{"paper_id":"44e69d89-3ec1-4f54-bf4b-64f408d4d25c","body_text":"Three types of endometriosis have been described: peritoneal endometriosis, ovarian endometriosis (known as endometrioma) and deep infiltrating endometriosis (DIE) ( 1 ). These are often\nassociated with endometriosis-related infertility\nand the extent and location affect the chances of pregnancy in women ( 2 ). While the American Society of\nReproductive Medicine (ASRM) classification is a\nuseful tool to compare studies, its relevance in predicting fertility outcomes according to endometriosis\nstage is debatable ( 3 ). Moreover, the ASRM classification does not take into account the presence of DIE\n( 2 ).\nDespite the limits of the ASRM classification, two\nrandomized studies ( 4 ,  5 ) and a meta-analysis ( 6 ) have\ndemonstrated the positive impact of removing endometriotic lesions in patients with I-II ASRM stages\non spontaneous fertility. A more recent meta-analysis\nhas demonstrated the absence of a positive impact of\nremoving endometriomas before  in vitro  fertilization\n(IVF) on fertility outcomes ( 7 ). Moreover, several\nstudies have underlined the negative effect of cystectomy for endometriomas on ovarian reserve, evaluated by anti-mullerian hormone (AMH) serum level\nor antral follicle count and response to IVF stimulation, particularly in patients with bilateral cysts ( 8 ,  9 ).\nSome controversy remains over the impact of DIE\non fertility. Stepniewska et al. ( 10 ) suggested that the\nremoval of DIE was associated with enhancement of\nboth spontaneous pregnancy and increased fertility\nresults in assisted reproductive therapy (ART). Moreover, this study revealed that incomplete resection of\nDIE was associated with a lower pregnancy rate compared with patients undergoing complete removal.\nBarri et al. ( 11 ) also demonstrated that the best option for infertile patients with endometriosis, depending on their age, was the combination of surgery and\nIVF. However, Mathieu d’Argent et al. ( 12 ) reported\nsimilar ICSI-IVF pregnancy rates in patients with\nDIE and colorectal involvement as in those with tubal\nor male infertility. This raises the question of whether\nsurgery, which exposes patients to the risk of severe\ncomplications, is a legitimate option to enhance fertility outcomes in ART. However, none of these authors\nwere able to demonstrate whether the association of\nendometrioma with DIE, a common occurrence, has\nan impact on IVF results.\nThe aims of this study were therefore to evaluate the\nimpact on pregnancy rate of endometrioma associated\nwith DIE after a first ICSI-IVF cycle and to evaluate\ndeterminant factors to establish a pragmatic approach.\n\nWe retrospectively identified 104 women with endometrioma who had undergone ICSI-IVF treatment\nafter at least 1 year of infertility in the Department of\nGynecology-Obstetrics at Tenon hospital (France)\nfrom January 2007 to June 2010. The investigation\nof fertility included a hormonal blood test in the third\nday of the cycle [serum level measurements of estradiol (E2), follicle stimulating hormone (FSH), inhibin\nB and anti-mullerian hormone (AMH)], a hysterosalpingography, transvaginal sonography and semen\nanalysis for the partner. The diagnosis of endometriosis was made with physical examination, transvaginal\nsonography and magnetic resonance imaging (MRI)\nusing previously published imaging criteria ( 13 ). DIE\nwas diagnosed with physical examination when lesions on the posterior vaginal fornix were found or\nwhen we identified some infiltration or nodule on\nthe torus uterinus or uterosacral ligaments. With the\ntransvaginal sonography, DIE was diagnosed when\none of these structures were involved: vagina, uterosacral ligament, rectovaginal septum, rectosigmoid\ncolon and bladder. The diagnosis of DIE was done\non abnormal hypoechoic linear thickening and nodules. With MRI, the diagnosis of DIE was done on\nthe combined presence of signal abnormalities on the\nsame structure mentioned before.\nThe patients were divided into two groups; the\nendometrioma group (37 women) consisting of patients with proved endometrioma without DIE and\nthe endometrioma-DIE group (67 women) consisting of patients with endometrioma and DIE.\nThree different forms of down regulation were used:\na long gonadotropin-releasing hormone (GnRH) agonist, a short agonist or an antagonist protocol. Ovarian\nstimulation was done with doses of recombinant FSH\nbetween 75 and 450 IU/d depending on patient age,\nbody mass index (BMI), antral follicle count (AFC),\nAMH, size and number of follicles, and E2 levels.