Abstract. Introduction

Transcription

1 RBMOnline - Vol 13. No Reproductive BioMedicine Online; on web 29 September 2006 Article GnRH-antagonists in ovarian stimulation for IVF in patients with poor response to gonadotrophins, polycystic ovary syndrome, and risk of ovarian hyperstimulation: a meta-analysis Georg Griesinger studied medicine at the University of Vienna from 1992 to 1998 and obtained his MD (1999) in trophoblast research under Professor M Knoefler at the Department of Obstetrics and Gynecology of the University Clinic of Vienna, Austria. Post-doctoral studies were conducted in clinical IVF with Dr P Kemeter in Vienna, and at University College, London, where he gained an MSc in Prenatal Genetics and Fetal Medicine in Currently he is a resident at the Department of Obstetrics and Gynecology at the University Clinic of Schleswig-Holstein, under Professor K Diedrich. His research interests include various aspects of assisted reproduction, particulary optimizing ovarian stimulation. Dr Georg Griesinger G Griesinger 1,3, K Diedrich 1, BC Tarlatzis 2, EM Kolibianakis 2 1 Department of Obstetrics and Gynecology, University Clinic of Schleswig-Holstein, Campus Luebeck, Luebeck, Germany; 2 Unit for Human Reproduction, 1st Department of Obstetrics and Gynaecology, Aristotle University of Thessaloniki, Greece 3 Correspondence: georg.griesinger@frauenklinik.uni-luebeck.de Abstract This article is a systematic review of the literature on utilization of gonadotrophin-releasing hormone antagonists (GnRH-ant) for ovarian stimulation for IVF in special patient groups. Summarized by meta-analysis are the data from randomized controlled trials (RCT) in which GnRH-agonist (GnRH-a) and GnRH-ant were compared (eight RCT for poor response, four RCT for PCOS). Also reviewed are the data from two RCT and 13 retrospective or observational trials in which patients at risk of ovarian hyperstimulation syndrome (OHSS) were triggered with GnRH-agonist instead of HCG. For poor responders, no differences in clinical outcomes were found, except a significantly higher number of cumulus oocyte complexes in GnRH-antagonist multiple dose protocol as compared to GnRH-agonist long protocol (P P = 0.05). For PCOS patients, no differences in outcomes were found, except a significantly shorter duration of stimulation, when GnRH-antagonist multiple dose protocol and GnRH-agonist long protocol are compared (P P < 0.01). However, sample sizes are still small and power to detect subtle differences is therefore limited. For OHSS risk patients triggered with GnRH-agonist, reports on the efficacy of this measure vary in the literature. GnRH-agonist triggering appears to be associated with a reduction in the incidence of mild and moderate OHSS. For prevention of severe OHSS, as yet, only very limited evidence is available. Keywords: cetrorelix, ganirelix, GnRH-antagonist, ovarian hyperstimulation syndrome, polycystic ovary syndrome, poor response 628 Introduction GnRH-antagonists (GnRH-ant; cetrorelix, ganirelix) were introduced to the European market in 1999 for use in ovarian stimulation for IVF (Diedrich et al., 1994; Olivennes et al., 1994). Initially, it was expected that these compounds would rapidly replace GnRHas (GnRHa) for pituitary suppression in ovarian stimulation, as the use of GnRH-ant is associated with several advantages (Felberbaum and Diedrich, 2003). However, uptake by clinicians was slow (Fauser and Devroey, 2005), most likely in response to the results of the initial comparative phase III studies, which indicated a lower probability of clinical pregnancy per treatment cycle in GnRH-ant protocols as compared with long luteal GnRHa protocols (summarized by Al-Inany and Aboulghar, 2002). Registry data (FIVNAT 2003, DIR 2004) indicate that the vast majority of IVF treatment cycles are still performed using the GnRHa long protocol regimen. GnRH-ant, in contrast, are apparently more often used as second-line agents, in patients who are older, or who were unsuccessful in previous treatment

2 attempts (Griesinger et al., 2005; Engel et al., 2006). Some theoretical considerations support the use of GnRH-ant in special patient groups. Since GnRH-ant avoids suppression of endogenous gonadotrophin secretion at the stage of follicular recruitment, it appears rational to use antagonists in patients with expected or proven decreased ovarian response to exogenous gonadotrophins (Craft et al., 1999; Tarlatzis et al., 2003). Furthermore, GnRH-ant administration can be tailored to the individual ovarian response (Ludwig et al., 2003), resulting in a pituitary suppressive effect only for the short time period in ovarian stimulation during which inhibition of premature LH surge is necessary. Regarding ovarian hyperstimulation syndrome (OHSS), although a significant reduction of the incidence of its severe forms has as yet not been shown in general population patients (excluding polycystic ovarian syndrome) (Al-Inany and Aboulghar, 2002), it is likely that GnRH-ant use will result not only in a reduced number of growing follicles, but also in a reduction of the incidence and severity of OHSS overall. More importantly, the pituitary remains responsive to a bolus dose of GnRHa under concomitant suppression with GnRH-ant (Felberbaum et al., 1995; Fauser et al., 2002). Thus, GnRHa can induce release of LH from the pituitary gland, similarly to a spontaneous mid-cycle LH surge, which provides an alternative to the administration of human chorionic gonadotrophin (HCG) in GnRH-ant-based stimulation protocols (Aboulghar and Al-Inany, 2005). Since triggering of final oocyte maturation with GnRHa has been suggested to prevent OHSS, its use has been proposed as a safer, and at the same time, efficacious way to perform IVF (Kol, 2003; Orvieto, 2005). Finally, polycystic ovary syndrome (PCOS) patients are distinctly different from general population patients, as they have cycle irregularities, an often-inverted FSH/LH ratio, a high incidence of hyperandrogenism, and often increased body mass index (BMI). An observational study on GnRHant utilization in PCOS patients has indicated a worse reproductive outcome in PCOS patients with high BMI (>29 kg/m 2 ), stimulated with higher doses of recombinant FSH, as compared with lower BMI patients stimulated with lower FSH doses (Kolibianakis et al., 2003). As yet, little comparative data from randomized studies are available on GnRH-ant in PCOS patients. Theoretical concerns regarding the use of antagonists refer to an absent suppression of elevated endogenous LH in early follicular phase by the antagonist protocol, and a probably increased incidence of premature LH surges, as PCOS patients are on average more obese and thus have a higher distribution volume. In contrast, it is expected that PCOS patients might possibly benefit