Interventions for women with endometrioma prior to assisted reproductive technology

Background Endometriomata are cysts of endometriosis in the ovaries. As artificial reproductive technology (ART) cycles involve oocyte pickup from the ovaries, endometriomata may interfere with the outcome of ART.

To determine the effectiveness and safety of surgery, medical treatment, combination therapy or no treatment for improving reproductive outcomes among women with endometriomata, prior to undergoing ART cycles. Search methods The review authors searched: Cochrane Menstrual Disorders and Subfertility Group Specialised Register of trials, CENTRAL (The Cochrane Library), EMBASE, MEDLINE, PubMed, PsycINFO, CINAHL, DARE, trial registers for ongoing and registered trials, citation indexes, conference abstracts on the ISI Web of Knowledge, Clinical Study Results, OpenSIGLE (July 2010) and handsearched Fertility and Sterility (2008 to 2010). Selection criteria Randomised controlled trials of any medical, surgical or combination therapy or expectant management for endometriomata prior to ART. Data collection and analysis The trials were independently identified and assessed for risk of bias by two authors. The authors of the trials that were potentially eligible for inclusion were contacted for additional information. Outcomes were expressed as Peto odds ratios and mean differences (MD). Main results Eleven trials were identified of which seven were excluded and four with 312 participants were included. No trial reported live birth outcomes. One trial compared gonadotropin‐releasing hormone (GnRH) agonist with GnRH antagonist. There was no evidence of a difference for clinical pregnancy rate (CPR), however the number of mature oocytes retrieved (NMOR) was greater with GnRH agonists (MD ‐1.60, 95% CI ‐2.44 to ‐0.76) and the ovarian response was increased (estradiol (E2) levels on day of human chorionic gonadotropin (hCG) injection) (MD ‐456.30, 95% CI ‐896.06 to ‐16.54). Surgery (aspiration or cystectomy) versus expectant management (EM) showed no evidence of a benefit for clinical pregnancy with either technique. Aspiration was associated with greater NMOR (MD 0.50, 95% CI 0.02 to 0.98) and increased ovarian response (E2 levels on day of hCG injection) (MD 685.3, 95% CI 464.50 to 906.10) compared to EM.Cystectomy was associated with a decreased ovarian response to controlled ovarian hyperstimulation (COH) (MD ‐510.00, 95% CI ‐676.62 to ‐343.38); no evidence of an effect on the NMOR compared to EM. Aspiration versus cystectomy showed no evidence of a difference in CPR or the NMOR. Authors' conclusions There was no evidence of an effect on reproductive outcomes in any of the four included trials. Further RCTs of management of endometrioma in women undergoing ART are required.

Female; Humans; Pregnancy; Reproductive Techniques, Assisted; Endometriosis; Endometriosis/complications; Endometriosis/drug therapy; Endometriosis/surgery; Gonadotropin‐Releasing Hormone; Gonadotropin‐Releasing Hormone/agonists; Gonadotropin‐Releasing Hormone/antagonists & inhibitors; Infertility, Female; Infertility, Female/etiology; Ovarian Hyperstimulation Syndrome; Ovarian Hyperstimulation Syndrome/prevention & control; Randomized Controlled Trials as Topic; Sperm Injections, Intracytoplasmic Plain language summary Interventions for women with endometrioma prior to assisted reproductive technology Endometriomata are a form of ovarian endometriosis, classified as cysts within the ovaries. They are a common cause of subfertility and pelvic pain. This review aimed to determine which treatment approach was better for women with subfertility and endometriomata who were undergoing assisted reproductive technology (ART). Four trials were identified. A gonadotropin‐releasing hormone (GnRH) agonist showed a positive treatment effect on the ovarian response to controlled ovarian hyperstimulation (COH) and the number of mature oocytes retrieved compared to GnRH antagonist. The evidence for surgery was limited but aspiration was associated with a greater ovarian response than expectant management (a wait and see approach). Further randomised controlled trials of interventions for the management of endometrioma in women undergoing ART are required.

Description of the condition Endometriosis is a common condition which may be present in up to 22% of asymptomatic women and up to 45% of women with pelvic pain (Hart 2008). Endometriomata (plural of endometrioma) are a form of endometriosis located on the ovaries and are also known as endometrial or chocolate cysts. The latter were first described by Sampson 1921. Endometriomata arise out of deposits of endometrial cells derived from the uterus that relocate within the ovaries. What causes these cells to implant in the ovaries is not clear. Sampson 1927 speculated that endometrial tissue develops on the surface of the ovary causing adhesions which pull the ovary to the uterus, creating an ovarian cavity. In contrast, Hughesdon 1957 and Brosens 1994 hypothesised that a cavity is formed by an invaginated cortex (folding inwards of the outermost layer) of the ovary causing adhesions and superficial ectopic endometrial implants (implants outside the uterus). Vercellini 2003a proposed that the endometrial cells regurgitate through the fallopian tubes as a result of menstrual spill, causing ovarian adhesions that involve the pelvic peritoneum which in turn cause progressive invagination of the ovary. This would imply that the wall of the endometrioma is formed out of the lining of the ovarian cortex, making the endometrioma, in fact, a pseudocyst (false cyst). Expectant management (a 'wait and see' policy) is still often not an option for women with endometriomata because of severe symptoms. One of the primary indications for treatment is that the endometriomata may impair the outcome of artificial reproductive technology (ART). Other indications for treatment are pelvic pain and dyspareunia (pain during intercourse) (Hart 2008; Yanushpolsky 1998). Other symptoms include dysmenorrhoea (uterine pain during menstruation) and decreased quality of life (Cobellis 2004; Gilmour 2008; Jones 2004). In addition, risks of the endometriomata themselves are rupture and torsion of the ovary. Untreated endometriomata rarely become malignant (Nishida 2000). Both medical and surgical treatments exist for the treatment of endometriomata, and treatments may be combined. Medical treatment will result in a reduction in size of the endometrioma up to 57%, but there may be a rapid return to the original size on withdrawal of medication (Farquhar 1998). Transvaginal ultrasound‐guided aspiration seemed a safe procedure but the recurrence rate after aspiration was found to be unacceptably high (Chan 2003). Nowadays a laparoscopic approach is the first choice when endometriomata are treated surgically. Compared to traditional laparotomy, laparoscopy is associated with fewer complications and a shorter hospital stay (Hart 2008). Surgery can cause damage to the ovaries and therefore reduce the egg quality. This can decrease the success rate of ART. Description of the intervention Treatment options for women who wish to undergo ART because of subfertility associated with endometrioma include: 1. surgical treatment to remove or destroy the endometrioma prior to ART; 2. medical treatment of the endometrioma prior to ART; 3. a combination of surgical and medical treatments prior to ART; 4. undergoing ART without treatment (expectant management or placebo).(Adamson 2003; Cahill 2002; Tim 2001). How the intervention might work The aim of surgical interventions is to restore the normal anatomy of the ovaries. Removal of the endometriomata can potentially improve accessibility of the fallopian tubes and therefore improve spontaneous fertility rates as well as ART outcomes including the response to ovarian stimulation. Removal of any adhesions can also potentially reduce pain. However, removal of the endometrioma may involve removing ovarian tissue with a possible negative impact on future fertility. Medical treatment is intended to create an amenorrhoeic state (the absence of a menstrual period in a woman of reproductive age). This might decrease the endometriomata because the growth of endometriosis is stimulated by the hormone oestrogen. Nevertheless, medical treatment alone in the treatment of endometriomata induced subfertility is not recommended because the cysts often recur (Hughes 2010). Why it is important to do this review ART cycles in the presence of an endometrioma are frequently complicated by difficulty with monitoring ovarian response by ultrasound, poor ovarian response to stimulation and, rarely, post‐ovum pick up infection (Matson 1986; Oehninger 1988; Pellicer 1995). The disadvantages of the surgical option include: side effects of surgery, trauma to the ovary and subsequent reduced ovarian function with resulting poor ovarian response or even premature ovarian failure (Busacca 2006; Loo 2005; Muzii 2002; Ragni 2005). The medical option is associated with temporary reduction in the size of the endometrioma and side effects of the treatment (Farquhar 1998; Winkel 2003). The advantages and disadvantages of these different approaches should be evaluated in order to inform practice decisions for women with endometriomata who are undergoing ART cycles.

To determine the effectiveness and safety of surgery, medical treatment, combination therapy or no treatment for improving reproductive outcomes among women with endometriomata, prior to undergoing ART cycles.