\nThis stimulation was begun once pituitary desensitization (E2 level <50 pg/mL) had been achieved.\nTransvaginal oocyte retrieval was scheduled 35-36\nhours after hCG injection and embryo transfer (ET)\nwas performed 2-3 days later. On day 2, individually\ncultured embryos were evaluated on the basis of the\nnumber of blastomeres, blastomere size, fragmentation rate and presence of multinucleated blastomeres\n( 14 ). The top quality embryos were defined as having\nfour regular blastomeres with <20% fragmentation.\nThe luteal phase was supported by vaginal administration of micronized P (400 mg/d) from the day\nof ovarian puncture to the day of the pregnancy test.\nPregnancies were diagnosed by an increasing concentration of serum β-human chorionic gonadotropin\n(β-hCG) which was tested 14 days after ET. Clinical pregnancies were confirmed by the presence of\na gestational sac on vaginal ultrasound examination\nduring the fifth week.\nFor embryo transfer, a soft catheter was used which is inserted through the cervical canal into\nthe uterine cavity. Ultrasound guidance and anesthesia was not required.\nUnivariate analysis was performed using Student’s t test or Wilcoxon test for continuous variables and Chi-square test or Fisher’s exact test\nfor qualitative variables. We tested epidemiological, biological and radiological characteristics in a\nmultivariate analysis for association with pregnancy rate. A p value of less than 0.05 was considered\nsignificant.\nRecursive partitioning (RP) was used to determine cut-offs for each variable predicting an improvement in pregnancy rate. RP is a technique\nwhich can be applied to examine large datasets\nto uncover hidden patterns within the data and\nto elucidate statistically significant sub-groupings within the data. RP is non-parametric in\nnature, imposing no a priori restrictions on the\ndistributional forms of the predictor variables.\nThe central result is a simple and intuitive RP\nalgorithm. At each step, the RP program determines for each variable cut-points that optimally separate patients into homogeneous groups.\nA multiple logistic regression (MLR) was performed; including all variables that were correlated to the conception rate. Only independent\nfactors of pregnancy rate were included in a RP\nmodel. All analyses were performed using the R\npackage with the Verification, Design, Hmisc,\nDiagnosisMed, ROCR and Presence Absence\nlibraries.\nAll the patients gave informed consent to participate in the study. The protocol was approved by\nthe Ethics Committee of the Collège National des\nGynécologues et Obstétriciens Français (CNGOF).\n\nOne hundred and four women with endometrioma and proved infertility who had undergone\nICSI-IVF cycles were included with only the first\ncycles being analyzed. The epidemiological characteristics of the whole population are summarized\nin table 1. The median age of the population study\nwas 32 years and the median BMI was 22.4 kg/m 2 . The median duration of infertility was 3 years.\nEpidemiological characteristics of the 104 patients with endometrioma with or without DIE\nEndometriomas were unilateral and associated\nwith DIE in 55.7 and 64.4% of cases respectively. The median number of endometriomas was\ntwo and the median size of the largest endometrioma was 33 mm. The proportion of patients\nwith a major endometrioma measuring less than\n3 cm, between 3-5 cm and more than 5 cm was\n36.5, 51 and 12.5% respectively. Before the ICSI-IVF cycle, AMH serum levels were 4 ng/ml\nin women with an endometrioma diameter size\nlower than 3 cm or between 3-5 cm and 3 ng/ml\nin those with an endometrioma diameter size\nover 5 cm. No difference in AMH serum levels\naccording to endometrioma sizes was observed.\nAMH serum levels in patients with or without prior surgery for endometrioma were 2.7 ng/ml\nand 3.9 ng/ml respectively (p=0.2) and in those\nwith or without prior surgery for DIE were 2.9\nand 3 ng/ml respectively (p=0.9).\nThe epidemiological characteristics of patients\nwith endometrioma with or without DIE are summarized in table 2. No difference in the median age,\nsmoking, duration of infertility or rate of associated tubal and male infertility was found. BMI was\nhigher in the group of patients with endometrioma\nand DIE (p=0.01). AMH serum levels in patients\nwith endometrioma with or without DIE were 3.2\nand 3.4 ng/ml respectively. No difference in AMH\nserum levels was found between the groups.