from mild stimulation regimens using GnRH-ant protocols (Hohmann et al., 2003). Aim The aim of the present review is to summarize the clinical experience with GnRH-ant for the following indications: (i) poor ovarian response to exogenous gonadotrophins, (ii) patients with PCOS, and (iii) patients at risk of OHSS. Method of review The present review is based on two recent comprehensive systematic reviews conducted by the authors on clinical efficacy of GnRH-ant for ovarian stimulation in IVF (Kolibianakis et al., 2006), and GnRHa for triggering final oocyte maturation in GnRH-ant protocols (Griesinger et al., 2006). Briefly, a literature search was performed using the bibliographic databases MEDLINE, CENTRAL and EMBASE. Additionally, references of retrieved articles and meeting proceedings were hand-searched. The initial literature search covered the period until March 2005 (Griesinger et al., 2006) and December 2005 (Kolibianakis et al., 2006). A comprehensive explanation of the literature search has been given in the initial publications. For the purpose of the present review, this database of literature was updated in February Included are randomized, controlled trials (RCT) from which comparative data on clinical outcome after GnRHa or GnRHant usage for ovarian stimulation in poor responder patients and PCOS patients could be retrieved. Outcomes assessed are clinical (or ongoing) pregnancy rate per woman randomized; duration of stimulation (days), gonadotrophin consumption (ampoules); number of cumulus oocyte complexes (COC) retrieved; and OHSS incidence per woman randomized. Dichotomous results for each unit of analysis were expressed as an odds ratio (OR) with 95% confidence interval (CI). Continuous variables were expressed as standardized differences of the mean (Hedges adjusted effect size defined as the difference between two groups expressed in standard deviation units) with 95% CI. The results were combined for meta-analysis with the computer software program Comprehensive Meta Analysis, version (Biostat Inc., USA), using random or fixed effects model (depending on the presence or absence of heterogeneity as assessed by Cochrane's Q text; P < 0.05 considered significant) and Mantel Haenszel statistics, where appropriate. For GnRHa triggering of final oocyte maturation in GnRHant protocols in OHSS-risk patients, very little comparative data from RCT are available; therefore, evidence on GnRHa triggering mostly from uncontrolled observational or retrospective studies is reviewed in a systematic way, but a formal meta-analysis could not be performed. GnRH-ant in patients with poor response The comprehensive literature search identified eight studies (seven manuscripts, one abstract), encompassing 575 patients treated with GnRHa or antagonist for ovarian stimulation in RCT. Table 1 summarizes clinical features of these studies. In two studies, the long agonist protocol was used, whereas six studies used short agonist protocol. In two studies, a fixed antagonist protocol was used. All studies used multiple dose regimens of GnRH-ant. The majority of studies employed true randomization, with concealment of allocation. 629

3 Table 1. Main characteristics of randomized controlled trials (RCT) on patients with expected, or history of, poor response. Reference Sample Randomization Criteria for Agonist Antagonist Gonadosize poor response protocol protocol trophin type (ITT) Full publication Akman et al., True, allocation Previous cycle: Short, flare-up Flexible, HMG + concealed bfsh >15 IU/l or E2 (leuprolide), multiple dose ufsh (dhcg) <500 pg/ml pretreatment with (cetrorelix) or COC <4 OCP a Martínez et al., True, allocation Previous poor Short, flare-up Fixed, multiple rfsh + concealed response (triptorelin), dose (cetrorelix), HMG OCP pretreatment OCP pretreatment Cheung et al., True, allocation Repeated bfsh > Long, luteal Fixed, multiple rfsh concealed 10 IU/l or previous (buserelin), OCP dose (cetrorelix), cycle with <3 COC pretreatment OCP pretreatment Marci et al., True, allocation Previous cycle: E2 Long, luteal Flexible, rfsh concealed (dhcg) <600 pg/ml (leuprolide) multiple dose and <3 COC (ganirelix) Malmusi et al., True, allocation Previous cycle: <5 Short, flare-up Flexible, rfsh concealment COC or no ovarian (triptorelin) multiple dose unclear response when 300 (ganirelix) IU FSH for 15 days Schmidt et al., True, allocation Previous cycle: E2 Short, flare-up Flexible, rfsh + concealed (dhcg) 850 pg/ml (leuprolide), multiple dose HMG and/or 4 COC and OCP pretreatment a (ganirelix) bfsh <13 miu/ml De Placido et al., True, allocation 37 yrs or bfsh 9 Short, flare-up Flexible, rfsh + concealment IU/l, regular cycle (triptorelin) multiple dose rlh unclear (cetrorelix) Abstract Mollo et al., Randomization bfsh >9 IU/l and/or Short, flare-up Flexible, rfsh + method not >37 yrs (decapeptyl) multiple dose uhcg described (ganirelix) ITT = intention-to-treat; yrs = years; b = basal (referring to measurement of endogenous FSH in early follicular phase); E2 = oestradiol; dhcg = day of human chorionic gonadotrophin administration; COC = cumulus oocyte complex; OCP = oral contraceptive pill; HMG = human menopausal gonadotrophin; r = recombinant; u = urinary. a Co-intervention in agonist group: OCP pretreatment. 630 Definition of poor response was heterogenous; the majority of studies defined poor response as inappropriate ovarian response to ovarian stimulation in a previous cycle. Two studies (Mollo et al., 2005; De Placido et al., 2006) used prospectively defined criteria, e.g. age and basal endogenous FSH concentrations. Figure 1 summarizes clinical outcome in poor responder patients. Duration of stimulation (Figure 1a) and consumption of gonadotrophins (Figure 1b) were lower with GnRH antagonists; however, this did not reach statistical significance, there was significant heterogeneity in outcome between studies in all comparison groups (P < 0.01), and most of the effect accounts to the comparison of long agonist versus antagonist protocol. Subgroup analysis on short agonist versus antagonist multiple dose protocol studies suggests similar results as regards stimulation duration, gonadotrophin consumption and number of oocytes retrieved; however, in this subgroup analysis, there was also significant heterogeneity between studies (see Figure 1 for P values). Figure 1c indicates that the number of oocytes is higher in antagonist stimulation protocols, as compared with agonist long protocols (OR 0.41, 95% CI: , P = 0.05). The likelihood of clinical pregnancy is higher after antagonistbased stimulation; however, this did not reach statistical significance (OR 1.28, 95% CI: ; P = 0.25; fixed effects model; no heterogeneity) (Figure 1d). No clinically relevant OHSS was reported in the poor responder studies, as expected.