Criteria for considering studies for this review Types of studies Randomised controlled trials comparing treatment options for endometriomata were included. Quasi and pseudo‐randomised trials were excluded. There were no crossover trials eligible for inclusion. In an update of this review only the first‐phase data of crossover trials will be included in meta‐analyses. Types of participants Inclusion criteria Women with endometriomata who underwent surgical, medical or combination treatment or expectant management prior to ART. The endometriomata were diagnosed by laparoscopy or imaging tests such as ultrasound and magnetic resonance imaging (MRI). Exclusion criteria Women with gynaecological cancer. Women with ovarian cysts other than an endometrioma. Types of interventions The following medical interventions were considered in women with endometriomata, prior to ART cycles. Antiprogestins e.g. danazol and gestrinone. These reduce the release of follicle stimulating hormone (FSH) and luteinizing hormone (LH) from the pituitary gland, which in turn decreases the activity of the ovaries. Progestagens. Oral contraceptives. Gonadotropin‐releasing hormone agonists (GnRHa) or antagonists. The following surgical interventions were considered in women with endometriomata, prior to ART cycles. Ablation of the cyst wall following drainage. Aspiration only of the endometrioma (laparoscopic or ultrasound guided). Cystectomy, excision or stripping of the cyst wall. Additional surgical techniques that are also used at the time of surgery for endometriosis include: excision or ablation of endometriotic implants and adhesiolysis. We considered: medical versus medical + surgical; medical versus expectant management or placebo; medical versus medical; surgical versus medical + surgical; surgical versus medical; surgical versus surgical; surgical versus expectant management or placebo; surgical + medical versus expectant management or placebo. Types of outcome measures All outcomes are per woman. Primary outcomes Live birth, defined as delivery of a live fetus after 20 completed weeks of gestation. Adverse outcomes (including miscarriage, ectopic pregnancy, drug side effects, multiple pregnancies, ovarian hyperstimulation syndrome (OHSS) rate and ovum pick up pain or infection). Secondary outcomes Clinical pregnancy, defined as presence of an intra‐uterine gestational sac or fetal heartbeat visualised by an ultrasound scan. Quality of life measured by participant satisfaction or objective quality of life scales. Pain measured by visual analogue scores or other continuous scores or dichotomous data at the time of the ART cycle. Recurrence of endometriomata. Number of mature oocytes retrieved after therapy. Ovarian response to controlled ovarian hyperstimulation (COH) defined as the oestradiol levels in pg/ml on the day of human chorionic gonadotropin (hCG) injection. Search methods for identification of studies All published and unpublished randomised controlled trials (RCTs) of endometriomata treatment versus a different treatment were obtained using the following strategy, constructed by the Cochrane Menstrual Disorders and Subfertility Group (MDSG) Trials Search Coordinator Marian Showell. No language restrictions were applied. Electronic searches The output of our search was a total of 2591 titles and abstracts. The following electronic databases, trial registers and web sites were searched (from inception to 19 July 2010): MEDLINE (Appendix 1); Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library) (Appendix 2); EMBASE (Appendix 3); Menstrual Disorders and Subfertility Group (MDSG) Specialised Register of controlled trials (Appendix 4); PsycINFO (Appendix 5); CINAHL; PubMed. The MEDLINE search was combined with the RCT filter from the Cochrane highly sensitive search strategy for identifying randomised trials, which appears in the Cochrane Handbook for Systematic Reviews of Interventions (version 5.0.2, chapter 6, 6.4.11) (Higgins 2009). The EMBASE search was combined with trial filters developed by the Scottish Intercollegiate Guidelines Network (SIGN). The search terms that were used in CINAHL, PubMed and the electronic sources listed under Electronic searches are the following: 'endometrioma', 'endometriomata', 'IVF' and 'endometriosis'. Other electronic sources of trials that were searched The DARE database was searched in The Cochrane Library to find reviews with included potentially useful RCTs Trial registers for ongoing and registered trials: Current controlled trials (www.controlled‐trials.com); ClinicalTrials.gov, a service of the US National Institutes of Health (http://clinicaltrials.gov/ct2/home); the World Health Organization International Trials Registry Platform search portal (www.who.int/trialsearch/default.aspx). Citation indexes (http://scientific.thomson.com/products/sci/) Conference abstracts in the ISI web of Knowledge (http://isiwebofknowledge.com/) Clinical Study results for clinical trial results of marketed pharmaceuticals (www.clinicalstudyresults.org/) OpenSigle database (http://opensigle.inist.fr/) Google for grey literature Searching other resources The reference lists of articles retrieved by the search were handsearched and personal contact was made with experts in the field to locate any additional trials. Any relevant journals and conference abstracts that were not covered in the MDSG Specialised Register were handsearched in liaison with the Trials Search Coordinator. These included American Society of Reproductive Medicine (ASRM) abstracts in the supplement of Fertility and Sterility (2008 to 2010) and the Australian and New Zealand Journal of Obstetrics and Gynaecology (ANZJOG) (to 2010). Data collection and analysis Data collection and analysis were conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (version 5.0.2) (Higgins 2009). Selection of studies One author (LB) read 2591 titles and abstracts retrieved by the Trials Search Coordinator. Those that were irrelevant were removed and the full texts of the 13 remaining trials were retrieved. Authors LB and NP independently examined the full text articles for compliance with the inclusion criteria and selected four trials as eligible for inclusion in the review. As required, LB corresponded with trial investigators to clarify trial eligibility. Disagreements were resolved by consensus. Data extraction and management LB and NP designed and pilot‐tested a data extraction form which was used to extract data from eligible trials. The main trial report was used as the reference in trials that had multiple publications (Demirol 2006;Pabuccu 2004 ,Pabuccu 2007) and additional details were supplemented from secondary papers. LB corresponded with the trial investigators of all included trials in order to resolve any data queries. LB and NP independently extracted the data. Assessment of risk of bias in included studies The included trials were assessed for risk of bias using the Cochrane risk of bias assessment tool to assess: sequence generation; allocation concealment; blinding of participants, providers and outcome assessors; completeness of outcome data; selective outcome reporting; and other potential sources of bias. LB and NP assessed these six domains, with any disagreements resolved by consensus or by discussion with CF. The conclusions are presented in the risk of bias figures. It was planned to conduct a sensitivity analysis if there was a risk of selection bias (bias in allocation sequence and allocation concealment). Lack of provider blinding was expected because the clinicians performing surgical interventions cannot be blinded. Selective reporting bias was also expected because many fertility trials do not report live birth. None of the identified trials did report this outcome, which is a form of reporting bias. We accepted selective reporting where identified trials failed to report the primary outcome of live birth but did report interim outcomes such as clinical pregnancy. Measures of treatment effect For dichotomous data such as the pregnancy rate, the number of events in the control and intervention groups of all trials were used to calculate the Peto odds ratios (OR). For continuous data such as number of oocytes retrieved, mean differences (MD) between treatment groups were calculated if all trials reported exactly the same outcomes. If similar outcomes were reported on different scales, the standardised mean difference (SMD) would have been calculated. Ordinal data, such as quality of life scores, would have been treated as continuous data. For all reported outcomes of interest, the 95% confidence intervals (CI) were presented and the data from the included primary trials were entered into RevMan 5 software. Unit of analysis issues The primary analyses were per woman randomised. Any reported data not allowing valid analyses will be briefly summarized in an additional table in an update of this review. Multiple live births were not reported but will be counted as one live birth event in an update of this review. There were no crossover trials to include. Only first‐phase data from crossover trials will be included in an update of this review. Dealing with missing data The data were analysed on an intention‐to‐treat basis as far as possible and attempts were made to obtain missing data from the original investigators (Alborzi 2007; Demirol 2006; Pabuccu 2004; Pabuccu 2007). The study authors were contacted through e‐mail, fax and letters. For one trial (Alborzi 2007), we obtained additional information from Hart RJ, who used this trial for inclusion in a Cochrane review (Hart 2008). Live birth was assumed not to have occurred in participants with unreported outcomes. When trials reported sufficient details to calculate mean differences but had no information on associated standard deviations (SD), the outcome was assumed to have a standard deviation equal to the highest SD from other trials within the same analysis. For the outcomes clinical pregnancy, number of oocytes retrieved after therapy, ovarian response to COH and miscarriage rate, only the available or obtained data were analysed. Any imputation undertaken will be subjected to sensitivity analysis in an update of this review. There were no trials available for pooled analysis. Assessment of heterogeneity LB and CF considered whether the clinical and methodological characteristics of the included trials were sufficiently similar for meta‐analysis to provide a meaningful summary. Statistical heterogeneity was assessed with the I2statistic. An I2 > 50% was taken to indicate substantial heterogeneity (Higgins 2009). Substantial heterogeneity (I2 > 75%) was present in some analyses and therefore we decided not to pool them. Assessment of reporting biases In view of the difficulty in detecting and correcting for publication bias and other reporting biases, LB and NP aimed to minimise their impact by ensuring a comprehensive search for eligible trials and by being alert for duplication of data. Where there were 10 or more trials in an analysis, a funnel plot would have been used to explore the possibility of small trial effects. We were attentive to searching for the risk of reporting bias within trials, such as trials failing to report obvious outcomes or reporting them in insufficient detail to allow inclusion. Data synthesis The data from primary trials are displayed using the fixed‐effect model for the above comparisons (seeTypes of interventions). Pooling was not possible as there were only single trials in each comparison. For dichotomous variables, results for each intervention were expressed as Peto odds ratios (OR) with 95% CIs. An increase in the odds of a particular outcome, which could be beneficial (clinical pregnancy) or detrimental (adverse event such as miscarriage), was displayed graphically in the forest plots to the right of the centre‐line. A decrease in the odds of an outcome was displayed to the left of the centre‐line. For continuous outcome data such as number of oocytes retrieved after therapy, results from each intervention were expressed as the difference in means with 95% CIs. Results will be combined in a meta‐analysis to calculate mean difference (MD) in an update of this review if further studies are identified. If different scales are used then a standardised mean difference (SMD) will be calculated. Subgroup analysis and investigation of heterogeneity Subgroup analysis was planned but could not be conducted because there were not enough trials. In the protocol we stated that there had to be a minimum of 10 trials. If data are available and heterogeneity present (I2 > 50%), in an update of this review, subgroup analysis will be conducted to determine the separate evidence within the following subgroups: women with unilateral endometrioma; women with bilateral endometrioma; women with single endometrioma; women with multiple endometrioma. Sensitivity analysis In this review the cause of the heterogeneity was the lack of trials, therefore we did not conduct a sensitivity analysis. If substantial heterogeneity is detected in an update of this review, possible explanations will be explored in sensitivity analyses. There was no evidence of a high risk of bias in the four included trials, therefore no sensitivity analysis for risk of bias was conducted. Sensitivity analysis was planned but not conducted for our primary outcome live birth, to determine whether the conclusions were robust enough to support decisions made regarding eligibility and analysis. We will conduct a sensitivity analysis in future updates if live birth is reported. These analyses will include consideration of whether conclusions would have differed if: eligibility was restricted to trials without high risk of bias; trials with outlying results had been excluded; alternative imputation strategies had been adopted; a random effects model had been adopted.

Description of studies See the Characteristics of included studies and Characteristics of excluded studies tables. Four randomised controlled trials met the inclusion criteria (Alborzi 2007; Demirol 2006; Pabuccu 2004; Pabuccu 2007). Pabuccu 2007 investigated the outcome of intracytoplasmic sperm injection (ICSI) cycles after COH with GnRH antagonist or GnRH agonist in women with mild‐to‐moderate endometriosis and endometrioma. Eighty‐one women who had ovarian surgery for endometrioma were included from this study. Pabuccu 2004 investigated whether transvaginal aspiration of the endometriomata or expectant management before COH improved the ICSI outcomes. The 41 women with endometriomata and no history of previous surgery and the 40 women with non‐aspirated endometriomata were included from this study. Demirol 2006 investigated whether cystectomy of the endometrioma (49 women) before COH or direct treatment with COH (50 women) resulted in better ICSI outcomes. Alborzi 2007 investigated whether fenestration and coagulation (ablation) of the endometrioma or cystectomy before COH improved intra‐uterine insemination (IUI) outcomes. Twenty‐four women with unilateral single endometrioma underwent ablation of their endometrioma and 41 women underwent cystectomy.