\nComparison of epidemiological characteristics of patients with endometrioma with or without DIE\nThe pre ICSI-IVF biological characteristics\nand responses to hormonal ovarian stimulation\nof the patients with endometrioma with or without DIE are summarized in table 3. No differences in the AMH, inhibin B, E2, AFC, number of ICSI or IVF procedures, types of ovarian\nstimulation, total dose of gonadotrophin used,\nnumber of mature follicles >14 mm, total number of oocytes retrieved, total number of day-2\nfresh embryos, number of top day-2 fresh embryos, thickness of endometrium, number of\ntop day-2 embryos transferred and number of\nembryos cryopreserved were found between\nthe groups. An association between the requirement for ICSI and male infertility was observed\n(p<0.0001).\nComparison of epidemiological characteristics of patients who conceived and those who\ndid not is given in table 4. Using univariable\nanalysis, the number of patients who conceived\nwas lower in the group of patients with endometrioma and DIE (in the group of patients with\nendometrioma and DIE: patients who conceived\nn=22 (51.1%) vs. patients who did not conceive\nn=45 (73.8%); p=0.03).\nBiological characteristics and responses to ovarian stimulation of patients with endometriomas with or without DIE\nCharacteristics of patients who conceived and who did not conceive\nMultivariable analysis identified three independent factors of pregnancy rate. A lower\nrate was associated with the presence of DIE\n(OR=0.24, 95% CI: 0.085-0.7, p=0.009) and\nthe use of ICSI (OR=0.23, 95% CI: 0.07-0.8,\np=0.02) and higher rate with an AMH serum\nlevel above 1 ng/ml (OR=4.3, 95% CI: 1.1-\n19, p=0.049). After RP, the presence of DIE\nemerged as the most likely determinant factor\nof pregnancy ( Fig 1 ). The calibration of the\nmodel was good with an ROC AUC (95% CI)\nof 0.70 (0.65-0.75) ( Fig 2 ).\nRecursive partitioning model to predict pregnancy\nrate.\nCalibration of the model with an ROC AUC.\n\nThis study demonstrates that the presence of\nDIE in patients with endometrioma requiring an\nICSI-IVF for infertility has a negative impact on\npregnancy rate.\nIndications to treat endometrioma before\nICSI-IVF in infertile patients have been the\nsource of controversy. A recent meta-analysis including four trials ( 7 ) demonstrated that\nlaparoscopic aspiration or cystectomy of endometrioma prior to ICSI-IVF did not show evidence of benefit over expectant management\non clinical pregnancy rates. However, these\nauthors did not evaluate the impact of associated DIE on fertility outcomes. While Redwine\n( 15 ) reported that isolated endometriomas were\nobserved in less than 1.1% of patients suggesting that endometrioma management cannot be\nanalyzed independently of the presence of other\nlocations of endometriosis, this author did not\ndifferentiate patients with superficial peritoneal\nendometriosis from those with DIE. In the present study, about two-thirds of infertile women\nwith endometrioma had associated DIE proved\nby clinical examination, trans-vaginal sonography and MRI. Before ICSI-IVF, the failure\nrate of intra-uterine insemination (IUI) was significantly higher in patients with endometrioma\nand DIE underlining the need to distinguish between infertile patients with endometrioma and\nDIE, and those without. Stepniewska et al. ( 10 )\nfound that removal of DIE in infertile women\nincreased the pregnancy rate of ART including\nIUI and IVF but did not take into account the\nassociation of endometrioma with DIE. Pabuccu et al. ( 16 ) investigated the outcome of ICSI\ncycles in women with mild-to-moderate endometriosis and endometrioma but none of the\npatients included in the trial had DIE. Finally,\nonly one of the four trials of the meta-analysis included patients with associated infertility\nfactors such as male sub-fertility, cervical and\ntubal factor ( 17 ). This is particularly important\nas more than half of the patients in this study\nhad associated tubal infertility often subsequent\nto anatomical distortion of the fallopian tubes\nlinked to DIE and nearly half of the patients\nalso had associated male infertility. Therefore,\nthe conclusion of this meta-analysis is relevant\nonly for the small subgroup of patients with\nisolated endometrioma and this is far from the\nreality of clinical practice.