4 Figure 1. Clinical outcome from comparative studies on gonadotrophin-releasing hormone antagonist (GnRH-ant) versus gonadotrophin-releasing hormone agonist (GnRHa) protocol in poor responder patients stratified for type of agonist protocol. Depicted are the standardized differences of the mean or OR, with 95% confidence intervals. Favor agonist indicates that an outcome was comparatively higher in quantitative variables or more often observed in categorical variables in the agonist group than in the antagonist group and vice versa). (a) Duration of stimulation (random effects model; significant heterogeneity, P < 0.01). (b) Gonadotrophin consumption (random effects model; significant heterogeneity, P < 0.01). (c) Number of cumulus oocyte complexes (fixed effects model; no significant heterogeneity, P = 0.14). (d) Clinical pregnancy rate per randomized patient (fixed effects model; no significant heterogeneity, P = 0.79). 631

5 Are GnRH-ant protocols superior to GnRHa protocols in patients with poor response? The number of analysed studies is currently small. It appears, however, that no major differences exist between the short agonist and the antagonist protocol. Regarding the comparison between the GnRH-ant and the long agonist protocol, only two studies are available; however, it appears that the long agonist protocol in poor responders is associated with significantly fewer COCs retrieved compared with the GnRH-ant protocol (OR 0.41, 95% CI: , P = 0.05). Additional studies are necessary to allow more solid conclusions to be drawn. GnRH-ant in patients with PCOS Table 2 summarizes characteristics of RCT on antagonist versus agonist protocols in patients with PCOS. Four studies were identified (three manuscripts, one abstract publication), encompassing 305 randomized patients. All studies employed long agonist protocols, with pretreatment with the oral contraceptive pill, except one study (Hwang et al., 2004). All studies employed multiple dose antagonist regimens for comparison, in two of which antagonist administration started in early follicular phase (Hwang et al., 2004; Kim et al., 2004). True randomization with concealment of allocation was reported in two manuscripts (Hwang et al., 2004; Bahçeci et al., 2005). Figure 2a indicates that the duration of stimulation is significantly reduced in favour of GnRH-ant when GnRH-ant multiple dose and GnRHa long protocol are compared (OR 0.86, 95% CI 1.14 to 0.59, P < 0.01), although this was not associated with a significant reduction in gonadotrophin consumption (Figure 2b). No significant difference was found regarding the number of cumulus oocyte complexes retrieved (Figure 2c) and the likelihood of clinical pregnancy (Figure 2d). No severe OHSS occurred in the studies reporting OHSS (Hwang et al., 2004; Bahçeci et al., 2005). No significant difference in the incidence of OHSS (grades I II) was present (Figure 2e). Table 2. Main characteristics of randomized controlled trials (RCT) on patients with polycystic ovarian syndrome (PCOS). Reference Sample Randomization Main inclusion Agonist Antagonist Gonadosize criteria protocol protocol trophin (ITT) type Full publication Hwang et al., 56 True, allocation Oligo/amenorrhoea and Long, follicular OCP pretreatment for HMG 2004 concealed sonomorphological PCO (buserelin) 3 cycles, cetrorelix and LH/FSH ratio >2 or ( mg) T >0.8 ng/ml and <39 starting day before yrs and bfsh <12 IU/l 1 HMG a Bahçeci et al., 148 True, allocation Primary infertility, OCP pretreatment OCP pretreatment for ufsh 2005 not concealed oligomenorrhoea, clinical for 21 days, start 21 days, antagonist hyperandrogenism, with leuprolide flexible, multiple dose reversed FSH/LH ratio from day 14 of (cetrorelix) and sonomorphological preceding cycle PCO Ashrafi et al., 60 Randomization Oligomenorrhoea and OCP pretreatment, OCP pretreatment, HMG 2004 method not hyperandrogenism and long, start on day flexible, multiple described LH/FSH >2.5 and 21 (buserelin) dose (cetrotide ultrasonographic features or ganirelix) b of PCOS Abstract Kim et al., 41 Randomization PCOS and 36 yrs OCP pretreatment, OCP pretreatment, rfsh 2004 method not long luteal mg on days 1 described (triptorelin) and 2 followed by 0.25 mg flexible start, multiple dose (cetorelix) 632 ITT = intention-to-treat; yrs = years; b = basal (referring to measurement of endogenous FSH in early follicular phase); OCP = oral contraceptive pill; HMG = human menopausal gonadotrophin; r = recombinant; u = urinary; PCO = polycystic ovaries; T = testosterone. a Co-intervention in antagonist group: OCP pretreatment; b Co-intervention in antagonist group: patients at risk of OHSS were triggered with GnRHa instead of HCG.

