{"paper_id":"fa0dbd8d-eae4-40d0-81cc-99f051f0cd55","body_text":"Endometriosis is defined as the presence of endometrium-like epithelium and/or stroma\noutside the uterus, typically associated with an inflammatory process ( International Working Group of AAGL, ESGE, ESHRE and WES  et\nal. , 2021 ). This condition remains a ‘hot topic’ in the public eye, with\nongoing delays in diagnosis ( Ghai  et\nal. , 2020 ) and no recent improvements in the treatment efficacy. It can\ncause a wide range of pelvic pain symptoms and subfertility ( Becker  et al. , 2022 ) which has negative impact on\nwomen’s quality of life (QOL) ( Nnoaham  et\nal. , 2011 ).\nBetter understanding of the aetiology and pathophysiology of endometriosis is required to\ndevelop novel treatments and preventative measures, which are currently lacking. Existing\ntheories include those of retrograde menstruation, coelomic metaplasia, Müllerian remnants,\nbenign metastasis, and the ability of progenitor cells from the bone marrow to differentiate\ninto endometrial tissue ( Burney and Giudice,\n2012 ;  Zubrzycka  et al. ,\n2020 ;  Saunders and Horne, 2021 ).\nDevelopment of this disease is, however, likely to be multifactorial.\nA recent pilot study observed that acute haemoperitoneum, which is managed conservatively,\ncould be a precursor of deep endometriosis ( Bean\n et al. , 2019a ). Over time, blood clots became more organized and\nowing to gravitational effects, settled in dependant parts of the pelvis where the formation\nof deep endometriotic lesions was subsequently seen on an ultrasound scan. This was the\nfirst study to demonstrate the development of deep endometriosis  de novo ,\nwhich is an important finding in helping to identify women at risk and target preventative\nstrategies.\nThe sample size used in this study was, however, small, with only six women with\nhaemoperitoneum who were followed up. Furthermore, there was difficulty in standardizing\nfollow-up for women, as the rates of clot resolution and development of deep endometriosis\nwere unknown when the study was commenced.\nIn the present study, a larger consecutive cohort of premenopausal, non-pregnant, and\nhaemodynamically stable women attending our gynaecology clinic with severe acute lower\nabdominal pain were sonographically observed over a standardized period of time. The aim of\nthis study was to determine whether the presence of significant haemoperitoneum in these\nwomen would lead to  de novo  development of deep endometriosis and\nassociated symptoms.\n\nThis was a single-centre, prospective observational cohort study performed in an acute\ngynaecology unit between August 2019 and March 2022. All women aged 18–50 years who\nconsecutively presented to our clinic with severe acute lower abdominal pain were eligible\nfor the study. Severe acute lower abdominal pain was described as recent onset pain,\nresulting in attendance to the emergency department. Further inclusion criteria were being\nclinically stable, able to tolerate transvaginal ultrasound scan (TVS), and suitable for\nconservative management.\nExclusion criteria were prior history or sonographic evidence of endometriosis on the\ninitial encounter, previous hysterectomy, bilateral oophorectomy, or postmenopausal\nstatus; the latter was described as ≥12 months of amenorrhoea, which was not secondary to\nbreastfeeding, exogenous hormones, or endocrine conditions. We also excluded women who\nwere pregnant, as the presence of significant haemoperitoneum is not suitable for\nconservative management in this group.\nFollowing the initial assessment, women who met the inclusion criteria were approached\nabout joining the study. Those who agreed were asked to sign a consent form and complete\nthe standardized ‘BSGE pelvic pain questionnaire’ ( https://www.bsge.org.uk/history-of-the-endometriosis-centre-project/ ;  Byrne  et al. , 2018 ).\nClinically stable women with sonographic evidence of significant haemoperitoneum, who were\nmanaged expectantly, formed the study group. Women without haemoperitoneum formed the\ncontrol group.