of the search Six databases and 15 sources of grey literature were searched. In total 2591 abstracts and titles were screened for potential trials for inclusion in this review. The full texts of 13 trials were retrieved. Included studies See Characteristics of included studies. Four trials met the criteria for inclusion (Alborzi 2007; Demirol 2006; Pabuccu 2004; Pabuccu 2007). The analyses for this review were based on the data from these four trials which involved a total of 312 randomised women. Two trials were conference abstracts that were subsequently published (Demirol 2006; Pabuccu 2007). The trials were all published in English. Three of them were conducted in Turkey (Demirol 2006; Pabuccu 2004; Pabuccu 2007) and one in Iran (Alborzi 2007). Three trials were single centred (Demirol 2006; Pabuccu 2004; Pabuccu 2007) and one was multi‐centred (Alborzi 2007). Participants Data on 312 women were considered. The duration of subfertility was at least two years. The age range of the women was from 25 to 36 years. The size of the endometrioma ranged between ≥ 1.28 cm and < 6 cm. The women's body mass index (BMI) was within the normal range (18.5 to 25) for at least one trial (Alborzi 2007) and in the range of 19.8 to 29 for two other included trials (Pabuccu 2004; Pabuccu 2007). Demirol 2006 did not mention the BMI range. None of the participants had undergone ovarian surgery or medical therapy in the six months prior to randomisation. One trial reported other causes of subfertility in addition to endometriomata. These included male factor subfertility, cervical factor and tubal factor (Demirol 2006). Of the participants in this trial, 64% had a diagnosis of minimal endometriosis in their medical history. The other trials excluded women with causes of subfertility other than endometriomata. A total number of 357 cycles were conducted. In three trials the women each underwent one cycle (Demirol 2006; Pabuccu 2004; Pabuccu 2007). The women in Alborzi 2007 underwent at least two cycles if pregnancy did not occur in the first cycle. In total there were 235 women with unilateral endometriomata and 77 women with bilateral endometriomata. A total of 211 women had single endometrioma and 101 women had multiple endometrioma. Exclusion criteria were: previous history of ovarian or pelvic surgery, previous IVF cycle or hormonal therapy, other causes of subfertility, hydrosalpinx, documented tuberculosis, male factor infertility, thaw cycles, multiple bilateral endometriomata, diagnosis by transvaginal ultrasonography of particular neoplastic features, acute haemorrhage, multi‐ovulatory or hyperechoic wall foci in the ovaries. Interventions The following medical interventions were considered in women with endometriomata prior to ART cycles. Antiprogestins e.g. danazol and gestrinone: there were no trials eligible for inclusion that considered these treatments. Progestagens: none of the included trials included progestogen treatment. Oral contraceptives: none of the included trials included these. Gonadotropin‐releasing hormone agonists (GnRHa) or antagonists: one of the included trials compared GnRHa treatment with GnRH antagonist treatment (Pabuccu 2007). The following surgical interventions were considered in women with endometriomata prior to ART cycles. Ablation of the cyst wall following drainage: one included trial compared ablation with excision (Alborzi 2007). Aspiration only of the endometrioma (laparoscopic or ultrasound guided): one included trial compared this intervention with expectant management (Pabuccu 2004). Cystectomy: one trial compared cystectomy with ablation and the other trial compared cystectomy with expectant management (Alborzi 2007; Demirol 2006). We considered: medical versus medical + surgical: no trials that made this comparison were suitable for inclusion; medical versus expectant management or placebo: no trials that made this comparison were suitable for inclusion; medical versus medical: one trial made this comparison (Pabuccu 2007); surgical versus medical + surgical: no trials that made this comparison were suitable for inclusion; surgical versus medical: no trials that made this comparison were suitable for inclusion; surgical versus surgical: one trial made this comparison (Alborzi 2007); surgical versus expectant management or placebo: two trials made this comparison (Demirol 2006; Pabuccu 2004); surgical + medical versus expectant management or placebo: no trials that made this comparison were suitable for inclusion. Medical interventions Long protocol with GnRH agonist before ICSI: pituitary desensitization performed with 100 mg of triptrolein (decapeptyl 0.1 mg/day injected subcutaneously, initiated on day 21 of the previous cycle) Multiple doses of GnRH antagonist protocol before ICSI: daily injections of cetrorelix (0.25 mg subcutaneously) (Cetrotide; Serono, Geneva, Switzerland) initiated when the leading follicle was 14 mm in diameter with serum levels of estradiol (E2) exceeding 600 pg/mL and maintained until the day of hCG injection (Pabuccu 2007). Surgical interventions Cystectomy: laparoscopic drainage and stripping of endometriomata before ICSI. Drainage of the endometrioma cyst and dissection of the pseudocapsule of the endometrioma from the underlying stroma by gentle traction and counter traction in the right plane. Gentle bipolar coagulation was performed to the ovarian stroma when necessary versus expectant management, direct start of the ICSI cycle (Demirol 2006). In the trial by Alborzi 2007 the inner lining of the cyst was stripped from the normal ovarian tissue with the use of two atraumatic grasping forceps pulled slowly in opposite directions and removed from the abdominal cavity. Transvaginal aspiration of endometriomas by ultrasound‐guided puncture, on day 3 of the menstrual cycle (starting day for COH). After vaginal cleansing by povidone iodine solution and paracervical local anaesthetic injection the cyst was flushed several times using 0.9% saline solution until totally aspirated versus expectant management, direct start of the ICSI cycle. In addition, women who had aspiration also received a single dose of ceftazidime 2 g IM (Pabuccu 2004). Ablation: drainage and coagulation of the cyst wall (Alborzi 2007). All surgical interventions were done by laparoscopy. All women underwent controlled ovarian hyperstimulation (COH) after the intervention, or immediately in the case of expectant management. The COH procedure differed amongst the trials. Two trials used recombinant FSH 300 IU/L starting the third day of the menstrual cycle (Pabuccu 2004; Pabuccu 2007). Pabuccu 2007 combined the FSH from day 21 of the menstrual cycle with 100 μg triptolein, decreasing to 50 μg from the third day of the menstrual cycle. Women who underwent laparoscopic cystectomy of their endometriomata (Demirol 2006) started with COH after three months. They received 300 IU/L of recombinant FSH and started with 1.0 mg of leuprolide acetate in their luteal phase. The women who underwent ablation or cystectomy (Alborzi 2007) started two months after surgery with COH. Clomiphene citrate was started in a 100 mg/daily dose from day 5 for five consecutive days of the menstrual cycle. Human menopausal gonadotropin (hMG) was injected from day 8 of the menstrual cycle (2 ampoules/day). Ultrasound follicular monitoring was performed from day 10 or 11 of each cycle. Human chorionic gonadotropin (HCG) 10,000 IU was administered to all participants when at least two or three follicles reached a mean diameter of 17 mm and the serum estradiol concentration was > 500 pg/ml. Outcomes of interest for the review The primary outcome was number of live births, defined as delivery of a live fetus after 20 weeks of gestational age. None of the four trials reported live birth. All four trials reported clinical pregnancy. This outcome was defined as the presence of at least one gestational sac with detectable fetal cardiac activity by transvaginal ultrasound (defined as such in Pabuccu 2004 and Pabuccu 2007). One trial defined clinical pregnancy as the presence of a fetal heart rate confirmed by ultrasound (Demirol 2006). One trial defined clinical pregnancy by the presence of an intrauterine gestational sac on ultrasound (Alborzi 2007). Quality of life, measured by participant satisfaction or objective quality of life scales. None of the trials reported on this. None of the trials reported on pain, measured either by visual analogue scores or dichotomous data at the time of the ART cycle. Recurrence of endometriomata was not reported in any of the four trials. Number of mature oocytes retrieved was reported in all four trials (Alborzi 2007; Demirol 2006; Pabuccu 2004; Pabuccu 2007). Some trials describe this outcome as 'the number of mature oocytes retrieved' and some describe it as 'the number of metaphase II oocytes'. Both descriptions were accepted as one and the same outcome in this review because mature oocytes are in the metaphase II stage. Demirol 2006 described this outcome as 'the number of mature oocytes retrieved' and the text described that ICSi was performed for all metaphase II oocytes. Pabuccu 2004 and Pabuccu 2007 described the number of metaphase II oocytes. Demirol 2006 described this outcome as 'the number of dominant follicles'. The ovarian response to COH, measured as the estradiol level in pg/ml on the day of hCG injection, was reported in three trials (Demirol 2006; Pabuccu 2004; Pabuccu 2007). Miscarriage was reported in two trials (Pabuccu 2004; Pabuccu 2007). Ectopic pregnancy, drug side effects, multiple pregnancies, OHSS rate and post‐ovum pick up infection were not reported in any of the four trials. Excluded studies See Characteristics of excluded studies Seven trials were excluded (Cooke 1989; Dicker 1992; Gelbaya 2009; Kim 1997; Neveu 1987; Oehninger 1990; Rizk 1990). Four trials did not include women with endometriomata (Cooke 1989; Dicker 1992; Neveu 1987; Rizk 1990). In two trials the trial groups were not comparable (Kim 1997; Oehninger 1990). One trial was excluded because the trial was aborted due to inability to recruit enough participants (Gelbaya 2009). Risk of bias in included studies Allocation In all four trials randomisation was done by computer (Alborzi 2007; Demirol 2006; Pabuccu 2004; Pabuccu 2007). Pabuccu 2004 randomised in blocks of ten. In all four trials the personnel who performed the randomisation process were not involved in the trial itself. All four trials used serially numbered opaque envelopes for allocation concealment. We asked the author of one trial (Alborzi 2007) if he used a different kind of randomisation because there was a substantial difference in the number of women in group one (24) and group two (41). The author responded that a different type of randomisation was not used and that randomisation was done with computerized numbering. Unfortunately we could not find a reason for the big difference in number between groups. Therefore, we should take this into account as a potential bias. Blinding In two trials the participants, investigators, clinicians and outcome assessors were not blinded (Alborzi 2007; Demirol 2006). In one trial the investigators and outcome assessors were blinded but it was not possible to blind the participants and clinicians (Pabuccu 2004). In one trial all were blinded (Pabuccu 2007). Incomplete outcome data In three trials there were no losses to follow up or women excluded after randomisation (Alborzi 2007; Demirol 2006; Pabuccu 2004). In the GnRH agonist and GnRH antagonist groups there were eight participants cancelled after randomisation and starting treatment. Reasons were mentioned in the article and are shown in the Characteristics of included studies table (Pabuccu 2007). All four trials were long enough in duration to detect the stated outcomes. Selective reporting For three trials there was no trial protocol (Demirol 2006; Pabuccu 2004; Pabuccu 2007). We should have asked the authors if they measured other outcomes but did not report them. Instead we asked them if they reported the outcomes we were interested in (see Types of outcome measures). This made clear that in all four trials there was reporting bias because none of them reported the live birth outcome. Pregnancy rate was mentioned in every trial. We considered this as a surrogate for the outcome live birth. Surrogate outcomes are often used when observations of clinical outcomes require long follow up. The follow‐up times were not mentioned in any of the articles but the duration of all four trials was at least 36 months which is enough time to account for live births. Pabuccu 2007 did not describe the outcome measures 'post‐ovum pick up infection', 'drug side effects' and 'OHSS rate'. These are important outcomes because this trial used medical treatment and hormonal stimulation. Demirol 2006 and Pabuccu 2004 did not describe the outcome measures 'post‐ovum pick up infection' and 'OHSS rate', important because the trials used hormonal stimulation and oocyte