6 Figure 2. Clinical outcome from comparative studies on gonadotrophin-releasing hormone antagonist (GnRH-ant) versus gonadotrophin-releasing hormone agonist (GnRHa) protocol in PCOS patients. Depicted are the standardized differences of the mean or OR, with 95% confidence intervals. (a) Duration of stimulation (fixed effects model; no significant heterogeneity, P = 0.19). (b) Gonadotrophin consumption (random effects model; no significant heterogeneity, P = 0.07). (c) Number of cumulus oocyte complexes (random effects model; significant heterogeneity, P = 0.02). (d) Clinical pregnancy rate per randomized patient (fixed effects model; no significant heterogeneity, P = 0.92). (e) Ovarian hyperstimulation syndrome incidence (grades I II) II) (fixed effects model; no significant heterogeneity, P = 0.64). 633

7 634 Are GnRH-ant protocols superior to GnRHa protocols in patients with PCOS? The limited evidence from the literature suggests that in PCOS patients duration of stimulation is shorter using GnRH-ant and that GnRH-ant are associated with a non-significant lower consumption of gonadotrophins, as compared with treatment with the GnRHa long protocol. However, non-outcome benefits associated with the antagonist protocol, such as the option to trigger final oocyte maturation with GnRHa instead of HCG in PCOS patients at OHSS risk, might be of greater importance in clinical decision-making on the protocol of first choice. GnRHa triggering in GnRH-ant protocols in OHSS-risk patients In March 2005, a systematic review identified three randomized controlled trials comparing GnRHa with HCG for triggering final oocyte maturation in IVF patients, published in peer reviewed journals (Griesinger r et al., 2006). In these trials, only normal responder patients were included, as a randomized trial on OHSS-risk patients is difficult to pursue due to ethical considerations. In this population, it was shown that the likelihood of clinical pregnancy was significantly reduced after GnRHa triggering with 0.2 mg triptorelin or 0.5 mg buserelin combined with vaginal progesterone and oral oestradiol for luteal phase support. This lower likelihood of pregnancy achievement appears to be associated only with fresh embryo transfer cycles, i.e. to a defective luteal phase, possibly in combination with insufficient luteal phase support following GnRHa triggering (Yding Andersen and Humaidan, 2005; Griesinger r et al., unpublished data). Although it is likely that the mechanisms accounting for the lower pregnancy rate after GnRHa triggering in a general population will still be present in patients at risk of OHSS, this might not be the case when different types of luteal phase support are employed (Engmann et al., 2005a,b). Furthermore, as patients at risk for OHSS are usually those with good prognosis (young age; high number of oocytes, and thus more embryos available for selection), the reduced chance of pregnancy achievement following agonist triggering might be less pronounced in OHSS-risk patients. What is the efficacy of GnRHa triggering in OHSS-risk patients? Thirteen retrospective or observational studies (two manuscripts, 11 abstracts), and two randomized controlled trials (one manuscript, one abstract), reporting pregnancy rates in OHSSrisk patients that were triggered with agonist were identified in the literature (Table 3). The sample sizes, definition of OHSS risk, and the luteal phase support (LPS) varied in these studies. For triggering final oocyte maturation, most often a single dose of 0.2 mg triptorelin or 1 2 mg leuprolide acetate was used. The majority of authors used intramuscular (i.m.) progesterone with or without oral or trans-dermal oestradiol for LPS. Clinical pregnancy rates range from 0 (Itskovitz-Eldor et al., 2000) to 75% (Engmann et al., 2005a). Table 3 suggests that in studies in which i.m. progesterone was used, higher pregnancy rates are achieved (Chun, 2005; Engmann et al., 2005a,b; Erden et al. 2005), as compared with studies with vaginal progesterone administration (Bracero et al., 2001; Kol and Muchtar, 2005; Körösi et al., 2006). However, baseline success rates, embryo transfer policies, and selection of patients were not consistently reported and might differ between the studies; therefore, the notion that i.m. progesterone performs better than vaginal progesterone for LPS should be considered tentative. The preliminary results from one of the randomized control trials (Engmann et al., 2005b) suggest that GnRHa and HCG are equally effective when i.m. progesterone combined with transdermal oestradiol is used for LPS. However, in this RCT the sample size is small, the study design is biased due to systematic differences in stimulation protocols and LPS in the groups compared, while the external validity of this finding is unclear, as the overall pregnancy rate in both study arms is exceptionally high. In contrast, the RCT of Babayof et al. (2006), in which i.m. progesterone and oral oestradiol were also used for LPS, finds live birth rate per embryo transfer not in favour of agonist triggering (1/15 [6.7%] in agonist versus 2/11 [18.2%] in HCG patients). In this trial, early pregnancy loss occurred in 80% of pregnancies after agonist triggering as compared with 50% after HCG. The pregnancy rates reported in some studies in Table 3 would appear acceptable for OHSS-risk patients, when the chance of success has to be balanced with the risk of OHSS following HCG. However, these positive preliminary results are currently available exclusively from abstract publications and need confirmation in