\nTVS was repeated after 2 and 6 months for the study group, and after 6 months for the\ncontrol group, unless earlier scans were clinically indicated. These time frames were\nchosen based on the previous pilot study demonstrating that the median time interval\nbetween the initial visit to completion of follow-up for women in the study group was\n159 days (5–6 months) and 119 days (∼4 months) for women in the control group ( Bean  et al. , 2019a ). When pain\nand/or haemoperitoneum persisted, we followed women up until spontaneous resolution of\nsymptoms or medical/surgical intervention was required. All women were asked to complete\nthe ‘BSGE pelvic pain questionnaire’ at each clinic visit.\nThe primary outcome of this study was the sonographically confirmed presence of deep\nendometriosis following the occurrence of significant acute haemoperitoneum.\nSecondary outcomes were the severity of pelvic pain symptoms and health-related QOL\n(HR-QOL) at baseline and 6 months.\nAll eligible women were clinically assessed. A comprehensive demographic and clinical\nhistory was taken. Demographic variables included patient age, BMI, ethnicity, gravidity,\nparity, and smoking status. BMI was calculated using a calibrated scale and stadiometer in\nour clinic. Self-reported history of the following conditions or surgeries was recorded:\nCaesarean section, other abdominal surgery, autoimmune diseases, chronic fatigue syndrome,\nchronic pain syndrome, fibromyalgia, irritable bowel disease, irritable bowel syndrome,\nmigraines, anxiety, and depression.\nWomen were offered a detailed TVS examination per routine practice, which was conducted\nmethodically (see below). The scan was performed using a 7.5-MHz probe (Voluson E8, GE\nMedical Systems, Milwaukee, WI, USA).\nSignificant haemoperitoneum was defined as the presence of blood clots and echogenic\nfluid within the peritoneal cavity. Fresh blood clots are seen as thick, hyperechoic,\ninhomogeneous, and avascular lesions on TVS. They can be distinguished from free fluid and\ncompressed on palpation with the transvaginal probe ( Bean  et al. , 2019a ). When blood and clots were\nonly seen in the pouch of Douglas (POD), the haemoperitoneum was classified as moderate.\nWhen blood and clots were also seen anterior to the uterus, in the uterovesical fold, it\nwas regarded as severe haemoperitoneum ( Rajah\n et al. , 2017 ;  The\nESHRE Working Group on Ectopic Pregnancy  et al. , 2020 ).\nDeep endometriosis was defined as lesions consisting of endometrium-like tissue, present\nin the abdomen, extending on or under the peritoneal surface. The lesions are typically\nnodular, have the ability to invade neighbouring structures, and are associated with\nfibrosis and anatomical distortion ( International\nWorking Group of AAGL, ESGE, ESHRE and WES  et al. , 2021 ).\nThe systematic approach described by The IDEA Group Consensus Statement was used to\nexamine the pelvis for signs of deep pelvic and ovarian endometriosis ( Guerriero  et al. , 2016 ). The\nuterus, adnexa, anterior, and posterior compartments were scanned to identify\nendometriotic nodules, hydro- and haematosalpinges, endometriotic tubo-ovarian complexes,\npelvic adhesions, and ureteric dilatation. Deep endometriotic nodules were diagnosed when\nhypoechoic, avascular, solid lesions were seen on TVS. They could have smooth or irregular\ncontours. They were often tender on palpation with the transvaginal probe and at the point\nof fixation of adjacent organs. The ovaries were then examined. Endometriomas were\ndiagnosed when thick-walled, avascular ovarian cysts containing fluid of ‘ground-glass’\nappearance were observed ( Van Holsbeke  et\nal. , 2010 ). In contrast, fresh haemorrhagic cysts typically displayed\na ‘spider’s web’ appearance ( Okaro and Valentin,\n2004 ). When it was difficult to differentiate between these cyst types, TVS was\nconducted again 6 weeks later, by which time haemorrhagic cysts would have resolved.\nPelvic adhesions and POD obliteration were assessed using the ‘sliding organs sign’ ( Guerriero  et al. , 2016 ) or the\n‘flapping sail sign’ when filmy adhesions were present ( Savelli  et al. , 2004 ).