pick up. Alborzi 2007 also did not report on 'OHSS rate' though the trial used hormonal stimulation. Two trials did not describe the miscarriage rate as an outcome (Alborzi 2007; Demirol 2006). Miscarriage can have a negative impact on quality of life (depression and anxiety) and can be accompanied by rare though serious complications such as prolonged bleeding, infection and Asherman's syndrome. None of the trials reported 'pain' as an outcome. This is a major flaw because, just like subfertility, pain has a great impact on a woman's life. Another potential form of selective reporting bias was detected in Pabuccu 2007, this was publication bias. The trial originally included four trial groups. The fourth trial group was excluded from the trial because of the advanced stage of disease, which had adverse effects on tubal function. This should not be done, nevertheless it did not affect this review because we used the data of only one of the four trial groups. Other potential sources of bias There was attrition bias in two trials because no intention‐to‐treat analysis was performed (Alborzi 2007; Pabuccu 2007). In one trial a power calculation was performed. It was calculated that a sample size of 45 women in each group would have a 87.85% power to detect a difference in means of 1.0 (SD 1.5) (Demirol 2006). In one trial the author replied that it was not possible to do a power calculation because no data existed for antagonist treatment before IVF‐ICSI cycles in women with advanced endometriosis. They could not calculate the required sample size for a pretest power estimation (Pabuccu 2007). In the trial by Pabuccu 2004, it was also not possible to do a power calculation. The author stated that this was because there was a very low prevalence of endometriomata. Alborzi 2007 did not give a reason for not performing a power calculation. One of the four included trials received additional funding (Alborzi 2007). For all four trials there was ethical approval. It seemed that there was the possibility of duplication bias for Pabuccu 2004 and Pabuccu 2007 because the trials were conducted by the same author and in the same hospital. Also, the second trial (Pabuccu 2007) started one month after the first trial (Pabuccu 2004) ended. We asked the author if the same participants were used in both trials. The author confirmed that different participants were used, therefore there was no duplication bias. In the trial by Alborzi 2007 there was a substantial difference in the numbers in the two groups (24 versus 41). The author stated that 'the method of randomisation for women with unilateral endometrioma (group 1 and 2) was computerized numbering, and that the difference was expected because of loss to follow up'. The randomisation can not be influenced by losses to follow up, therefore this could be a potential source of bias. Effects of interventions 1. One medicine versus a different medicine prior to ART (GnRH antagonist versus GnRH agonist) Live birth Live birth was not reported. 1.1 Clinical pregnancy See Analysis 1.1 In the long protocol with GnRH antagonist before ICSI, there were eight out of 33 women with a positive pregnancy test (Pabuccu 2007). In the GnRH antagonist protocol there were seven out of 34 women who became pregnant. There was no evidence of difference in treatment effect between GnRH antagonist and GnRH agonist for the clinical pregnancy outcome (Peto OR 0.81, 95% CI 0.26 to 2.54). 1.2 Number of mature oocytes retrieved See Analysis 1.2 The number of mature oocytes retrieved in the GnRH antagonist and GnRH agonist groups were 4.9 ± 1.6 and 6.5 ± 4.2, respectively (Pabuccu 2007). Analysis showed a positive treatment effect for GnRH agonists compared to GnRH antagonists (MD ‐1.60, 95% CI ‐2.44 to ‐0.76). The CI was small, which made the result more reliable. Heterogeneity was not applicable because there was only one trial. 1.3 Ovarian response to COH, defined as the estradiol level on the day of hCG injection See Analysis 1.3; Figure 3 The estradiol level on the day of hCG injection was 1512.2 ± 448 pg/ml in the GnRH antagonist group and 19,685 ± 1211 pg/ml in the GnRH agonist group (Pabuccu 2007). Treatment with GnRH agonist showed a positive treatment effect compared to GnRH antagonist on the ovarian response to COH (MD ‐456.30, 95% CI ‐896.06 to ‐16.54). The CI was wide, which made the results less reliable. Heterogeneity was not applicable as there was only one trial. 1.4 Miscarriage See Analysis 1.4 Miscarriage was considered as an important outcome measure in this review because it can have a negative impact on quality of life and can be accompanied by rare though serious complications such as prolonged bleeding, infection and Asherman's syndrome. There was one miscarriage out of 34 participants in the GnRH antagonist group and one out of 33 participants in the GnRH agonist group (Peto OR 0.97, 95% CI 0.06 to 15.85) (Pabuccu 2007). Heterogeneity was not applicable as there was only one trial. 2. Surgery (aspiration or cystectomy) versus expectant management prior to ART Live birth Live birth was not reported. 2.1 Clinical pregnancy See Analysis 2.1; Figure 4 There was no evidence of an effect on clinical pregnancy with either intervention. Aspiration versus expectant management had a Peto OR of 1.29 (95% CI 0.45 to 3.64) (Pabuccu 2004) and cystectomy versus expectant management had a Peto OR of 1.15 (95% CI 0.52 to 2.55) (Demirol 2006). 2.2 Number of mature oocytes retrieved See Analysis 2.2 In the cystectomy group the number of mature oocytes retrieved was 7.8 ± 3.0 versus 8.6 ± 2.82 in the expectant management group (Demirol 2006). There was no evidence of difference in effect between the therapies (MD ‐0.80, 95% CI ‐1.96 to 0.36). For the second trial, in the aspiration group the number of mature oocytes retrieved was 6.1 ± 1.1 versus 5.6 ± 1.2 in the expectant management group (Pabuccu 2004). Aspiration showed a positive treatment effect compared to expectant management (MD 0.50, 95% CI 0.02 to 0.98). 2.3 Ovarian response to COH, defined as the estradiol level on the day of hCG injection See Analysis 2.3 In the cystectomy group the oestradiol level on the day of hCG injection was 1170 ± 417.14 pg/ml versus 1680 ± 428.69 pg/ml in the expectant management group (Demirol 2006). Expectant management showed a positive treatment effect compared to cystectomy (MD ‐510.00, 95% CI ‐676.62 to ‐343.38). In the aspiration group the estradiol level was 1632 ± 670 pg/ml versus 946.7 ± 264 pg/ml in the expectant management group (Pabuccu 2004). Aspiration showed a positive treatment effect compared to expectant management on the ovarian response to COH (WMD 685.3, 95% CI 464.50 to 906.10). 2.4 Miscarriage See Analysis 2.4 In the aspiration group there were four miscarriages out of 41 women compared with four out of 40 women in the expectant management group (Pabuccu 2004). There was no difference in occurrence of miscarriage between the two groups (Peto OR 0.97, 95% CI 0.23 to 4.15). Heterogeneity was not applicable as there was only one trial for this analysis. 3. Surgery versus different surgery prior to ART (ablation versus cystectomy) Live birth Live birth was not reported. 3.1 Clinical pregnancy See Analysis 3.1 In the ablation group there were seven out of 24 women with a positive pregnancy test versus 15 out of 41 women in the cystectomy group (Alborzi 2007). Statistical analysis showed that one treatment did not show a more positive effect than the other (Peto OR 0.72, 95% CI 0.25 to 2.08). The CI was wide and crossed zero. Heterogeneity was not applicable because there was only one trial for this analysis. 3.2 Number of mature oocytes retrieved See Analysis 3.2 In the ablation group the number of mature oocytes retrieved was 2.6 ± 1.6 versus 3.2 ± 1.2 in the cystectomy group (Alborzi 2007). Statistical analysis showed that there was no evidence of difference in treatment effect between ablation and cystectomy (Peto OR ‐0.60, 95% CI ‐1.32 to 0.12). The CI was wide and crossed zero. Heterogeneity was not applicable because there was only one trial for this analysis. The following outcomes were not reported in any of the four included trials: quality of life, pain, ectopic pregnancy, drug side effects, multiple pregnancies, OHSS rate and post‐ovum pick up infection.

Summary of main results The aim of the this review was to determine the best strategy for women undergoing an ART cycle in the presence of endometriomata. Therapies considered included surgery, medicines, expectant management or a combination. We did not identify any randomised controlled trials for inclusion where the approach to the management of the endometriomata was by combination therapy. After assessment of risk of bias of the trials, we included four trials involving 312 women. No data were suitable for pooling and none of the trials reported live birth outcomes. One trial compared GnRH agonist with GnRH antagonist and there was no evidence of a difference in their clinical pregnancy. However, the number of mature oocytes retrieved was greater with the GnRH agonist (MD ‐1.60, 95% CI ‐2.44 to ‐0.76) and the ovarian response (E2 levels on day of hCG injection) was increased (MD ‐456.30, 95% CI ‐896.06 to ‐16.54). Two trials compared surgery (aspiration and cystectomy) with expectant management. There was no evidence of a benefit with one or the other technique for clinical pregnancy. Aspiration was associated with greater numbers of mature oocytes retrieved (MD 0.50, 95% CI 0.02 to 0.98) and increased ovarian response (E2 levels on day of hCG injection) (mean difference 685.3, 95% CI 464.50 to 906.10) compared to expectant management. Cystectomy was associated with a decreased ovarian response to COH (MD ‐510.00, 95% CI ‐676.62 to ‐343.38) and there was no evidence of an effect of cystectomy on the number of mature oocytes retrieved compared to expectant management. One trial compared aspiration with cystectomy and there was no evidence of a difference in clinical pregnancy rate or in the number of mature oocytes retrieved. Overall completeness and applicability of evidence We were primarily interested in the outcome live birth. A longer duration of follow up is necessary to assess live births than in assessing clinical pregnancy. All trials lasted at least 36 months, which is enough time to account for live births. All the trials reported clinical pregnancy rate but, as pregnancy loss occurs, this can not be extrapolated to live births. None of the trials reported quality of life or any of the adverse events of pain, ectopic pregnancy, drug side effects, multiple pregnancies, OHSS rate and post‐ovum pick up infection. Pain and multiple pregnancies are common adverse events in women with endometrioma and IVF treatment. The other adverse events are less common, nevertheless they are of great significance to the participant. Therapies that have overall high pregnancy rates but which are accompanied with severe adverse events are not desirable for women or clinicians alike. The ovarian response to COH and the number of mature oocytes retrieved are secondary outcomes to this review but provide some useful clinical information about ovulation induction and the resulting oocytes. Quality of the evidence The quality of evidence was graded with the five factors of the grades of recommendation, assessment, development and evaluation approach (developed by the GRADE Working Group 2004) (Guyatt 2008a; Guyatt 2008b;Schünemann 2006b ). The first factors that the GRADE approach addresses, to rate the quality of evidence, are the limitations in the design and implementation of available trials. All four trials meeting our criteria for inclusion had methodological limitations. It was not possible to draw one overall conclusion regarding the methodological quality for each individual trial and quality was therefore assessed per methodological item. The methodological quality graph (Figure 1) shows that there was a low risk of bias amongst three of the included trials regarding sequence generation. Sequence generation was adequately done with computerized randomisation. In the trial by Alborzi 2007 there was a substantial difference in the number of people in the two groups (24 versus 41). Therefore we asked the author if there was a special kind of randomisation method used. The author replied that: 'the method of randomisation for women with unilateral endometrioma (group 1 and 2) was computerized numbering, and that the difference was expected because of loss to follow up'. All women were analysed in the group they were initially randomised to. Therefore we do not think the difference can be caused by losses to follow up and we think this as a major source of bias. Allocation concealment is a very important item in determining methodological quality and was initially unclear in all four trials. We had to get back to the authors for clarification. The allocation concealment was adequately done in all four trials, with the use of opaque envelopes. Potential