appropriately sized, prospective, observational studies, in which selection of patients, luteal phase support and assessment of both pregnancy and OHSS are clearly described. As yet, the only three studies on agonist triggering in OHSSrisk patients published in peer-review journals (Itskovitz-Eldor et al., 2000; Kol and Muchtar, 2005; Babayof et al., 2006) are not in favour of this measure, as far as likelihood of pregnancy achievement is concerned (Table 3). Should embryo transfer in OHSSrisk patients be performed independently from the luteal phase following ovarian stimulation and GnRHa? The cause of the reported lower pregnancy likelihood after GnRHa triggering of final oocyte maturation and fresh embryo transfer observed in the RCT in general patient population (Griesinger r et al., 2006) is unknown, but is likely to be associated with a defective luteal phase following agonist triggering (Fauser et al., 2002; Nevo et al., 2003; Humaidan et al., 2005). Potentially, this defective luteal phase will also impair likelihood of pregnancy in OHSS-risk patients, despite luteal phase support with intramuscular instead of vaginal progesterone. Possibly, the pregnancy rates achieved after fresh transfer as depicted in Table 3 do not reflect the true potential of pregnancy achievement in a highly selected, good

8 Table 3. Main characteristics of studies on GnRHa-triggering of final oocyte maturation in GnRH-ant protocols in patients at risk of developing ovarian hyperstimulation syndrome. In comparative studies, only results from agonist-triggered patients are depicted. Reference Trial type n OHSS-risk GnRH-ant GnRHa LPS Pregnancy OHSS criteria protocol (daily) Kol and Itskovitz- Observational, 8 >20 follicles Fixed, multiple 0.2 mg 50 mg i.m. CPR/ET: No signs or Eldor, 2000 fp uncontrolled >11 mm and/or dose (ganirelix) TR P + 2 mg 0% symptoms E2 >3000 pg/ml oral E2 of OHSS Kol and Muchtar, Observational, 6 History of Flexible, 0.2 mg 600 mg OPR/ET: No OHSS 2005 fp uncontrolled OHSS multiple dose TR vag P + 4 1/6 leading to (cetrorelix) mg oral E2 (16.76%) hospitalization Babayof et al., RCT 15 PCOS Flexible, 0.2 mg 50 mg i.m. OPR/ET: None 2006 fp multiple dose TR P + 4 mg 1/15 (cetrorelix) oral E2 (6.67%) Bracero et al., Retrospective, 8 20 follicles Flexible, 2 1 mg 200 mg CPR: 2/8 No signs 2001 ab comparative 15 mm or E2 multiple dose LA vag P (25.0%) or symptoms 3000 pg/ml (ganirelix) of OHSS Meltzer et al., Observational, 35 History of Flexible, 0.25 mg Not stated CPR: 11/35 1/35 (2.89%) 2002 ab uncontrolled OHSS or PCOS multiple and TR (31.4%) (moderate single dose OHSS requiring (cetrorelix) hospitalization) Bankowski et al., Retrospective, 97 E Not stated 1 mg LA Not stated PR: 11/97 No severe case 2004 ab comparative pg/ml (11.3%) of OHSS Carone et al., Observational, 10 PCOS ( 20 Flexible, 0.2 mg Not stated CPR: 3/10 No signs and 2005 ab uncontrolled follicles 15 multiple dose TR (33.3%) symptoms of mm and/or E2 (ganirelix) OHSS 3500 pg/ml) Chun, 2005 ab Retrospective, 26 >20 follicles > Not stated mg i.m CPR: 7/26 1/26 (3.8%) cohort 14 mm or E2 > mg LA P mg (26.9%) (severe, late pg/ml oral E2 onset in pregnant woman) Erden et al., Retrospective, 97 a PCOS Flexible, 0.2 mg 100 mg i.m. CPR: 32/79 Not reported 2005 ab cohort multiple dose TR P (41%) (cetrorelix) Shapiro, 2005 ab Retrospective, follicles Not stated 4 or 8 mg Not stated CPR: 9/29 No OHSS cohort LA (31.0%) requiring aspiration of ascites Engmann et al., Retrospective, 20 PCOS or Not stated L A i.m. P + E2 CPR: 15/20 None 2005a ab comparative previous high transdermal (75.0%) responder Engmann et al., RCT 12 <40 yrs with Flexible, 1 mg LA 50 mg i.m. CPR: 8/12 None 2005b ab PCOS or history multiple dose P mg (66.7%) of OHSS (ganirelix) E2/2n day Bukulmez et al., Retrospective, 10 Oocyte donors Not stated LA Not stated NA None 2005 ab comparative at OHSS risk Bar-Hava et al., Observational 67 Previous OHSS Single and 0.25 mg Not stated CPR/ET: 1/67 (1.54%) 2005 ab II-III or peak E2 multiple dose TR 30% Hospital 2500 pg/ml (cetrorelix) admission and >7 follicles with mild >15 mm OHSS Körösi et al., Retrospective 25 E2 >2500 pg/ml Not stated 0.1 or 600 mg vag OPR/ET: Not reported 2006 ab comparative 0.2 mg P μg 7/25 TR transdermal (28.0%) E2 fp denotes full publiction; ab denotes abstract; a Number of cycles. n = number of patients triggered with agonist; OHSS = ovarian hyperstimulation syndrome; PCOS = polycystic ovarian syndrome; LPS = luteal phase support; i.m. = intramuscular; vag = vaginal; yrs = years; P = progesterone; ET = embryo transfer; TR = triptorelin; LA = leuprolide acetate; PR = pregnancy rate; CPR = clinical pregnancy rate; OPR = ongoing pregnancy rate; E2 = oestradiol; RCT = randomized controlled trial. 635