\nThe pelvis was also searched for other gynaecological and non-gynaecological\nabnormalities during the ultrasound examination. Acute pathology, including acute\nappendicitis, acute pelvic inflammatory disease, ovarian hyperstimulation, ovarian\ntorsion, and ureteric calculi, was identified in line with current literature ( Timor-Tritsch  et al. , 1998 ;\n Molander  et al. , 2002 ;\n Nastri  et al. , 2015 ;\n Bean  et al. , 2019b ;  Moro  et al. , 2020 ). Other\nabnormalities were diagnosed according to the following; adenomyosis and fibroids as\ndescribed in the Morphological Uterus Sonographic Assessment (MUSA) Group Consensus\nStatement ( Van den Bosch  et al. ,\n2018 ), cervical and endometrial polyps as per The International Endometrial\nTumour Analysis (IETA) Group Consensus Statement ( Leone  et al. , 2010 ) as well as more current evidence ( Wong  et al. , 2017 ), congenital\nuterine anomalies following the revised American Society for Reproductive Medicine (ASRM)\nclassification for Müllerian anomalies ( Pfeifer\n et al. , 2021 ), accessory cavitated uterine malformation (ACUM)\nand dilated pelvic veins using guidance from recent papers ( Amin  et al. , 2019 ;  Naftalin  et al. , 2020 ), and non-endometriotic\novarian cysts by pattern recognition ( Valentin\n et al. , 2001 ).\nThe kidneys were examined for hydronephrosis and abnormalities, such as renal cysts,\nusing a transabdominal 3.5-MHz ultrasound probe (Voluson E8, GE Medical Systems).\nThe examiners were highly experienced in the ultrasonic diagnosis of endometriosis and\nother gynaecological pathologies (EFSUMB Level 2) ( Education and Practical Standards Committee  et al. , 2006 ),\nhaving worked in a unit with a tertiary endometriosis centre for over 3 years. They were\nsupervised by consultant gynaecologists who were all expert gynaecological ultrasound\nexaminers (EFSUMB Level 3).\nThe British Society of Gynaecological Endoscopy (BSGE) questionnaire was used to measure\nthe severity of pelvic pain symptoms and HR-QOL patients were experiencing ( Byrne  et al. , 2018 ) ( British Society for Gynaecological Endoscopy,\n2024 ). Pelvic pain symptoms assessed included pre-menstrual and menstrual pain,\nnon-cyclical pelvic pain, dyspareunia, menstrual and non-menstrual dyschezia, lower back\npain, dysuria, and difficulty emptying the bladder, measured on an 11-point numerical\nrating scale. Frequency and urgency of bowel movements, sensation of incomplete emptying,\nconstipation, and menstrual haematochezia were graded on a 5-point Likert scale. Hormonal\ncontraceptive usage was assessed from dichotomous data (‘yes’ and ‘no’ for each type of\ntreatment). HR-QOL was measured using the EuroQol 5D-3L (EQ-5D-3L) questionnaire, a\nsimple, generic tool, which is validated for clinical use and incorporated in the BSGE\nquestionnaire ( Rabin and de Charro, 2001 ).\nThe EQ-5D-3L consists of two parts, one which assesses mobility, self-care, daily\nactivities, pain and discomfort, and anxiety and depression through five questions. The\nEQ-5D index score is computed from the responses and ranges from 0 (death) to 1 (perfect\nhealth), The second component comprises a 100-point visual analogue scale referred to as\nthe ‘EQ Visual Analogic Scale’ (EQ-VAS), where the user rates their overall health status,\nwith higher scores describing better health.\nAll clinical and sonographic data were stored on our dedicated clinic database (Viewpoint\nBildverabeitung GmbH, Munich, Germany).\nBased on a previously published study ( Bean\n et al. , 2019a ), ∼50% of women with significant haemoperitoneum\nwere expected to develop deep endometriosis and only 5% of women in the control group.\nFrom this, we calculated that a sample size of 30, with 15 in each group, would be\nrequired, to achieve a power of 80% and CI of 95%.\nNormally distributed data was reported using mean and SD, and non-normally distributed\ndata with median and interquartile range (IQR). Data distribution was determined by\nassessing skewness and kurtosis. The chi-square test or Fisher’s exact test was used to\ncompare categorical variables between groups. Continuous variables were compared between\ngroups using the unpaired Student’s  t -test or Mann–Whitney test,\ndepending on normality of the sample.\nChanges in pain and QOL scores between timepoints for individuals were analysed using the\npaired Student’s  t -test for normally distributed variables or the\nWilcoxon matched pairs test for non-normally distributed variables. The Wilcoxon matched\npairs test or the paired exact test were used to examine changes over time.