\nUsing multivariable analysis, the present study has demonstrated that endometrioma associated\nDIE, AMH serum levels and the type of ART (IVF\nor ICSI) were independent prognostic factors of\npregnancy. In a study comparing conservative surgery for rectovaginal endometriosis with expectant management, Vercellini et al. (3) reported a\n12-month and a 24-month cumulative probability\nof conception of 20.5 and 44.9% respectively in the\nformer group and 34.7 and 46.8% respectively in\nthe latter (not significant) suggesting that excision\nof rectovaginal endometriosis does not improve\nthe likelihood of pregnancy nor reduce time-toconception. In a review of the literature, these\nauthors concluded that the purported benefit of\nexcision of rectovaginal endometriosis in infertile patients reported by several authors may be\nattributed to treatment of co-existing peritoneal\nand ovarian endometriosis ( 18 ). These results\ncontrast with those of other authors suggesting\nthat the removal of DIE enhanced both spontaneous pregnancy and increased pregnancy rates\nin IUI and IVF treatments ( 10 ). Moreover, in a\nrandomized trial comparing laparoscopy to open\nsurgery for colorectal resection of endometriosis, Daraï et al. ( 19 ) found that removal of lesions enhanced spontaneous pregnancy even in\npatients with prior failure of IVF. Appasamy et\nal. ( 20 ) reported that a cumulative score using\nbasal FSH, basal AMH, delta E2, delta inhibinB, AFC and age was the best predictor of ovarian reserve with a ROC AUC of 0.91. In the\npresent study, among biological parameters and\nAFC, the sole independent factor was AMH serum level. These results are in agreement with\nthose of Buyuk et al. ( 21 ) who reported that patients with elevated AMH serum level ≥0.6 ng/\nmL had twice the number of oocytes retrieved,\na greater number of day-3 embryos and a higher\nclinical pregnancy rate compared with patients\nwith an AMH serum level below this value. In a\nlogistic regression analysis, La Marca et al. ( 22 )\nfound that AMH and age were the only independent predictive criteria of live birth but with\na sensitivity of 79.2% and a specificity of only\n44.2%. A lower pregnancy rate was associated\nthe use of ICSI. Indeed, ICSI bypasses the selective biological barrier of the zona pellucida\nand increases the probability of introducing an\nabnormal spermatozoa into the oocyte ( 23 - 25 )\nwhich is detrimental to embryo development.\nA few studies have focused on patient and\nendometriosis characteristics that may be useful to evaluate the individual probability of\npregnancy in infertile patients. Younis et al.\n( 26 ) recommended the use of a scoring system\ntaking into account both epidemiological and\nbiological characteristics and the antral follicle\ncount to predict fertility results in IVF. However, this score does not take into account the\npresence of DIE which appears the most relevant predictive factor of pregnancy rate in our\nmodel. Similarly, Adamson et al. ( 27 ) recommended the use of a fertility index to evaluate\nthe probability of obtaining spontaneous pregnancy in patients with endometriosis taking\ninto account patient age, duration of infertility,\nprior pregnancy, tubal and total ASRM scores\nwithout distinguishing between patients with\nor without DIE. As previously mentioned, the\ncovariates of our model are clinically significant and concordant with the published data\nunderlining its potential use in routine practice. Thus, further studies are required to evaluate the calibration of the model, an important\nparameter reflecting the accuracy of prediction\nfor continuous models by giving an idea of the\nmodel’s performance when extrapolated to a\nnew patient population.\nSome limitations of the present study have to\nbe underlined. First, the retrospective nature of\nthis study cannot exclude all potential biases.\nSecondly, the true impact of DIE in patients\nwith infertility associated with endometrioma\ncan only be truly assessed by a prospective trial\ncomparing fertility results of ART in patients\nwith DIE compared to those after removal of\nDIE. Third, the higher BMI in patients with endometrioma and DIE in our study constitutes\na compounding factor. However, the number\nof obese patients (BMI >30 kg/m²) was low\n(2.6%) and no difference in response to hormonal ovarian stimulation was observed among\nthe groups. Fourth, we used only the data from\nthe first ICSI-IVF cycle of each patient to develop the model which means that the cumulative pregnancy rate after several cycles could\nnot be evaluated. Finally, further external validation studies are required before the use of the\npresented model in clinical practice.\n\nThe data in this study support that DIE associated with endometrioma in infertile patients has a\nnegative impact on the pregnancy rate in first cycle\nICSI-IVF. Moreover, the resultant predictive model of pregnancy rate could provide better prediction for couples about the chances of conceiving,\nthereby contributing to a comprehensive strategy\nof infertility management.","source_license":"CC0","license_restricted":false}