9 636 prognosis cohort. Indicative of this notion are the following observations. Erden et al. (2005) found a lower clinical pregnancy rate in fresh transfer cycles as compared with frozen thawed embryo replacement (FRET) cycles in the same cohort of OHSS-risk women (62 versus 41% in frozen and fresh cycles respectively) after GnRHa triggering. This is especially relevant as the women undergoing FRET cycles were negatively selected with respect to their chance of pregnancy, as they did not become pregnant after fresh transfer. Here, LPS was provided with 100 mg progesterone i.m. daily in both fresh and FRET cycles. Similarly, Shapiro et al. (2005) report a higher pregnancy rate in FRET cycles after GnRHa triggering as compared with fresh cycles in patients with at least 20 mature follicles (86 versus 30%). Here, no information on LPS was provided. The authors own experience in normal responder population indicates that the likelihood of pregnancy achievement is higher in FRET cycles as compared with fresh transfer cycles. Interestingly, one of the rare cases of severe OHSS following agonist triggering reported in the literature so far occurred as a late-onset form of the syndrome in a patient who was pregnant after fresh transfer (Chun, 2005). Taking these observations together suggests that transfer of embryos in FRET cycles (independently of the effects of ovarian stimulation and GnRHa) in OHSS-risk patients deserves further evaluation, and might be more successful as compared with fresh cycle transfer, even when accounting for the loss of embryos due to cryopreservation. Does GnRHa triggering in antagonist protocols prevent OHSS in risk patients? The studies summarized in Table 3 indicate that the incidence of clinically relevant OHSS (requiring hospitalization) after agonist triggering appears to be much less than would be expected in a high-risk population triggered with HCG (although from the abstract publications, definition of OHSS and method of assessment are mostly unclear). However, this is not convincingly substantiated in the studies having control populations triggered with HCG, also in which no severe OHSS is explicitly reported: Bracero et al. (2001) report mild OHSS in 2/11 HCG patients; Babayof et al. (2005) report OHSS in 6/11 HCG patients (or moderate-to-severe in 4/13; Babayof et al., 2006); Bukulmez et al. (2005) report moderate OHSS in 2/13 HCG patients; Engmann et al. (2005a) report moderate OHSS in 1/23 HCG patients; and Engmann et al. (2005b) report mild OHSS in 4/13 hcg patients. In contrast, two OHSS requiring hospitalization or classified as severe (lateonset) are reported after agonist triggering (Meltzer et al., 2002; Chun et al., 2005). Considering that the incidence of severe OHSS as defined by Golan et al. (1989) is a relatively rare event even in high-risk patients, a higher number of patients after agonist triggering need to be evaluated regarding OHSS occurrence to reach the confident conclusion that GnRHa triggering in antagonist protocols eliminates the occurrence not only of mild and moderate, but also severe forms of the syndrome, as previously suggested (Itskovitz-Eldor et al., 2000). Summary GnRH-ant ovarian stimulation protocols offer a number of potential advantages (Tarlatzis et al., 2006) compared with the still accepted gold standard treatment, the GnRHa long protocol, including the option of triggering final oocyte maturation with GnRHa instead of HCG. Outcome differences between stimulation protocols with either GnRHa or antagonists are possibly more prominent in special patient subgroups, as compared with a more general population. However, there are a variety of ways in which both antagonists and agonists can be employed for pituitary suppression in ovarian stimulation, which inevitably introduces heterogeneity into the available comparative trials. This is illustrated by the findings of the current review. For patients with poor response to gonadotrophins, the outcome differences between GnRHa short protocol and GnRH-ant multiple dose protocol appear to be small, whereas more prominent differences appear to exist when only the GnRHa long protocol and the GnRH-ant multiple dose protocol are compared. Nevertheless, significant heterogeneity was present in most subgroup analysis in the present work. Bearing in mind the small sample size and limited power, statistically significant differences between agonist and antagonists were only found for two outcomes, both in favour of antagonists: a statistically significantly higher number of cumulus oocyte complexes was retrieved in poor response patients in antagonist versus agonist long ovarian stimulation. In PCOS patients, the duration of stimulation was shorter in GnRH-ant patients. However, a larger number of studies are necessary to further substantiate these findings, and to also detect more subtle differences with confidence between antagonists and agonists in poor response and PCOS patients. For GnRHa triggering of final oocyte maturation, a number of issues warrant further investigation: the reason for the observed lower likelihood of pregnancy after GnRHa triggering and luteal phase support with vaginal progesterone and oral oestradiol (Humaidan et al., 2005; Kolibianakis et al., 2005) is unclear; an optimal protocol (including optimal luteal phase support) has not yet been established, and the potential of GnRHa to prevent severe OHSS in high-risk patients should be explored further by larger-sized observational studies. Therefore, GnRHa triggering in OHSS-risk patients should still be considered an experimental procedure. References Aboulghar M, Al-Inany H 2005 Use of GnRHa to induce ovulation in GnRH antagonist cycles. Reproductive BioMedicine Online 11, 520. Akman MA, Erden, HF, Tosun, SB et al Comparison of agonistic flare-up-protocol and antagonistic multiple dose protocol in ovarian stimulation of poor responders: results of a prospective randomized trial. Human Reproduction 16, Al-Inany H, Aboulghar M 2002 GnRH antagonist in assisted reproduction: a Cochrane review. Human Reproduction 17,