\nTo assess change of categorical variables over time, regression methods were used, with\nthe outcome variable being the value at 6 months, with the baseline value included as a\ncovariate. By adjusting for the baseline value, this is akin to examining the change over\ntime. Ordinal logistic regression was used for the ordinal outcomes, whilst logistic\nregression was used for the binary variables.  P -values of <0.05 were\ndeemed statistically significant. Statistical calculations were performed via Stata\nversion 15.1 (StataCorp., College Station, TX, USA).\nEthical approval was obtained from the NHS Research Ethics Committee (Reference:\n19/NI/0107) on 21.05.2019.\n\nDuring the study period, 282 eligible women presented to our clinic, of which 59 fulfilled\ninclusion criteria and consented to participate. Fifty-one women attended all follow-up\nappointments and formed the final study sample. The patient flow is presented in  Fig. 1 .\nFlowchart summarizing inclusion, exclusion, and diagnosis of women attending our\nacute gynaecology unit during the study period.  TVS, transvaginal ultrasound;\nTRS, transrectal ultrasound; PID, pelvic inflammatory disease.\nThe demographic and clinical characteristics of the study participants are listed in  Table 1 , and concomitant ultrasound abnormalities\nin  Supplementary Table S1 . There\nwere no statistically significant differences between the study and control groups at their\ninitial visit. The participants’ analgesic use at baseline and 6 months, as well as\nproportions trying for pregnancy, are displayed in  Supplementary Tables S2  and  S3 .\nDemographic and clinical characteristics of the study participants (N = 51),\ncategorized by presence or absence of haemoperitoneum at initial presentation.\nNo women reported a history of chronic fatigue syndrome, chronic pain syndrome,\nfibromyalgia, inflammatory bowel disease, or inflammatory bowel syndrome in either\ngroup. IQR, interquartile range.\nTable 2  shows the primary diagnosis made on\nTVS at the initial presentation, including the causes for haemoperitoneum. The most common\nreason for haemoperitoneum was a ruptured functional haemorrhagic ovarian cyst, present in\n13/15 cases (87%). The use of hormonal contraception was less frequent in women who\ndeveloped deep endometriosis, but the difference was not statistically significant (47% vs\n20%,  P  =   0.1). There were 10/36 (28%) women in the ‘no\nhaemoperitoneum’ group who were taking the combined oral contraceptive pill (COCP) or the\nprogesterone-only pill, compared to no women in the haemoperitoneum group\n( P  =   0.02). Furthermore, all 3/15 (20%) women in the\nhaemoperitoneum group who were using hormonal contraception, had a levonorgestrel-containing\nintrauterine device (Lng-IUS)  in situ , compared to only 1/36 (2.8%) in the\n‘no haemoperitoneum’ group ( P  =   0.07).\nClinical diagnosis causing acute pelvic pain in the study participants (N = 51),\ncategorized by presence or absence of haemoperitoneum at initial presentation.\nWhere N differs from 44 women in the group who did not develop endometriosis and\nseven in the group who did develop endometriosis, this was due to missing data. TVS,\ntransvaginal ultrasound; IUCD, intrauterine contraceptive device; CMV,\ncytomegalovirus.\nAfter completion of follow-up, 7/15 (46.7%; 95% CI 21.3–71.4%) women with haemoperitoneum\ndeveloped sonographic evidence of deep endometriosis, compared to 0/36 (0%; 95% CI 0.0–9.7%)\nwomen in the control group, who had no signs of haemoperitoneum at inclusion. Of the seven\nwomen who developed deep endometriosis, only one patient also had evidence of an ovarian\nendometrioma. Of the 7/15 (46.7%) women who developed deep endometriosis, 4/7 (57%; 95% CI\n20.5–93.8%) presented with severe haemoperitoneum and 3/7 (43%; 95% CI 6.2–79.5%) with\nmoderate haemoperitoneum.\nOf the women with significant haemoperitoneum at their initial visit, the proportion who\nhad sonographic evidence of deep endometriosis at their follow-up scans is demonstrated in\n Fig. 2 . The formation of deep endometriosis\nover time is illustrated in  Fig. 3 . Of the 8/15\n(53%) women who did not develop deep endometriosis following haemoperitoneum, the pelvic\nblood clots had resolved by their first follow-up visit, which occurred at 2 months in 6/8\n(75%) women. In 2/8 women, this was undertaken earlier for clinical reasons, 17 and 25 days\nfrom the initial attendance.