bias was derived from the fact that in two trials the participants and clinicians were not blinded (Demirol 2006; Pabuccu 2004). We have to put this in perspective because in both trials an expectant management group was compared with a group that underwent surgery. In this case it is only possible to blind the outcome assessors; the clinician could be blinded to the patient data, but this most often does not happen. We accept this bias, as we mentioned in the protocol beforehand. Incomplete outcome data in all four trials (Alborzi 2007; Demirol 2006; Pabuccu 2004; Pabuccu 2007) were adequately addressed and therefore comprised a low risk of bias. A low risk of bias was derived from loss to follow up. In two trials there were no losses to follow up (Demirol 2006; Pabuccu 2004) and in one trial an explanation was given for the losses (Pabuccu 2007). The trial of Alborzi 2007 did not show any losses, but as explained before we do think there was loss to follow up. In all four trials there was a high risk of bias from selective reporting of outcomes. None of the trials reported the live birth primary outcome, which is a major flaw for trials that concern fertility. Besides 'miscarriage', in two trials (Pabuccu 2004; Pabuccu 2007), no other adverse outcomes were reported in any of the included trials. There was attrition bias in one trial because no intention‐to‐treat analysis was performed (Pabuccu 2007). In this trial 20 women were excluded after randomisation and not included in the final analysis. In the other three trials the authors stated that they did not perform an intention‐to‐treat analysis (Alborzi 2007; Demirol 2006; Pabuccu 2004). Nevertheless, were all participants analysed in the groups they were first randomised to and none of them switched between treatment groups. Other sources of bias were noted in Pabuccu 2007 as it excluded one of its original four groups from the trial because advanced stage of disease had adverse effects on tubal function. The second factor that the GRADE approach addresses in rating the quality level of evidence is the indirectness of evidence. Amongst other things this embraces the comparators used within a trial. In two trials (Demirol 2006; Pabuccu 2004) the intervention group underwent surgery (aspiration or cystectomy) and the comparison group expectant management. The latter will be less effective compared to the intervention in most settings. This was also the case in the trial by Pabuccu 2004 for the outcomes 'number of mature oocytes retrieved' and 'ovarian response to COH'. Nevertheless was cystectomy more effective then expectant management for the outcome 'ovarian response to COH' in the trial by Demirol 2006. Another form of indirectness was caused by the outcomes assessed; outcome measures that are not of direct practical importance but are believed to reflect outcomes that are important are called surrogate outcomes (Bandolier). Participant important data such as live birth were not available in any of the four trials, unlike pregnancy rate which was reported in all trials. Pregnancy rate is a surrogate outcome for live births. Surrogate outcomes are often used when clinical outcomes require long follow‐up periods. The follow‐up times were initially not reported in any of the trials, but the duration of all four trials was at least 36 months, which is enough time to account for life births. The third factor that the GRADE approach addresses is the presence of unexplained heterogeneity, or inconsistency of results. There were few trials in this review and they did not evaluate the same kind of therapy, this is called clinical heterogeneity. It was to be expected. Because of the clinical and statistical heterogeneity we could not pool the results in a meta‐analysis. We did pool the results of two trials for the outcome clinical pregnancy (Demirol 2006; Pabuccu 2004). The fourth factor of the GRADE approach considers imprecision of results. Few participants were included in all four trials, which caused wide confidence intervals. Wide confidence intervals are a measure of imprecision. Sometimes trials are small because the sponsor has excluded participant groups based on the fact that they show an unfavourable outcome for the product. This does not seem to be the case for most of the included trials in this review because three out of four trials were not sponsored (Demirol 2006; Pabuccu 2004; Pabuccu 2007). The last factor the GRADE approach addresses is publication bias. There seems to be no high probability of publication bias in this review. If there had been a large number of small, sponsored trials, then the heavy involvement of sponsors would raise questions of whether unpublished trials suggesting no benefit existed. As mentioned before, three out of four of the included trials were not sponsored. When abstracts presented at conference meetings are not published in full text, we consider that as direct evidence of publication bias. We performed a thorough search through the conference abstracts of the meetings that related to fertility problems. All the conference abstracts that could have been of any interest to this review were published as full text articles. In the protocol we mentioned that we would develop a funnel plot to detect any publication bias if we had at least 10 trials for inclusion. We ended up with four trials, which meant it was not possible to draw a funnel plot. Regarding the objective of this review, to determine whether surgical, medical, combination therapy or no treatment was better in the management of women with endometriomata prior to ART, we can not draw an overall conclusion with the evidence we have at present. There was too much clinical and statistical heterogeneity between the trials to compare their results in a meta‐analysis. The included trials also did not provide relevant data to draw a conclusion concerning our primary outcome, live birth. The surrogate outcome pregnancy rate is inadequate. Potential biases in the review process The methods section in this review gives a broad description of the methods we used to try to prevent the introduction of any form of bias. The authors did their searching, trial selection, data collection and analysis independently from each other. They are confident that this was done in the correct way and that they did not introduce any bias such as publication bias. If there were any uncertainties or missing data, we contacted the authors of the trials to provide us with the answers. All four trials introduced bias by selective reporting of outcomes. None of the trials reported live birth, they all reported the surrogate outcome pregnancy rate. A limitation in this review is that we could not provide any results regarding our primary outcome live birth. Agreements and disagreements with other studies or reviews The Cochrane review 'Excisional surgery versus ablative surgery for ovarian endometriomata' determined the most effective technique for treating an ovarian endometrioma; either excision of the cyst capsule or drainage and electrocoagulation of the cyst wall (Hart 2008). One of the endpoints assessed was the ovarian response to stimulation, addressed by the number of follicles developed with clomiphene and HMG intervention. The forest plot shows that cystectomy was favoured over ablation, based on the results of one article (Alborzi 2007). We consider this outcome as uncertain because the results are based on data that were analysed per cycle while the randomisation was done per woman. We did not detect a major difference in benefit on fertility outcomes between the use of GnRH agonist and GnRH antagonist. These results seem to be in accord with the results shown in the Cochrane review by Al‐Inany (Al‐Inany 2006). This review showed that GnRH has slightly improved pregnancy outcomes with GnRh agonist compared to GnRH antagonist cycles, although ovarian hyperstimulation was reduced with GnRH antagonists. The trial by Yamamoto 2009 is not used for discussion in this review because this trial is a retrospective analysis. Authors' conclusions Implications for practice. There is no evidence that the use of a GnRH antagonist for endometriomata prior to ART provides a more favourable outcome than the use of a GnRH agonist with regard to the clinical pregnancy rate and the occurrence of miscarriage. There was evidence that the use of a GnRH agonist had improved outcomes on the ovarian response to COH and the number of oocytes retrieved compared to a GnRH antagonist . Laparoscopic aspiration or cystectomy of endometriomata prior to ART did not show evidence of benefit over expectant management with regard to the clinical pregnancy rate. In one trial there was evidence that laparoscopic aspiration improved the ovarian response and had a positive treatment effect on the number of mature oocytes retrieved compared to a GnRH antagonist (Pabuccu 2004). One trial showed that the ovarian response to COH was greater after expectant management than after cystectomy (Demirol 2006). There is no evidence of effect that aspiration of endometriomata prior to ART provides an increase in outcome compared to expectant management on the clinical pregnancy rate and the occurrence of miscarriage. Implications for research. Future trials comparing the following interventions for the management of women with endometriomata prior to ART should be undertaken. Medical versus medical + surgical Medical versus expectant management or placebo Surgical versus medical + surgical Surgical versus medical Surgical + medical versus expectant management or placebo Future trials that investigate therapies for the management of endometriomata prior to ART should also include the following interventions. Excision or ablation of endometriotic implants and adhesiolysis Antiprogestins e.g. danazol and gestrinone Progestagens Oral contraceptives These upcoming trials should account for live births, therefore the duration of the follow up should be long enough to result in a live birth. Other outcomes such as ectopic pregnancy, drug side effects, multiple pregnancies, OHSS rate, post‐ovum pick up infection, quality of life, pain and the recurrence rate of endometriomata should also be investigated. 'Egg quality' should also be addressed as an outcome. We hope to look at egg quality as an outcome in an update of this review. Upcomming trials and reviews should pay attention to cost and availability of the various interventions. These could be important if there is no difference between different interventions in reproductive outcomes. Feedback Control intervention in Demirol 2006 Summary The authors report that Demirol 2006 compared cystectomy with expectant management. However, the Demirol paper states that 'for patients in group II, endometrioma was aspirated at the time of oocyte retrieval and cytological diagnosis was performed'. Although no intervention occurred up until oocyte retrieval, the subsequent aspiration of the endometrioma(ta) may have had an effect on the receptivity of the endometrium and the implantation of the embryo(s). I therefore question the validity of classifying this treatment arm as receiving expectant management. Reply Thank you for your comment. The authors agree that this point needs addressing and will do so in the next update of the review, due September 2012. Contributors Comment by Dr Luk Rombauts, Infertility specialist (30.6.11). Response by L Benschop, C Farquhar, N van der Poel, MJ Heineman (1.12.11) What's new | Date | Event | Description | |---|---|---| | 1 December 2011 | Feedback has been incorporated | Feedback received (summarised below), which will be addressed in the 2012 update of this review | History Protocol first published: Issue 7, 2010 Review first published: Issue 11, 2010 | Date | Event | Description | |---|---|---| | 15 September 2005 | New citation required and major changes | Substantive amendment |