10 Ashrafi M, Mohammadzadeh A, Ezabadi Z, Baghestani AR 2005 A comparative study of GnRH-ant and GnRH-ag protocols in IVF ICSI in PCOD patients. Iranian Journal of Reproductive Medicine 3, Babayof R, Margalioth EJ, Huleihel M et al Serum inhibin A, VEGF and TNFα levels after triggering oocyte maturation with GnRHa compared with HCG in women with polycystic ovaries undergoing IVF treatment: a prospective randomized trial. Human Reproduction Jan 26 [Epub ahead of print]. Babayof R, Huleihel M, Margalioth EJ et al Serum inhibin A and VEGF levels after triggering oocyte maturation with GnRHa (GnRH-a) versus hcg among women with polycystic ovary syndrome (PCOS) undergoing IVF treatment. Human Reproduction 20 (Suppl.1), 298. Bahçeci M, Ulug U, Ben-Shlomo I et al Use of a GnRH antagonist in controlled ovarian hyperstimulation for assisted conception in women with polycystic ovary disease: a randomized, prospective, pilot study. Journal of Reproductive Medicine 50, Bankowski B, Bracero N, King J et al Triggering ovulation with leuprolide acetate is associated with lower pregnancy rates. Abstracts of the 19th annual meeting of the ESHRE, Berlin, Germany, P-295, i103. Bar-Hava I, Rabinson J, Ben-Mordechay R et al Preventing OHSS by inducing oocyte maturation with GnRHa, instead of hcg, in high responders treated with GnRH antagonist protocols for IVF: a comparison between single dose (3mg) and daily dose (0.25mg) regimens. Abstracts of the Annual meeting of the ESHRE, Madrid, Spain, 2003, xviii 132 (P-389). Bracero NJ, Jurema MW, Posada MN et al Triggering ovulation with leuprolide acetate (LA) instead of human chorionic gonadotrophin (hcg) after the use of ganirelix for invitrofertilization-embryo transfer (IVF ET) does not compromise cycle outcome and may prevent ovarian hyperstimulation syndrome. Fertility and Sterility (suppl.), S93 (O-254). Bukulmez O, Rehman KS, Langley M et al Triggering ovulation by GnRHa leuprolide acetate does not adversely affect the number and quality of the oocytes as compared with recombinant hcg in oocyte donation cycles. Fertility and Sterility 84 (Suppl. 1), S314 (P-455). Carone D, Vizziello GM, Schonauer LM, D Amato G 2005 Safety and efficacy of GnRHa to trigger ovulation in controlled ovarian hyperstimulation for ART with recombinant FSH and GnRH antagonist in high responders (PCOD) patients. Abstract book of the 8th International Symposium on GnRH-analogues in Cancer and Human Reproduction, Salzburg, Austria, A24. Cheung LP, Lam, PM, Lok, IH et al GnRH antagonist versus long GnRHa protocol in poor responders undergoing IVF: a randomized controlled trial. Human Reproduction 20, Chun E 2005 Severe OHSS can be prevented in GnRH antagonist protocol using GnRHa to trigger ovulation. Fertility and Sterility 84 (Suppl. 1), S301 (P-149). Craft I, Gorgy A, Hill J et al Will GnRH-antprovide new hope for patients considered difficult responders to GnRHa protocols? Human Reproduction 14, De Placido G, Mollo A, Clarizia R et al Gonadotrophinreleasing hormone (GnRH) antagonist plus recombinant luteinizing hormone versus a standard GnRHa short protocol in patients at risk for poor ovarian response. Fertility and Sterility 85, Diedrich K, Diedrich C, Santos E et al Suppression of the endogenous LH-surge by the LH-RH antagonist cetrorelix during ovarian stimulation. Human Reproduction 9, Engel JB, Griesinger G, Schultze-Mosgau A et al GnRHas and antagonists in assisted reproduction: pregnancy rate. Reproductive BioMedicine Online 13, Engmann L, Hartnett J, Siano L et al. 2005a The Use of GnRHa to trigger final stages of oocyte maturation in patients with polycystic ovarian syndrome (PCOS) and high responders during IVF treatment. Fertility and Sterility 83 (Suppl. 2), S26 (P-40). Engmann L, Diluigi A, Schmidt D et al. 2005b Prevention of ovarian hyperstimulation syndrome (OHSS) with the use of gonadotrophin releasing hormone (GnRH) agonist to trigger final oocyte maturation after co-treatment with GnRH antagonist in patients with polycystic ovarian syndrome (PCOS) or previous high response undergoing IVF treatment a prospective randomized clinical trial. Fertility and Sterility 84 (Suppl. 1), S96 (O-233). Erden HF, Akman MA, Çiray N, Bahçeci M 2005 The results of fresh and frozen thaw embryo transfer cycles of PCOS patients in which GnRH-antwere used together with GnRHa for LH surge. Fertility and Sterility 84 (Suppl. 1), S164 (P-36). Fauser BC, Devroey P 2005 Why is the clinical acceptance of gonadotrophin-releasing hormone antagonist cotreatment during ovarian hyperstimulation for in vitro fertilization so slow? Fertility and Sterility 83, Fauser BC, de Jong D, Olivennes F et al Endocrine profiles after triggering of final oocyte maturation with GnRHa after cotreatment with the GnRH antagonist ganirelix during ovarian hyperstimulation for in vitro fertilization. Journal Clinical Endocrinology and Metabolism 87, Felberbaum RE, Diedrich K 2003 Gonadotrophin-releasing hormone antagonists: will they replace the agonists? Reproductive BioMedicine Online 6, Felberbaum RE, Reissmann T, Kupker W et al Preserved pituitary response under ovarian stimulation with HMG and GnRH-ant (Cetrorelix) in women with tubal infertility. European Journal of Obstetrics, Gynecology and Reproductive Biology 61, Golan A, Ron-El R, Herman A 1989 Ovarian hyperstimulation syndrome: an update review. Obstetrical and Gynecological Survey 44, Griesinger G, Diedrich K, Devroey P, Kolibianakis EM 2006 GnRHa for triggering final oocyte maturation in the GnRH antagonist ovarian hyperstimulation protocol: systematic review and metaanalysis. Human Reproduction Update 12, 159. Griesinger G, Felberbaum R, Diedrich K 2005 GnRH-antin ovarian stimulation: a treatment regimen of clinicians second choice? Data from the German national IVF registry. Human Reproduction 20, Hohmann FP, Macklon NS, Fauser BC 2003 A randomized comparison of two ovarian stimulation protocols