\nKaplan–Meier estimator illustrating the proportion of women who presented with\nacute pelvic pain and significant haemoperitoneum at initial scan, who remained free\nof deep endometriosis during the follow-up period.  Estimator is inclusive of 15\nwomen.\nFormation of endometriosis nodules following haemoperitoneum.  B-mode\ntransvaginal ultrasound images in the upper row, with corresponding schematic\nillustrations below. 0 weeks: Significant haemoperitoneum containing smaller clots (c)\nsecondary to a ruptured functional haemorrhagic cyst. 1 week: Free fluid in the pouch of\nDouglas is starting to reabsorb. A larger clot (C) is formed, located in the\nretrocervical region. Peritoneum shows reactive thickening (*). 8 weeks: This image\nillustrates the blood clot transitioning to an endometriotic nodule (C/N). The blood\nclot becomes smaller, more solid and hypoechoic over time, and adherent to the\nsurrounding peritoneum and bowel. The muscularis layer of the bowel (B) appears\nthickened, as does the surrounding peritoneum (*). 26 weeks: The resolving clot\ncontracts further in size and becomes a completely solid, incompressible, hypoechoic\nlesion (N). It is tender on palpation with the ultrasound probe and resembled the\nappearance of an endometriotic nodule. The nodule invades the muscularis layer of the\nanterior wall of the rectosigmoid colon (B). The transition of the peritoneal to a\nfibrotic state is visible (*). C, clot; Cx, cervix; C/N, the transitioning blood clot\ninto an endometriotic nodule; N, endometriotic nodule; B, bowel wall.\nThe median number of endometriotic nodules seen per person was 1 (range 1–2) and the\nlocations involved were the uterosacral ligaments (n = 3), retrocervical area (n = 3),\nrectosigmoid colon (n = 1), and bladder (n = 1). Partial obliteration of the POD was seen in\n3/7 women (43%; CI 6.2–79.5%). Other pelvic adhesions involving the organs neighbouring the\nnewly formed endometriotic nodule(s) were noted in 4/7 women (57%; CI 20.5–93.8%).\nWhen comparing women who developed endometriosis with those who did not, there were only a\nfew statistically significant differences in pelvic pain, bowel symptoms, and EQ-5D-3L\nscores ( Tables 3  and  4 ). A greater proportion of women who later developed\nendometriosis reported having constipation at baseline compared to women who did not develop\nendometriosis ( P  =   0.01), but this was not seen at\n6 months. Women who developed endometriosis also reported greater difficulty emptying their\nbladder at 6 months ( P  =   0.04) ( Table 3 ), and they also had a lower EQ-5D index score at both\nbaseline and 6 months ( P  =   0.002,\n P  =   0.03, respectively) ( Table 3 ), compared to women who did not develop endometriosis.\nSeverity and frequency of pain symptoms and EQ-5D-3L quality of life scores in groups\nwho did and did not develop endometriosis, at baseline and at 6 months.\nWhere N differs from 44 women in the group who did not develop endometriosis, and 7\nin the group who did develop endometriosis, this was due to either ‘N/A’ being\nselected on the questionnaire or missing data. (*) Number representing EQ-5D Index.\n(**)  P -values calculated from women who recorded scores for both\ntimepoints for each variable only (women who recorded ‘N/A’ for either timepoint were\nalso excluded), therefore N differed from values in table slightly in some cases. For\nfurther details, see  Supplementary\nTable S2 . VAS, Visual Analogue Scale. IQR, interquartile range. EQ-5D-3L,\nEuroQoL-5 Dimension-3 Level. EQ-5D Index, EuroQoL-5 Dimension Index. EQ-VAS,\nEuroQoL-Visual Analogue Scale.\nSeverity and frequency of bowel symptoms scores in groups who did and did not develop\nendometriosis, at baseline and at 6 months.\nWhere N differs from 44 women in the group who did not develop endometriosis, and 7\nin the group who did develop endometriosis, this was due to either ‘N/A’ being\nselected on the questionnaire or missing data. (*) Number representing EQ-5D Index.\n(**) P-values calculated from women who recorded scores for both timepoints for each\nvariable only (women who recorded ‘N/A’ for either timepoint were also excluded),\ntherefore N differed from values in table slightly in some cases.