Marian Showell and Jane Clarke of the Cochrane Menstrual Disorders and Subfertility Group assisted in the trial search and provided general support. Appendices Appendix 1. MEDLINE search strategy 1 exp adnexal diseases/ or exp endometriosis/ (91965) 2 adnexal disease$.tw. (120) 3 exp Ovarian Cysts/ (12505) 4 Ovar$ Cyst$.tw. (3683) 5 (endometrioma$ or endometrioid or endometrial cyst$).tw. (3800) 6 (endometrial tumour$ or endometrial tumor$).tw. (513) 7 blood cyst$.tw. (95) 8 chocolate cyst$.tw. (87) 9 ovar$ tumo?r$.tw. (8324) 10 (ovar$ adj2 neoplasm$).tw. (1434) 11 or/1‐10 (95149) 12 exp fertilization in vitro/ or exp sperm injections, intracytoplasmic/ (22856) 13 in vitro fertili?ation$.tw. (13189) 14 invitro fertili?ation$.tw. (8) 15 (ivf or icsi).tw. (13654) 16 intracytoplasmic sperm injection$.tw. (3525) 17 intra cytoplasmic sperm injection$.tw. (83) 18 or/12‐17 (28379) 19 11 and 18 (2356) 20 randomised controlled trial.pt. (280513) 21 controlled clinical trial.pt. (80002) 22 randomized.ab. (190268) 23 placebo.tw. (118525) 24 clinical trials as topic.sh. (146365) 25 randomly.ab. (138281) 26 trial.ti. (82343) 27 (crossover or cross‐over or cross over).tw. (43892) 28 or/20‐27 (665943) 29 (animals not (humans and animals)).sh. (3344852) 30 28 not 29 (615198) 31 19 and 30 (274) 32 from 31 keep 1‐274 (274) Appendix 2. CENTRAL search strategy 1 exp adnexal diseases/ or exp endometriosis/ (2451) 2 adnexal disease$.tw. (0) 3 exp Ovarian Cysts/ (582) 4 Ovar$ Cyst$.tw. (96) 5 (endometrioma$ or endometrioid or endometrial cyst$).tw. (67) 6 (endometrial tumour$ or endometrial tumor$).tw. (5) 7 blood cyst$.tw. (2) 8 chocolate cyst$.tw. (1) 9 ovar$ tumo?r$.tw. (48) 10 (ovar$ adj2 neoplasm$).tw. (15) 11 or/1‐10 (2550) 12 exp fertilization in vitro/ or exp sperm injections, intracytoplasmic/ (1292) 13 in vitro fertili?ation$.tw. (1203) 14 invitro fertili?ation$.tw. (3) 15 (ivf or icsi).tw. (1927) 16 intracytoplasmic sperm injection$.tw. (343) 17 intra cytoplasmic sperm injection$.tw. (5) 18 or/12‐17 (2601) 19 11 and 18 (186) 20 from 19 keep 1‐186 (186) Appendix 3. EMBASE search strategy 1 adnexa disease/ (1659) 2 exp endometrium tumor/ (22498) 3 adnexa$ disease$.tw. (72) 4 endometrium tumor$.tw. (4) 5 (endometrioma$ or endometrioid or endometrial cyst$).tw. (3566) 6 (endometrial tumour$ or endometrial tumor$).tw. (473) 7 blood cyst$.tw. (74) 8 chocolate cyst$.tw. (82) 9 ovar$ tumo?r$.tw. (6833) 10 or/1‐9 (32028) 11 exp fertilization in vitro/ (22368) 12 exp intracytoplasmic sperm injection/ (6387) 13 in vitro fertili?ation$.tw. (11990) 14 invitro fertili?ation$.tw. (17) 15 (ivf or icsi).tw. (13720) 16 intracytoplasmic sperm injection$.tw. (3482) 17 intra cytoplasmic sperm injection$.tw. (84) 18 or/11‐17 (27891) 19 10 and 18 (370) 20 Clinical Trial/ (577791) 21 Randomized Controlled Trial (181145) 22 exp randomisation/ (27508) 23 Single Blind Procedure/ (9037) 24 Double Blind Procedure/ (76095) 25 Crossover Procedure/ (22410) 26 Placebo/ (137708) 27 Randomi?ed controlled trial$.tw. (37714) 28 Rct.tw. (3236) 29 random allocation.tw. (669) 30 randomly allocated.tw. (10800) 31 allocated randomly.tw. (1393) 32 (allocated adj2 random).tw. (572) 33 Single blind$.tw. (7925) 34 Double blind$.tw. (88703) 35 ((treble or triple) adj blind$).tw. (148) 36 placebo$.tw. (116204) 37 prospective study/ (90605) 38 or/20‐37 (758486) 39 case study/ (6833) 40 case report.tw. (126897) 41 abstract report/ or letter/ (525867) 42 or/39‐41 (657021) 43 38 not 42 (732163) 44 19 and 43 (73) 45 (2008$ or 2009$ or 2010$).em. (1350536) 46 44 and 45 (22) 47 from 46 keep 1‐22 (22) Appendix 4. MDSG search strategy

CONTAINS "endometrioma" or "endometriotic cysts" or "ovarian cyst" or "Ovarian Cysts" or "ovarian endometrioma" or Title CONTAINS"endometrioma" or "endometriotic cysts" or "ovarian cyst" or "Ovarian Cysts" or "ovarian endometrioma" AND

CONTAINS "ivf" or "icsi" or "in vitro fertilisation" or "in‐vitro fertilisation procedure" or "in‐vitro fertilisation techniques" or "in vitro fertilization" or "intracytoplasmic sperm injection" or "intracytoplasmic sperm injection cycle" or "intracytoplasmic sperm injection techniques" or "Sperm Injections, Intracytoplasmic" or Title CONTAINS "ivf" or "icsi" or "in vitro fertilisation" or "in‐vitro fertilisation procedure" or "in‐vitro fertilisation techniques" or "in vitro fertilization" or "intracytoplasmic sperm injection" or "intracytoplasmic sperm injection cycle" or "intracytoplasmic sperm injection techniques" or "Sperm Injections, Intracytoplasmic" There were 25 results Appendix 5. PsycINFO search strategy 1 adnexa$ disease$.tw. 2 endometrium tumor$.tw. 3 (endometrioma$ or endometrioid or endometrial cyst$).tw. 4 (endometrial tumour$ or endometrial tumor$).tw. 5 blood cyst$.tw. 6 chocolate cyst$.tw. 7 ovar$ tumo?r$.tw. 8 or/1‐7 9 exp reproductive technology/ 10 in vitro fertili?ation$.tw. 11 invitro fertili?ation$.tw. 12 (ivf or icsi).tw. 13 intracytoplasmic sperm injection$.tw. 14 intra cytoplasmic sperm injection$.tw. 15 or/9‐14 16 8 and 15 There were no results. Data and analyses Comparison 1. GnRH antagonist versus GnRH agonist prior to ART. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Clinical pregnancy | 1 | Peto Odds Ratio (Peto, Fixed, 95% CI) | Subtotals only | | | 2 Number of mature oocytes retrieved | 1 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | | | 3 Ovarian response to COH (oestradiol pg/ml on day of hCG injection) | 1 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | | | 4 Miscarriage | 1 | Peto Odds Ratio (Peto, Fixed, 95% CI) | Subtotals only | Comparison 2. Surgery (aspiration or cystectomy) versus expectant management prior to ART. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Clinical pregnancy | 2 | Peto Odds Ratio (Peto, Fixed, 95% CI) | Subtotals only | | | 1.1 Aspiration | 1 | 81 | Peto Odds Ratio (Peto, Fixed, 95% CI) | 1.29 [0.45, 3.64] | | 1.2 cystectomy | 1 | 109 | Peto Odds Ratio (Peto, Fixed, 95% CI) | 1.15 [0.52, 2.55] | | 2 Number of mature oocytes retrieved | 2 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | | | 2.1 Aspiration | 1 | 81 | Mean Difference (IV, Fixed, 95% CI) | 0.5 [‐0.00, 1.00] | | 2.2 Cystectomy | 1 | 99 | Mean Difference (IV, Fixed, 95% CI) | ‐0.80 [‐1.96, 0.36] | | 3 Ovarian response to COH (oestradiol pg/ml on day of hCG injection) | 2 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | | | 3.1 Aspiration | 1 | 81 | Mean Difference (IV, Fixed, 95% CI) | 685.3 [464.50, 906.10] | | 3.2 Cystectomy | 1 | 99 | Mean Difference (IV, Fixed, 95% CI) | ‐508.00 [‐676.62, ‐343.38] | | 4 Miscarriage | 1 | Peto Odds Ratio (Peto, Fixed, 95% CI) | Subtotals only | | | 4.1 Aspiration | 1 | 81 | Peto Odds Ratio (Peto, Fixed, 95% CI) | 0.97 [0.23, 4.15] | Comparison 3. Ablation versus cystectomy. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Clinical pregnancy | 1 | Peto Odds Ratio (Peto, Fixed, 95% CI) | Subtotals only | | | 2 Number of mature oocytes retrieved | 1 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | Characteristics of studies Characteristics of included studies [ordered by study ID] Alborzi 2007. | Methods | Randomised multi‐centre trial, conducted in two universities and private hospitals. Randomisation was done by computerized numbering and the personnel who performed the randomisation process was not involved in the trial. The allocation concealment was adequately done with the use of serial opaque envelopes. | | | Participants | Sixty‐five Iranian women from one infertility centre. They were diagnosed with single unilateral endometrioma by vaginal ultrasound, after which they were operated on by one out of two surgeries. After surgery they underwent ovulation induction with one single method and IUI at two universities and private hospitals. Inclusion criteria: endometrioma size ≥ 3 cm, single endometrioma, subfertility, reference for COH, BMI within normal range. Group 1 and 2 were comparable in terms of age, duration of subfertility and type of menstruation. Mean age in year in group 1 was 24.4 and in group 2 was 26.2. Duration of subfertility in years in group one was 5.9 and in group 2 was 5.1. The diameter of the endometriomata in mm in group 1 was 36.0 and in group 2 was 44.0. The American Fertility Society (AFS) score for classification of endometriosis severity at the time of laparoscopy in group 1 was 43 and in group 2 was 45. The number of cycles in group 1 was 40 and in group 2 was 70. So each woman had at least two COH cycles if pregnancy did not occur in the first cycle. The mean number of dominant follicles achieved during COH after treatment in group 1 was 2.6 ± 1.6 and in group 2 was 3.2 ± 1.1. The clinical pregnancy rate in group 1 was 7 and in group 2 was 15. Exclusion criteria: previous pelvic operation, other causes of infertility, previous IVF cycle, previous hormonal therapy. | | | Interventions | Group 1 consists of 24 women with unilateral endometrioma who underwent laparoscopic fenestration and coagulation (ablation). Group 2 consists of 41 women with unilateral endometrioma who underwent laparoscopic cystectomy. All laparoscopic operations were performed by the first author. All women had day 3 FSH assay before and one cycle after the operation, which showed normal ovarian reserve. All women were stimulated with the same method for at least 2 months after surgery. Clomiphene citrate (Iran Hormone, Tehran, Iran) was started in a 100 mg/daily dose from day 5 for 5 consecutive days of menstrual cycle. hMG (Menogon; Ferring, Mannheim, Germany) was injected from day 8 of menstrual cycle (2 ampoules/day). Ultrasound follicular monitoring was performed from day 10 or 11 of each cycle. When two follicles with an average diameter ≥ 18 mm were detected, ovulation was induced with 10,000 units of hCG. IUI with washed sperm was 34‐36 hours later performed. Duration of the trial was 44 months. The follow‐up time was two cycles. | | | Outcomes | Pregnancy rate, number of dominant follicles achieved during COH after using different methods of laparoscopy for treatment of endometrioma. | | | Notes | The trial was ethically approved by the Shiraz University of Medical Sciences Institutional Review Board. There was no power calculation done. The trial protocol is available. The trial received additional funding; it was supported by Shiraz University, as a standard therapeutic protocol. We asked the author if there was a special kind of randomisation method used because there was a difference in the amount of people between group 1 (n=24) and 2 (n=41). The author replied the following: 'the method of randomisation for women with unilateral endometrioma (group 1 and 2) was computerized numbering, and that difference was expected because of loss to follow up.' All women were analysed in the group they were initially randomised to. Therefore the difference can not be caused by the loss to follow up. So unfortunately we could not find a reason for the difference in the numbers of women between group 1 and 2. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Random sequence generation (selection bias) | Unclear risk | The trial reports that randomisation was done with computerized numbering for patients with unilateral endometrioma. The author responded that the personnel who performed the randomisation process were not involved in the trial. Nevertheless there is a big difference in the amount of people between group 1 and 2 (24 versus 41). Therefore we asked the author if there was a special kind of randomisation method used. The author replied the following: 'the method of randomisation for women with unilateral endometrioma (group 1 and 2) was computerized numbering, and that difference was expected because of loss to follow up.' All women were analysed in the group they were initially randomised to. Therefore we do not think the difference can be caused by the loss to follow up. | | Allocation concealment (selection bias) | Low risk | Serial opaque envelopes. | | Blinding (performance bias and detection bias) All outcomes | High risk | The clinician, patients, investigators and outcome assessors were all not blinded. The all could have been blinded. The clinician could not have been blinded for the surgery itself, but could have been blinded for the patient data, if the clinician and the investigator would not have been one and the same person. | | Incomplete outcome data (attrition bias) All outcomes | Low risk | There were no drop‐outs and no participants were excluded after randomisation. | | Selective reporting (reporting bias) | High risk | Contacted the author with the question: 'have you also reported other outcome measures?' His answer was: 'no'. Live birth was not an outcome in the trial. | | Other bias | Unclear risk | Attrition bias: the author responded that there was no intention to treat analysis performed. Nevertheless all participants were analysed in the groups they were first randomised in and none of them switched treatment groups. | Demirol 2006. | Methods | Randomised single‐centre trial. The randomisation was computer assisted and the personnel who performed the randomisation process were not involved in the trial. The allocation concealment was adequately done with the use of opaque envelopes. | | | Participants | Turkish women from one infertility centre. Ninety‐nine women with endometriomata with a diameter of ≥ 3 cm and < 6 cm awaiting to undergo intracytoplasmic sperm injection (ICSI). All women had endometriomata with a diameter of ≥ 3 cm and < 6 cm. The male factor infertility rate was similar in both groups. There was no significant difference between both groups with regard to body mass index (BMI). Sixty‐four per cent had a diagnosis of minimal endometriosis in their medical history. The rate of male factor infertility was similar in groups 1 and 2 (59.18% versus 62.00%). Mean age in year in group 1 was 35.2 ± 0.3 and in group 2 was 34.9 ± 0.2. The number of unilateral endometriomata in group 1 was 32 and in group 2 was 35. The number of bilateral endometriomata in group 1 was 17 and in group 2 was 15. The number of endometrioma in group 1 was 36 and in group 2 