with gonadotrophin-releasing hormone (GnRH) antagonist cotreatment for in vitro fertilization commencing recombinant folliclestimulating hormone on cycle day 2 or 5 with the standard long GnRHa protocol. Journal of Clinical Endocrinology and Metabolism 88, Humaidan P, Ejdrup Bredkjaer H, Bungum L et al GnRHa (buserelin) or hcg for ovulation induction in GnRH antagonist IVF/ICSI cycles: a prospective randomized study. Human Reproduction 20, Hwang JL, Seow KM, Lin YH et al Ovarian stimulation by concomitant administration of cetrorelix acetate and HMG following Diane-35 pre-treatment for patients with polycystic ovary syndrome: a prospective randomized study. Human Reproduction 19, Itskovitz J, Boldes R, Levron J et al Induction of preovulatory luteinizing hormone surge and prevention of ovarian hyperstimulation syndrome by gonadotrophin-releasing hormone agonist. Fertility and Sterility 56, Itskovitz-Eldor J, Kol S, Mannaerts B 2000 Use of a single bolus of GnRH-antagonist triptorelin to trigger ovulation after GnRHantagonist ganirelix treatment in women undergoing ovarian stimulation for assisted reproduction, with special reference to the prevention of ovarian hyperstimulation syndrome: a preliminary report. Human Reproduction 15, Kim CH, Lee YJ, Hong SH et al Efficacy of a GnRH antagonist during early and late controlled ovarian hyperstimulation period in women with polycystic ovary syndrome undergoing IVF ET. Human Reproduction 19, Kol S 2003 Prediction of ovarian hyperstimulation syndrome: why predict if we can prevent! Human Reproduction 18,

11 638 Kol S, Itskovitz-Eldor J 2000 Severe OHSS: yes, there is a strategy to prevent it! Human Reproduction 15, Kol S, Muchtar M 2005 Recombinant gonadotrophin-based, ovarian hyperstimulation syndrome-free stimulation of the high responder: suggested protocol for further research. Reproductive BioMedicine Online 10, Kolibianakis EM, Collins J, Tarlatzis B et al Among patients treated for IVF with gonadotrophins and GnRH analogues, is the probability of live birth dependent on the type of analogue used? A systematic review and meta-analysis. Human Reproduction Update August 18 [Epub ahead of print, doi: /humupd/dml038]. Kolibianakis E, Schultze-Mosgau A, Schroer A et al A lower ongoing pregnancy rate can be expected when GnRHa is used for triggering final oocyte maturation instead of hcg, in patients undergoing IVF with GnRH antagonists. Human Reproduction 20, Kolibianakis E, Zikopoulos K, Albano C et al Reproductive outcome of polycystic ovarian syndrome patients treated with GnRH-antand recombinant FSH for IVF/ICSI. Reproductive BioMedicine Online 7, Körösi T, Boga P, Barta C et al Use of GnRHa instead of hcg to trigger ovulation in GnRH antagonist IVF/ICSI cycles. Gynecological Endocrinology 22 (Suppl.1), 181 (P20). Ludwig M, Katalinic A, Banz C et al Tailoring the GnRH antagonist cetrorelix acetate to individual patients needs in ovarian stimulation for IVF: results of a prospective, randomized study. Human Reproduction 17, Malmusi S, La Marca A, Giulini S et al Comparison of a gonadotrophin-releasing hormone (GnRH) antagonist and GnRHa flare-up regimen in poor responders undergoing ovarian stimulation. Fertility and Sterility 84, Marci R, Caserta D, Dolo V et al GnRH antagonist in IVF poorresponder patients: results of a randomized trial. Reproductive BioMedicine Online 11, Martínez F, Coroleu B, Marqués L et al Comparación del Protocolo Corto versus Antagonistas con o sin Citrato de Clomifeno para estimulación en FIV de pacientes con baja respuesta Revista Iberoamericana de Fertilidad Meltzer S, Girsh E, Shults A, Katz N, Zohav E, Tur-Kaspa I 2002 Prevention of ovarian hyperstimulation syndrome in high responders undergoing IVF treatment with GnRH antagonist combined with single dose of GnRHa, instead of HCG, for the induction of oocyte maturation. Abstracts of the 18th Annual Meeting of the ESHRE, Vienna, Austria, O-258, p. 89 Mollo A, Clarizia R, Strina I et al GnRH antagonist plus hcg versus GnRHa short protocol in patients at risk for poor ovarian response. Human Reproduction 20 (Suppl.1), i126 (P-348). Nevo O, Eldar-Geva T, Kol S, Itskovitz-Eldor J 2003 Lower levels of inhibin A and pro-αc during the luteal phase after triggering oocyte maturation with a gonadotrophin-releasing hormone agonist versus human chorionic gonadotrophin. Fertility and Sterility 79, Olivennes F, Fanchin R, Bouchard P et al Scheduled administration of a gonadotrophin-releasing hormone antagonist (Cetrorelix) on day 8 of in-vitro fertilization cycles: a pilot study. Human Reproduction 10, Orvieto R 2005 Can we eliminate severe ovarian hyperstimulation syndrome? Human Reproduction 20, Schmidt DW, Bremner T, Orris JJ et al A randomized prospective study of microdose leuprolide versus ganirelix in in vitro fertilization cycles for poor responders. Fertility and Sterility 83, Shapiro BS, Daneshmand ST, Garner FC et al Eliminating severe ovarian hyperstimulation syndrome by using GnRHa instead of hcg. Fertility and Sterility 78 (Suppl. 1), S17 (P-12) Tarlatzis B, Fauser BC, Kolibianakis EM et al GnRH-antin ovarian stimulation for IVF. Human Reproduction Update, in press. Tarlatzis BC, Zepiridis L, Grimbizis G, Bontis J 2003 Clinical management of low ovarian response to stimulation for IVF: a systematic review. Human Reproduction Update 9, Yding Andersen C, Humaidan P 2005 Characteristics of follicular fluid from women receiving either buserelin or hcg to induce ovulation. Human Reproduction Supplement 1 (20), 65. Received 24 May 2006; refereed 8 June 2006; accepted 1 August 2006.