\nWe analysed the change in symptoms and QOL between baseline and 6 months ( Tables 3  and  4 ,  Supplementary Tables S4  and  S5 ), in the women who developed endometriosis and those who did not. The only\nobserved difference in change was EQ-5D index and EQ-VAS scores, which demonstrated that by\n6 months, there was a statistically significant improvement in QOL from baseline in the\ngroup who did not develop endometriosis ( P  <   0.001,\n P  <   0.001, respectively). In the group who did\ndevelop endometriosis, only the EQ-5D index score improved\n( P  =   0.04), not the EQ-VAS score\n( P  =   0.08) ( Table 3 ).\nThe women in the group who had significant haemoperitoneum were also reviewed at 2 months.\nIn the subgroup who developed endometriosis, EQ-5D index and EQ-VAS scores were\nsignificantly lower at initial presentation compared to the 2-month follow-up visit\n( P  =   0.03,  P  =   0.04)\n( Supplementary Table S6 ). There\nwere no other statistically significant changes in scores observed for pelvic pain, bowel,\nand urinary symptoms between baseline and 2 months, nor between 2 and 6 months ( Supplementary Tables S6  and  S7 ).\n\nOur study found that nearly half of women presenting with significant acute haemoperitoneum\ndeveloped deep endometriosis during follow-up, compared to none of the women without\nhaemoperitoneum. Our primary outcome was comparable to the findings in the study by  Bean  et al.  (2019a) , who reported\nthat 67% of women with haemoperitoneum developed endometriosis compared to only 3% of women\nwithout haemoperitoneum.\nWe observed on interval TVS examinations that in some women, blood clots did not resolve.\nInstead, they became more organized and solid over time, appearing more hypoechoic and\nsmaller in size, always remaining in the same position ( Fig. 3 ). They eventually resembled the characteristic\nincompressible, solid, hypoechoic appearance of endometriotic nodules, which were tender on\npalpation with the ultrasound probe ( Guerriero\n et al. , 2016 ). Nodules appear different from resolving blood\nclots, which tend to reside in the POD, whilst endometriotic nodules directly involve the\nbowel wall, bladder wall, sacro-uterine ligaments, or parametria, presenting as focal\nabnormalities of these organs. Nodules are also different from pelvic fibrosis involving the\npelvic organs and peritoneum, which typically appear hyperechoic on ultrasound scan, while\nthe endometriotic nodules are typically hypoechoic.\nIt has been previously demonstrated that endometrial epithelial cell colonies are prevalent\nin the peritoneal fluid of 79–90% of women, regardless of the presence of endometriosis\n( Halme  et al. , 1984 ;  Kruitwagen  et al. , 1991 ). As\nhypothesized by  Bean  et al. \n(2019a) , when peritoneal healing occurs over a blood clot, these endometrial cells\ncould become trapped underneath the peritoneal surface and trigger the development of deep\ndisease. A large amount of intraperitoneal blood translates into major oxidative stress that\nis extremely deleterious for the delicate mesothelial cells. Such cytotoxic effect on the\nperitoneal lining could open the way to the extracellular matrix for these endometrial cells\n( Wyatt  et al. , 2023 ).\nFurthermore, various molecules released by activated platelets, activation of immune cells,\nand neuroangiogenesis in response to the haemoperitoneum might cause the endometrial cells\nwithin the clot to undergo epithelial–mesenchymal transition and fibroblast–myofibroblast\ntransdifferentiation ( Yan  et al. ,\n2017 ). Activated platelets have also been shown to trigger endothelial to\nmesenchymal transition, which also occurs in endometriotic lesions ( Yan  et al. , 2020 ). These processes result in\nincreased cellular proliferation, contractility, migration, invasiveness, and collagen\nproduction, ultimately leading to fibrosis ( Guo,\n2018 ;  Yan  et al. ,\n2020 ), which is a characteristic feature of endometriosis ( Vigano  et al. , 2020 ). This fibrosis could then\nlead to adhesion formation and pelvic anatomical distortion ( Nisolle and Donnez, 1997 ), which was also observed in our\nstudy.\nBlood clots were most commonly found in the posterior compartment owing to gravitational\neffects. This would explain why deep endometriosis and adhesions typically form in the\nposterior compartment, often involving the anterior bowel wall and leading to obliteration\nof the POD ( Chapron  et al. ,\n2006 ;  Chaggar  et al. ,\n2023a ).