was 34. The number of endometriomata in group 1 was 13 and in group 2 was 16. Basal FSH (mIU/ml) in group 1 was 8.2 ± 0.38 and in group 2 was 7.9 ± 0.36. Number of cycles in group 1 was 49 and in group 2 was 50. The duration of sub fertility in group 1 was 3.3 ± 1.1 and in group 2 was 3.0 ± 1.0. Exclusion criteria: multiple bilateral endometriomata with a diameter between 3 and 6 cm, ovarian surgery before the trial. Diagnosis by transvaginal ultrasonography of particular neoplastic features, acute haemorrhage, multi‐ovularly or hyperechoic wall foci in the ovaries. Women in whom sutures were used for any reason during laparoscopy. | | | Interventions | Group 1: laparoscopic drainage of the endometrioma cyst and dissection of the pseudocapsule of the endometrioma 3 months before the ICSI cycle (49 cases). This is cystectomy. Group 2: underwent ICSI cycle directly (50 cases). All women were stimulated with luteal long protocol. Leuprolide acetate 1.0 mg was started at the luteal phase. Recombinant FSH was administered in a step‐down fashion, starting with 300 IU/day by the documentation of suppression during menses; after 5 days the dose was adjusted according to the ovarian response. HCG 10,000 IU IM was administered when at least two or three follicles reached a mean diameter of 17 mm and the serum estradiol concentration was >500 pg/ml. Transvaginal oocyte retrieval was scheduled 36 h after HCG injection. ICSI was performed for all metaphase II oocytes and embryo transfers were performed on day 3 for all women under ultrasound guidance. Duration of the trial was 50 months | | | Outcomes | Total FSH dosage (IU) consumed per cycle, stimulation days (day), peak estradiol (pg/ml), number of mature oocytes retrieved, fertilization rate in %, number of embryos transferred, implantation rate in % and clinical pregnancy rate in %. | | | Notes | The trial was ethically approved and there was no additional funding for the trial. A power calculation was performed. They chose to focus on oocyte number as the primary end point (p1=8 for group II, p2=7 for group I), it has been calculated that a sample size of 45 women in each group will have an 87.85% power to detect a difference in means of 1.0 (SD=1.5), using a t‐test, with P=0.05 two‐sided significance level. The trial did not receive additional funding. Additional data were retrieved from the conference proceeding for this trial. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Random sequence generation (selection bias) | Low risk | Computer assisted. The personnel who performed the randomisation process were not involved in the trial. | | Allocation concealment (selection bias) | Low risk | Serial number opaque envelopes. | | Blinding (performance bias and detection bias) All outcomes | High risk | The participants and clinicians could not be blinded because the intervention was surgery. The outcome assessor and investigators were also not blinded. | | Incomplete outcome data (attrition bias) All outcomes | Low risk | There were no drop‐outs and no participants were excluded after randomisation. | | Selective reporting (reporting bias) | High risk | Contacted the author with the question: 'have you also reported other outcome measures?' His answer was: 'no'. Live birth was not an outcome in the trial. | | Other bias | Low risk | Attrition bias: the author responded that there was no intention to treat analysis performed. Nevertheless all participants were analysed in the groups they were first randomised in and none of them switched treatment groups. | Pabuccu 2004. | Methods | Randomised single‐centre trial. Randomisation was done by computer in blocks of 10. The randomisation was computer assisted and the personnel who performed the randomisation process was not involved in the trial. The allocation concealment was adequately done with the use of opaque envelopes. | | | Participants | Turkish women from one centre. Age range 25‐35 years. Duration of subfertility was 3.5‐9.5 years. Group 1 consisted of women with endometriomata and no history of previous surgery (41 women). Group 2 consisted of women with non‐aspirated endometriomata (40 women). Inclusion criteria: Age ≤ 39 years. Mean age in year in group 1 was 30.2 ± 4.9 and in group 2 was 30.1 ± 4.5. Duration of subfertility in years in group one was 6.1 ± 3.3 and in group 2 was 5.6 ± 3. Body mass Index in kg/m2 in group 1 was 25.9 ± 3 and in group 2 was 25.2 ± 3.2. The diameter of the endometriomata in mm in group 1 was 22.3 ± 9.2 and in group 2 was 26.4 ± 10.9. The number of unilateral endometriomata in group 1 was 32 and in group 2 was 29. The number of bilateral endometriomata in group 1 was 9 and in group 2 was 11. The number of endometrioma in group 1 was 28 and in group 2 was 25. The number of endometriomata in group 1 was 13 and in group 2 was 15. The number of cycles in group 1 was 41 and in group 2 was 40. When the female age was < 30 years, up to two embryos were transferred. When the female age was between 30 and 36 years, up to three embryos were transferred. In group 1 were 14/41 women < 30 years of age and in group 2 was 12/40 < 30 years of age. Exclusion criteria: thaw cycles, male factor infertility, hydrosalpinx, previous surgery for control group, all other causes of subfertility and FSH: 20 IU/ml. | | | Interventions | Group 1: transvaginal aspiration of endometriomata at the beginning of COH. A single dose of ceftazidime 2 g was injected IM after aspiration. Group 2: expectant management, blood samples for CA125 were taken from group 2 women on day 3 of the menstrual cycle to confirm the diagnosis of endometriomata. All women underwent the same pituitary down‐regulation, ovarian stimulation, oocyte retrieval, embryo transfer, implantation and luteal support protocols. Triptolein 100 μg starting on day 21 of the previous cycle. COH with recombinant FSH 300 IU was started on the third day of the menstrual cycle. After the development of at least three follicles larger than 17 mm and serum E2 levels higher than 500 pg/mL, 10,000 IU hCG (Profasi; Serono) was injected. Oocyte retrieval was performed 35 hours after hCG injection. An embryo transfer was done on day 3 or 5. Methylprednisolone (16 mg/day) and doxycycline (100 mg bid.) were administered to all women as a routine protocol for 4 days, starting on the day of oocyte retrieval, to support implantation. All women had luteal support with micronized vaginal P after ET. Duration of the trial was 44 months. Follow‐up time was at least one year in each group. | | | Outcomes | Serum levels of day 2 FSH (IU/mL), day 2 E2 (pg/mL), number of antral follicles, duration of COH (days), E2 level on hCG day (pg/mL), total recombinant FSH ampoules per cycle (n), total number of follicles > 17mm, the average number of embryos transferred, mean fertilization rate in %, metaphase II oocytes, implantation rate in %, clinical pregnancy in % and spontaneous abortion in %. | | | Notes | The trial was ethically approved by the Ethics Committee of the Gulhane School of Medicine. There was no additional funding for the trial. There was no power calculation done. The author explained that the frequency of disease was to low to make a power calculation. The trial protocol was not available. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Random sequence generation (selection bias) | Low risk | Randomisation by computer in blocks of 10. The personnel who performed the randomisation process was not involved in the trial. | | Allocation concealment (selection bias) | Low risk | With the use of opaque envelopes. | | Blinding (performance bias and detection bias) All outcomes | High risk | The participants and clinicians could not be blinded because the intervention was surgery. The outcome assessor was blinded. | | Incomplete outcome data (attrition bias) All outcomes | Low risk | There were no drop‐outs and there were no women excluded after randomisation. | | Selective reporting (reporting bias) | High risk | Contacted the author with the question: 'have you also reported other outcome measures?' His answer was: 'no'. Live birth was not an outcome in the trial. | | Other bias | Low risk | Attrition bias: the author responded that there was no intention to treat analysis performed. Nevertheless all participants were analysed in the groups they were first randomised in and none of them switched treatment groups. The author responded that loss to follow up was not accounted for, but there was no loss to follow up. | Pabuccu 2007. | Methods | Randomised single centre trial. Randomisation was done according to the randomisation list generated by the computer. The personnel who performed the randomisation process was not involved in the trial. The allocation concealment was adequately done with the use of opaque envelopes. | | | Participants | Turkish women from one centre. 67 Women with unilateral or bilateral endometriomata and no history of previous ovarian surgery. Mean age in years in the GnRH antagonist group (group1) was 32.8 ± 5.2 and in the GnRH agonist group (group 2) was 31.2 ± 3.6. Duration of subfertility in years in group 1 was 8.4 ± 4.3 and in group 2 was 7.1 ± 3.7. Body mass Index in kg/m2 in group 1 was 24.5 ± 4.1 and in group 2 was 24.3 ± 5.1. The diameter of the endometriomata in mm in group 1 was 25.3 ± 11.2 and in group 2 was 26.9 ± 14.1. The number of unilateral endometriomata in group 1 was 20 and in group 2 was 22. The number of bilateral endometriomata in group 1 was 14 and in group 2 was 11. The number of endometriomata in group 1 was 10 and in group 2 was 13. The number of endometriomata in group 1 was 24 and in group 2 was 20. The number of antral follicles in the GnRH antagonist group was 4.3 ± 1.2 and in the GnRH‐agonist group was 4.1 ± 1.9. Number of cycles in GnRH antagonist group was 34 and in the GnRH agonist group was 33. Inclusion criteria: a normal uterine cavity documented by either hysterosalpingography or hysteroscopy done no longer than 1 year before the trial. Exclusion criteria: women with hydrosalpinx, documented tuberculosis, male factor infertility, thaw cycles and previous history of ovarian surgery. | | | Interventions | Thirty‐three participants received COH with GnRH agonist long protocol vs 34 participants COH with multiple doses GnRH antagonist. Long protocol: triptolein 100 μg starting on day 21 of the previous cycle. Multiple doses protocol: daily injections of cetrorelix when the leading follicle was 14mm in diameter with serum levels of E2 exceeding 600 pg/mL For both protocols: COH with recombinant FSH 225‐300 IU was started on the third day of the first following menstrual cycle. After the development of at least three follicles larger than 17 mm and serum E2 levels higher than 500 pg/mL, 10,000 IU hCG (Profasi; Serono) was injected. Oocyte retrieval was performed 35 hours after hCG injection. An embryo transfer was done on day 3 or 5. Methylprednisolone (16 mg/day) and doxycycline (100 mg twice a day) were administered for 4 days, starting on the day of oocyte retrieval, to support implantation. All women had luteal phase support with micronized vaginal progesterone after embryo transfer. Duration of the trial was 39 months. | | | Outcomes | Serum levels of day 2 FSH (IU/mL), day 2 E2 (pg/mL), number of antral follicles, duration of COH (days), E2 level on hCG day (pg/mL), total recombinant FSH ampoules per cycle (n), the number of follicles > 17 mm, the number of retrieved oocytes, the number of metaphase II oocytes, the fertilization rate in %, the number of available embryos, the number of transferred embryos, the clinical pregnancy in %, the abortion rate in %, the implantation rate in %. | | | Notes | The trial was ethically approved by the local Ethics Committee. The trial received no additional funding. There is no power calculation performed because no data exist for the antagonist treatment before IVF‐ICSI cycles in women with advanced endometriosis. They could not calculate the required sample size for pretest power estimation. We asked the author if the participants included in this trial were the same participants included in Pabuccu 2004. The author responded that they were not the same participants. Additional data were retrieved from the conference proceeding of this trial. