupture of the\nmaturing follicle over the ectopic endometrium and creates a local pro-haemorrhagic\nenvironment that increases the risk of heavy bleeding from the corpus luteum.\nA working hypothesis on the blood-based pathogenesis and pathophysiology of\nendometriosis.  The definition and progression of each step are based on\ninformation from  Bean  et al. \n(2019) ,  Brosens  et al. \n(1996) ,  Canis  et al. \n(2017) ,  Chaggar  et al. \n(2024) ,  Defrère  et al. \n(2011) ,  Donnez  et al. \n(2016) ,  Garcia Garcia  et\nal.  (2023) ,  Guo\n(2018) ,  Jain  et al. \n(2022) ,  Koninckx  et\nal.  (2016) ,  Hughesdon\n(1957) ,  Lousse  et al. \n(2009) ,  Ng  et al. \n(2020) ,  Vercellini  et\nal.  (2009) ,  Viganò\n et al . (2020) , and  Wyatt  et al.  (2023) .\nIn the case of blood entrapment, an endometrioma (extraovarian haematoma) may develop,\nwhereas in the case of blood flowing freely towards the pelvis due to gravitational effects,\nclots will form and accumulate in the most dependent pelvic pouches, leading to DIE\nformation. This macro-haemorrhagic phase may be positively associated with the total amount\nof ovarian bleeding, which may therefore act as a rate-limiting step in the progression from\nearly to extensive forms of endometriosis. Only a fraction of women with mostly undiagnosed\nsuperficial peritoneal endometriosis will develop the usually diagnosed endometriomas and\ninfiltrating fibrotic lesions. This may partly explain the apparent discrepancy between the\nsupposedly universal phenomenon of retrograde menstruation and the relatively low prevalence\nof diagnosed endometriosis. Most women may harbour micro-haemorrhagic endometriosis, albeit\ntransiently, whereas few women would develop stable macro-haemorrhagic endometriosis.\nOur pathogenic hypothesis is based on limited data, is highly speculative, does not\nconsider many potential additional contributing causes, undoubtedly appears simplistic, and\nmay not explain the onset of disease in many patients. There is no evidence on the real\nincidence of haemorrhagic corpora lutea in the general population and in the subgroup of\nindividuals with established superficial peritoneal endometriosis, nor on the likelihood of\ndeveloping deep lesions and endometriomas without transition through exposure to large\namounts of blood. It is not possible to infer, even indirectly, the minimum volume of blood\nrequired to trigger the development of DIE and endometriomas, and it cannot be excluded that\namounts smaller than those observed by  Chaggar\n et al.  (2024)  may be enough to prompt the process. In other\nwords, blood may act as the pivot of endometriosis progression even in the absence of easily\nidentifiable moderate or severe haemoperitoneum causing acute lower abdominal pain.\nWhat is now clear, thanks to the report by Chaggar and colleagues, is that refluxed\nendometrium alone is unlikely to be sufficient for the development of advanced\nendometriosis. The authors are to be commended for an extremely interesting and innovative\nstudy that may open a new window into the still-undefined pathogenesis of endometriosis. At\nthe very least, the findings of  Chaggar  et\nal.  (2024)  provide a proof of concept that should now be confirmed by\nlarger studies conducted by independent research groups. What better response to the ‘call\nfor new theories’ proposed by the  Endometriosis\nInitiative Group (2024) ?\n\nP.Ve. conceived the study and drafted the original version of the article. C.E.M.M. and\nP.Vi. contributed to the acquisition of published information. All authors revised\ncritically the drafts of the manuscript and approved its final version. All authors agree to\nbe accountable for all aspects of the work.","source_license":"CC0","license_restricted":false}