\nIn 87% of the women who presented with acute haemoperitoneum, this was secondary to a\nruptured functional haemorrhagic cyst. Patient characteristics that are positively or\nnegatively associated with ovulation, such as frequent menstrual cycles, early menarche,\nparity, and oral contraceptive use, have been consistently linked to endometriosis risk\n( Sangi-Haghpeykar and Poindexter, 1995 ;\n Arumugam and Lim, 1997 ;  Moen & Schei, 1997 ;  Vercellini  et al. , 1997 ). Our study offers a\npathophysiological explanation for this link. This association is further supported by a\nrecent, large cohort study, reporting a positive association between the presence of\nfunctional haemorrhagic ovarian cysts and endometriosis ( Chaggar  et al. , 2023a ). Although the overall use\nof hormonal contraception was not statistically less frequent in women who developed deep\nendometriosis (47% vs 20%,  P  =   0.1), given the large\ndifference in the two figures, a type II error may have occurred here. This is likely owing\nto the small sample size. Additionally, there was a trend towards more women using the\nLng-IUS in the haemoperitoneum group compared to the ‘no haemoperitoneum’ group\n( P  =   0.07). This is less likely to inhibit ovulation\nthan the COCP or oral progestogens, which were more commonly being used in the ‘no\nhaemoperitoneum’ group ( P  =   0.02).\nWhy not all women with blood clots in the pelvis develop deep endometriosis is uncertain.\n Kapczuk  et al.  (2023) \nreported that only 46% women who had developed retrograde menstruation caused by obstructive\nuterine anomalies were diagnosed with endometriosis at surgery, which is very similar to our\nfindings. The factors which determine the severity of inflammatory response to the presence\nof blood within the peritoneal cavity are currently unknown and this requires further\nresearch. Of the 8/15 (53%) women who did not develop deep endometriosis following\nhaemoperitoneum, the pelvic blood clots had resolved within 2 months. In contrast, in 5/7\n(71%) women who developed deep endometriosis, it took at least 6 months for the blood clots\nto completely resolve ( Fig. 2 ). Alternative\nexplanations could involve the role of genetic predisposition, inflammatory changes, and\nindividual immunological factors ( Burney and Giudice,\n2012 ;  Zubrzycka  et al. ,\n2020 ). Our sample size did not allow us to establish a reliable correlation between\ndemographic and clinical covariates and the risk of developing haemoperitoneum or\nendometriosis.\nWe documented clinical symptoms and HR-QOL in order to investigate the potential clinical\nsignificance of  de novo  formation of deep endometriosis nodules. Observed\ndifferences, such as a higher prevalence of constipation at baseline and difficulties\nemptying the bladder at 6 months, in the endometriosis group do not seem clinically relevant\ngiven the small size of the endometriosis group.\nWe observed a statistically significant increase in the EQ-5D scores between baseline and\n6 months for the group who developed endometriosis, but not for the EQ-VAS, unlike the ‘no\nendometriosis’ group, which noted improvements in both QOL variables. The improvements are\nlikely linked to baseline data being obtained while the participants were admitted with\nacute pain. Although it did not quite reach statistical significance, the EQ-VAS did also\nshow some improvement in the endometriosis group ( P =  0.08). These findings\ncould have also been influenced by the small sample size and the length of the follow-up\nperiod, which was probably too short to see a potential clinical effect of endometriosis,\nwhich is not always symptomatic. The median EQ-5D index scores at 6 months were 0.73 and\n0.85 for the endometriosis and ‘no endometriosis’ groups, respectively, both of which are\nconsidered clinically good.\nOur findings suggest that because women with significant haemoperitoneum are more likely to\ndevelop deep endometriosis, surgical management (laparoscopy and washout) could be offered\nas a preventative measure, even if they are clinically stable.