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Random sequence generation (selection bias) | Low risk | Randomisation list generated by computer. The personnel who performed the randomisation process was not involved in the trial. | | Allocation concealment (selection bias) | Low risk | They used sealed opaque envelopes. | | Blinding (performance bias and detection bias) All outcomes | Low risk | Participants, investigators, clinicians and outcome assessors were all blinded. | | Incomplete outcome data (attrition bias) All outcomes | Low risk | Initialy there were 266 eligible women enrolled and included in the trial. The group with unilateral or bilateral endometriomata and no history of previous ovarian surgery consisted of 75 participants. Of these 37 were randomised to GnRH agonist and 38 to GnRH antagonist. After randomisation and starting the treatment, 20 cases were cancelled because of insufficient ovarian response and risk of hyperstimulation (cycle cancellation). Of them were 8 in the group with unilateral or bilateral endometriomata and no history of previous ovarian surgery (n = 67; GnRH‐a, n = 33 and GnRH antagonist, n = 34). | | Selective reporting (reporting bias) | High risk | Live birth was not reported. | | Other bias | High risk | No intention to treat analysis was performed. The 20 women that were excluded after randomisation are not included in the final analysis. There was a fourth group before, this group was excluded from the trial because the advanced stage of disease had adverse effects on tubal function. | Characteristics of excluded studies [ordered by study ID] | Study | Reason for exclusion | |---|---| | Cooke 1989 | Does not include endometriomata and IVF | | Dicker 1992 | Includes endometriosis not endometriomata. | | Gelbaya 2009 | The trial has been aborted due to inability to recruit enough women | | Kim 1997 | trial groups are not comparable: they do not compare women who have endometriomata. | | Neveu 1987 | They excluded women with any form of endometriosis. | | Oehninger 1990 | trial groups are not comparable. | | Rizk 1990 | Ovarian cyst is not an endometrioma. | Differences between protocol and review For continuous data such as number of oocytes retrieved, mean differences (MD) between treatment groups were calculated if all trials reported exactly the same outcomes. The protocol described the weighted MD instead of the MD. Contributions of authors L Benschop took the lead in writing the review, with input from the other authors. Review authors (L Benschop and NA van der Poel) independently considered trials for inclusion, evaluated their risk of bias and extracted trial data. Differences in interpretation have been resolved by discussion or by a third author (MJ Heineman or C Farquhar). Sources of support Internal sources University of Amsterdam, Netherlands. External sources No sources of support supplied Declarations of interest None Edited (no change to conclusions), comment added to review

References to studies included in this review Alborzi 2007 {published and unpublished data} - Alborzi S. A comparison of follicular response of ovaries to ovulation induction after laparoscopic ovarian cystectomy or fenestration and coagulation versus normal ovaries in patients with endometrioma. Fertility and Sterility 2007;88(2):507‐9. [DOI] [PubMed] [Google Scholar] Demirol 2006 {published and unpublished data} - Demirol A. The effect of ovarian cystectomy of endometriosis on ART outcome. Middle East Fertility Society Journal 2005;10 Suppl 1:125. [Google Scholar] - Demirol A, Guven S, Baykal C, Gurgan T. Effect of endometrioma cystectomy on IVF outcome: a prospective randomised study. Reproductive BioMedicine Online 2006;12(5):639‐43. [DOI] [PubMed] [Google Scholar] Pabuccu 2004 {published and unpublished data} - Pabuccu R, Onalan G, Goktolga U, Kucuk T, Orhon E, Ceyhan T. Aspiration of ovarian endometriomas before intracytoplasmic sperm injection. Fertility and Sterility 2004;82(3):705‐11. [DOI] [PubMed] [Google Scholar] Pabuccu 2007 {published and unpublished data} - Pabuccu R, Onalan G, Kaya C. GnRH agonist and antagonist protocols for stage I‐II endometriosis and endometrioma in in vitro fertilization/intracytoplasmic sperm injection cycles. Fertility and Sterility 2007;88:832‐9. [DOI] [PubMed] [Google Scholar] - Pabuccu R, Onalan G, Salem B, Ceyhan T, Akar M, Onalan R. Comparison of GnRH agonist and antagonist protocols among patients with mild‐moderate endometriosis and endometrioma: a novel clinical approach. Fertility and Sterility 2005;84 Suppl 1:197‐8. [Google Scholar]

to studies excluded from this review Cooke 1989 {published and unpublished data} - Cooke ID, Thomas EJ. The medical treatment of mild endometriosis. Acta Obstetricia et Gynecologica Scandinavica. Supplement 1989;150:27‐30. [PubMed] [Google Scholar] Dicker 1992 {published and unpublished data} - Dicker D, Goldman JA, Levy T, Feldberg D, Ashkenazi J. The impact of long‐term gonadotropin‐releasing hormone analogue treatment on preclinical abortions in patients with severe endometriosis undergoing in vitro fertilization‐embryo transfer. Fertility and Sterility 1992;57(3):597‐600. [DOI] [PubMed] [Google Scholar] Gelbaya 2009 {published and unpublished data} - Gelbaya T. Management of endometriotic ovarian cyst before in vitro fertilisation (IVF). Current Controlled Trials 2009. Kim 1997 {published and unpublished data} - Kim CH, Chae HD, Kang BM, Chang YS, Mok JE. The immunotherapy during in vitro fertilization and embryo transfer cycles in infertile patients with endometriosis. The Journal of Obstetrics and Gynaecology Research 1997;23(5):463‐70. [DOI] [PubMed] [Google Scholar] Neveu 1987 {published and unpublished data} - Neveu S, Hedon B, Bringer J, Chinchole JM, Arnal F, Humeau C, et al. Ovarian stimulation by a combination of a gonadotropin‐releasing hormone agonist and gonadotropins for in vitro fertilization. Fertility and Sterility 1987;47(4):639‐43. [DOI] [PubMed] [Google Scholar] Oehninger 1990 {published and unpublished data} - Oehninger S, Rosenwaks Z. In vitro fertilization and embryo transfer: an established and successful therapy for endometriosis. Progress in Clinical and Biological Research 1990;323:319‐35. [PubMed] [Google Scholar] Rizk 1990 {published and unpublished data} - Rizk B, Tan SL, Kingsland C, Steer C, Mason BA, Campbell S. Ovarian cyst aspiration and the outcome of in vitro fertilization. Fertility and Sterility 1990;54(4):661‐4. [PubMed] [Google Scholar] Additional references Adamson 2003 - Adamson D. Surgical management of endometriosis. Seminars in Reproductive Medicine 2003;21:223‐34. [DOI] [PubMed] [Google Scholar] Al‐Inany 2006 - Al‐Inany HG, Abou‐Setta AM, Aboulghar M. Gonadotrophin‐releasing hormone antagonists for assisted conception.. Cochrane Database of Systematic Reviews 2006, Issue 3. Art. No.: CD001750. DOI: 10.1002/14651858.CD001750.pub2. 2006;3:Art. No.: CD001750. DOI: 10.1002/14651858.CD001750.pub2. [DOI] [PubMed] [Google Scholar] Bandolier - Surrogate endpoints. http://www.medicine.ox.ac.uk/bandolier/booth/glossary/surrog.html. Brosens 1994 - Brosens IA, Puttemans PJ, Deprest J. The endoscopic localization of endometrial implants in the ovarian chocolate cyst. Fertility and Sterility 1994;61:1034‐8. [DOI] [PubMed] [Google Scholar] Busacca 2006 - Busacca M, Riparini J, Somigliana E, Oggioni G, Izzo S, VignaliM, et al. Postsurgical ovarian failure after laparoscopic excision of bilateral endometriomas. American Journal of Obstetrics and Gynecology 2006;195:421‐5. [DOI] [PubMed] [Google Scholar] Cahill 2002 - Cahill DJ. What is the optimal medical management of infertility and minor endometriosis? Analysis and future prospects. Human Reproduction 2002;17:1135‐40. [DOI] [PubMed] [Google Scholar] Chan 2003 - Chan LY, So WW, Lao TT. Rapid recurrence of endometrioma after transvaginal ultrasound‐guided aspiration. European Journal of Obstetrics, Gynecology, and Reproductive Biology 2003;109:196‐8. [DOI] [PubMed] [Google Scholar] Cobellis 2004 - Cobellis L, Razzi S, Fava A, et al. A danazol‐loaded intrauterine device decreases dysmenorrhoea, pelvic pain, and dyspareunia associated with endometriosis. Fertility and Sterility 2004;82(1):239‐40. [DOI] [PubMed] [Google Scholar] Farquhar 1998 - Farquhar C, Sutton C. The evidence for the management of endometriosis. Current Opinion in Obstetrics and Gynecology 1998;10:321‐32. [DOI] [PubMed] [Google Scholar] Gilmour 2008 - Gilmour JA, Huntington A, Wilson HV. The impact of endometriosis on work and social participation. International Journal of Nursing Practice 2008;14:443‐8. [DOI] [PubMed] [Google Scholar] Guyatt 2008a - Guyatt GH, Oxman AD, Kunz R, Vist GE, Falck‐Ytter Y, Schünemann HJ. What is 'quality of evidence' and why is it important to clinicians?. BMJ 2008;336:995‐98. [DOI] [PMC free article] [PubMed] [Google Scholar] Guyatt 2008b - Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck‐Ytter Y, Alonso‐Coello P, Schünemann HJ. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008b;336:924‐26. [DOI] [PMC free article] [PubMed] [Google Scholar] Hart 2008 - Hart RJ, Hickey M, Maouris P, Buckett W. Excisional surgery versus ablative surgery for ovarian endometriomata. Cochrane Database of Systematic Reviews 2008, Issue 2. [DOI] [PubMed] [Google Scholar] Higgins 2009 - Higgins JPT, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.1 The Cochrane Collaboration, 2009. Available from www.cochrane‐handbook.org. Wiley‐Blackwell, 2009. [Google Scholar] Hughes 2010 - Hughes E, Brown J, Collins JJ, Farquhar C, Fedorkow DM, Vanderkerchove P. Ovulation suppression for endometriosis for women with subfertility (Review). Cochrane Database of Systematic Reviews 2010, Issue 1. [DOI] [PMC free article] [PubMed] [Google Scholar] Hughesdon 1957 - Hughesdon PE. The structure of endometrial cysts of the ovary. Journal of Obstetrics and Gynaecology 1957;44:481‐7. [DOI] [PubMed] [Google Scholar] Jones 2004 - Jones G, Jenkinson G, Kennedy S. The impact of endometriosis upon quality of life: a qualitative analysis. Journal of Psychosomatic Obstetrics and Gynaecology 2004;25(2):123‐33. [DOI] [PubMed] [Google Scholar] Loo 2005 - Loo TC, Lin MY, Chen SH, Chung MT, Tang HH, Lin LY, et al. Endometrioma undergoing laparoscopic ovarian cystectomy: its influence on the outcome of in vitro fertilization and embryo transfer (IVF‐ET). Journal of Assisted Reproduction and Genetics 2005;22:329‐33. [DOI] [PubMed] [Google Scholar] Matson 1986 - Matson PL, Yovich JL. The treatment of infertility associated with endometriosis by in vitro fertilization. Fertility and Sterility 1986;46:432‐4. [DOI] [PubMed] [Google Scholar] Muzii 2002 - Muzii L, Bianchi A, Croce C, Manci N, Panici PB. Laparoscopic excision of ovarian cysts: is the stripping technique a tissue sparing procedure?. Fertility and Sterility 2002;77:609‐14. [DOI] [PubMed] [Google Scholar] Nishida 2000 - Nishida M, Watanabe K, Sato N, Ichikawa Y. Malignant transformation of ovarian endometriosis. Gynecology and Obstetrics Investigations 2000;50:18‐25. [DOI] [PubMed] [Google Scholar] Oehninger 1988 - Oehninger S, Acosta AA, Kreiner D, et al. In vitro fertilization and embryo transfer (IVF/embryo transfer): an established and successful therapy for endometriosis. Journal of In Vitro Fertilization and Embryo Transfer 1988;5:249‐56. [DOI] [PubMed] [Google Scholar] Pellicer 1995 - Pellicer A, Oliveira N, Ruiz A, et al. Exploring the mechanism(s) of endometriosis‐related infertility: an analysis of embryo development and implantation in assisted reproduction. Human Reproduction 1995;10:91‐7. [DOI] [PubMed] [Google Scholar] Ragni 2005 - Ragni G, Somigliana E, Benedetti F, Paffoni A, Vegetti W, Restelli L, et al. Damage to ovarian reserve associated with laparoscopic excision of endometriomas: a quantitative rather than a qualitative injury. American Journal of Obstetrics and Gynecology 2005;193:1908‐14. [DOI] [PubMed] [Google Scholar] Sampson 1921 - Sampson JA. Perforating haemorraghic (chocolate) cysts of the ovary. Archives of Surgery 1921;3:245‐323. [Google Scholar] Sampson 1927 - Sampson JA. Peritoneal endometriosis due to the menstrual dissemination of endometrial tissue into the peritoneal cavity. American Journal of Obstetrics and Gynecology 1927;14:422‐69. [Google Scholar] Schünemann 2006b - Schünemann HJ, Jaeschke R, Cook DJ, Bria WF, El‐Solh AA, Ernst A, et al. An official ATS statement: grading the quality of evidence and strength of recommendations in ATS guidelines and recommendations. American Journal of Respiratory and Critical Care Medicine 2006;174:605‐14. [DOI] [PubMed] [Google Scholar] Tim 2001 - Child TJ, Tan SL. Endometriosis: Aetiology, pathogenesis and treatment. Drugs 2001;61:1735‐50. [DOI] [PubMed] [Google Scholar] Vercellini 2003a - Vercellini P, Chapron C, Giorgi O, Consonni D, Frontino G, Crosignani PG. Coagulation or excision of ovarian endometriomata?. American Journal of Obstetrics and Gynecology 2003;188:606‐10. [DOI] [PubMed] [Google Scholar] Winkel 2003 - Winkel CA. Evaluation and management of women with endometriosis. Obstetrics and Gynecology 2003;102:397‐408. [DOI] [PubMed] [Google Scholar] Yamamoto 2009 - Yamamoto S, Umeki M, Maeda T, Yamaguchi M, Hamano T. Poor pregnancy outcome following assisted reproductive technology among women operated on for bilateral endometriomas. Reproductive Medical Biology 2009;9:43‐9. [DOI] [PMC free article] [PubMed] [Google Scholar] Yanushpolsky 1998 - Yanushpolsky EH, Best CL, Jackson KV, Clarke RN, Barbieri RL, Hornstein MD. Effects of endometriomata on oocyte quality, embryo quality, and pregnancy rates in in vitro fertilization cycles: a prospective, case‐controlled study. Journal of Assisted Reproduction 1998;15:193‐7. [DOI] [PMC free article] [PubMed] [Google Scholar]