\nThe strengths of this study include the innovative hypothesis, the prospective design and a\nhigh quality of ultrasound examination with clearly defined diagnostic criteria. Although\nthis was not a single-operator study, which would theoretically reduce inter-observer\nvariability, all examiners were extensively trained in the ultrasound diagnosis of deep\nendometriosis. They belonged to the same academic group, and were using the same model of\nultrasound machines and transvaginal probes, allowing for a consistent approach to\nexaminations. Furthermore, recent studies have demonstrated high inter-observer\nreproducibility in the detection of deep endometriotic nodules ( Bean  et al. , 2020 ;  Chaggar  et al. , 2023b ).\nThis study was designed as validation of our previous research and was able to show that\nthe findings from the initial pilot study are reproducible. The pilot study provided\nvaluable insights that allowed us to refine the design of our current study, perform a more\naccurate sample size calculation, and establish a more standardized follow-up protocol.\nAdditionally, we were able to monitor participants’ pain scores over time, providing further\ninsight into the evolution of these scores over the study period. Our study also\ndemonstrated a strong relationship between haemorrhagic cysts and the development of deep\nendometriosis, as well as the potential protective nature of oral contraceptive pills.\nSeveral limitations of this study need to be acknowledged. Firstly, it cannot be said for\ncertain whether minimal, superficial endometriosis was already present at the start of the\nstudy and how this could have contributed to the development of deep endometriosis. However,\nthis limitation also applies to women without haemoperitoneum, which was the only\nidentifiable difference between the groups who did and did not develop deep endometriosis.\nFurthermore, while the ultrasound findings were in line with deep endometriosis, we have no\nhistological data to prove that they in fact represent endometriosis. This should be\ninvestigated in future studies.\nWhile we have investigated changes in HR-QOL and symptom scores, without finding clinically\nsignificant differences over time and between groups, the baseline score was taken when the\npatient was admitted with acute pain. This could likely have influenced the baseline towards\na lower score.\nAlthough the sample size in this study was calculated using findings from a previous\nsimilar study, it was still a small sample and larger studies would be required to draw\ndefinitive conclusions, particularly regarding analysis of consequences of the newly formed\nendometriotic lesions on fertility. Only a small proportion of our cohort was trying to\nconceive, preventing us from reaching sufficient conclusions regarding fertility.\nFurthermore, one of the most common causes of significant haemoperitoneum in pregnant\npatients is ruptured ectopic pregnancy, which cannot safely be managed expectantly. It is\npossible that the haemoperitoneum would have behaved in the same way in pregnant women,\nleading to the development of deep endometriosis. However, the morphology and behaviour of\nendometriosis in ongoing pregnancies are very different compared to non-pregnant women\n( Bean  et al. , 2023 ) and a\nseparate study would be necessary to look at this.\nIn addition, although none of the women recruited to our study had history of recent egg\ncollection, this is another important cause of significant haemoperitoneum. In fact, the\nprevious preliminary study ( Bean  et\nal. , 2019a ) did report that two of their cases of haemoperitoneum were\nsecondary to this.\nIn conclusion, our study provides further evidence to suggest that significant\nhaemoperitoneum may be a precursor of deep endometriosis in some women. Clinically stable\nwomen presenting with acute pelvic pain and significant haemoperitoneum should be counselled\nabout the risk of developing deep endometriosis and offered expectant or surgical\nmanagement. However, larger future studies need to be conducted to assess women over a\nlonger period of time to see whether the effect on pain and QOL worsens over time. Fertility\nimplications should also be assessed in more detail, as well as more confounding factors\nadjusted for in the QOL analysis. Suppression of ovulation and formation of functional\ncysts, e.g. with combined and progesterone-only contraceptive pills, should be investigated\nfor the prevention of significant haemoperitoneum and development of deep endometriosis.","source_license":"CC-BY-4.0","license_restricted":false}