Association of estradiol levels on the day of hcg administration and pregnancy achievement in IVF: a systematic review

Transcription

1 Human Reproduction Vol.19, No.11 pp , 2004 Advance Access publication October 7, 2004 doi: /humrep/deh473 Association of estradiol levels on the day of hcg administration and pregnancy achievement in IVF: a systematic review Ioannis P.Kosmas, Efstratios M.Kolibianakis 1 and Paul Devroey Center for Reproductive Medicine, Dutch-speaking Brussels Free University, Laarbeeklaan 101, 1090 Brussels, Belgium 1 To whom correspondence should be addressed. stratis@easynet.be BACKGROUND: Evaluation of the association between estradiol (E 2 ) levels on the day of hcg administration and pregnancy achievement in IVF has so far yielded conflicting results. The purpose of the present study was to systematically review the above association in cycles down-regulated with GnRH analogues. METHODS: Literature search was performed using Medline, Embase ( ) and the Cochrane Library. Additionally, references of retrieved articles were hand-searched. Only full articles published in peer-reviewed medical journals were considered for analysis. RESULTS: All the eligible studies (n 5 9) involved the use of GnRH agonists and were retrospective. Two studies (including 191 patients) suggested that the higher the E 2 levels on the day of hcg administration, the higher the probability of pregnancy. However, five studies (including 1875 patients), did not support an association between E 2 levels on the day of hcg administration and pregnancy rates. Moreover, two of the studies including (1286 patients) suggested that high E 2 levels on the day of hcg administration are associated with a decreased probability of pregnancy. If we consider only studies in which criteria used for administering hcg include follicular development but not E 2 levels (including 2687 patients), there is no study suggesting a positive association between E 2 levels on the day of hcg administration and pregnancy achievement. CONCLUSIONS: Currently there is no high-quality evidence to support or deny the value of E 2 determination on the day of hcg administration for pregnancy achievement in IVF cycles, where pituitary down-regulation is performed with GnRH agonists. Existing retrospective studies suggest that there is no positive association. However, in order to arrive at recommendations for clinical practice, there is a need to perform well-designed prospective studies in both agonist and antagonist cycles. Key words: estradiol/gnrh agonist/ivf/pregnancy rate Introduction Establishment of pregnancy in human IVF requires transfer of morphologically adequate embryos in the uterus. This is achieved by the retrieval of multiple cumulus oocyte complexes (COC) after ovarian stimulation. Ovarian stimulation is necessary for multiple follicular development and is accompanied by supraphysiological serum E 2 levels. Assessment of the role of E 2 levels for IVF outcome has been the focus of interest for many years. High E 2 levels have been shown to affect embryonic adhesion (Valbuena et al., 2001), while a negative association between the probability of pregnancy and E 2 levels on the day of hcg has been reported (Table II). On the other hand, several studies have suggested that pregnancy achievement is independent of E 2 levels on the day of hcg administration or that there is a positive association between the two (Table II). At present, the importance of supraphysiological E 2 levels on the day of hcg administration for the probability of pregnancy in IVF remains unclear. The objective of the present study was to systematically review the association between E 2 levels on the day of hcg administration and pregnancy achievement in IVF cycles, in which GnRH analogues were used for down-regulation. Materials and methods Identification of studies Literature search was performed using Medline, Embase ( ) and the Cochrane Library, for publications in English, French, German or Italian. Additionally, references of retrieved articles were hand-searched. The search strategy was based on the following clinical question: are E 2 levels on the day of hcg administration associated with pregnancy achievement in IVF cycles where downregulation is used? The search terms used in all databases were: oestradiol, estradiol, pregnancy, pregnancy rate(s), implantation, in-vitro fertilisation, IVF, ART, GnRH agonist, GnRH antagonist. Meeting proceedings were not considered, since unpublished studies cannot be adequately evaluated for their design and quality Human Reproduction vol. 19 no. 11 q European Society of Human Reproduction and Embryology 2004; all rights reserved

2 E 2 and pregnancy achievement Moreover, it has been shown that although a considerable publication deficit is present in reproductive medicine, this is not accompanied by publication bias for randomized controlled trials (RCT) (Evers, 2000). Eligibility criteria All studies in which patients underwent ovarian stimulation were eligible regardless of the number of patients included. Studies examining the association between E 2 levels in the follicular phase on days other than the day of hcg administration or in the luteal phase were excluded from further screening. Moreover, studies which assessed the association between E 2 levels and pregnancy rates in IVF during a natural cycle or studies in which suppression of gonadotrophins was not performed, were not included in the current review. All studies examining the association between E 2 on the day of hcg and pregnancy rates were considered for inclusion in the systematic review, regardless of the analogue used. However, no studies exist to date regarding the association between E 2 levels on the day of hcg administration and pregnancy achievement in GnRH antagonist cycles. Therefore, only studies in which patients treated with GnRH agonists and gonadotrophins for IVF were identified and reviewed (Figure 1). The studies excluded from the analysis are listed in Table I and the studies included are listed in Table II. Data extraction The following data were recorded from each eligible study: journal in which the manuscript was published, country and period of enrolment, number of patients included, number of cycles performed, year of publication, baseline characteristics of the patients included, type and protocol of ovarian stimulation, type of gonadotrophin administered, criteria used for hcg administration, day of embryo transfer, type of luteal support administered and method of E 2 assessment. Quality evaluation of the studies identified Quality parameters included the type of the study (retrospective or prospective), the description of inclusion and exclusion criteria, the presence of power calculations and the use of well-specified study outcomes. Evaluation also considered whether quality characteristics of the studies identified changed over time. Any disagreement Table I. Excluded studies and reason for their exclusion Author Forman et al. (1988) Gratton et al. (1993) Pelicer et al. (1996) Simon et al. (1998) Loumaye et al. (1998) Sharara and McClamrock (1999a) Lindheim et al. (1999) Levi et al. (2001) Reason for exclusion Stimulation performed without pituitary down-regulation Down-regulation was not used in all cycles performed Duplication of data a ; analysis was not performed according to estradiol levels on the day of hcg a The same patients have been analysed in Sharara and McClamrock (1999a). Table II. Classification of the studies included in the review according to type of association between estradiol levels on the day of hcg administration and pregnancy rates Positive association Chenette et al. (1990), Gelety and Buyalos (1995) Mettler and Tavmergen (1989), Sharara and McClamrock (1999b), Dor et al. (1992), Papageorgiou et al. (2002), Chi-Huang Chen et al. (2003) Negative association Simon et al. (1995), Ng et al. (2000) between the authors about exclusion/inclusion of studies in the current review, their relevance or their validity was resolved through discussion. Quality evaluation of the retrospective studies in this review did not include a quality score. Up to date, quality assessment scales that have been used for retrospective studies have not been fully validated or shown to include criteria that are associated with the effect size (outcome) in empiric studies (Cochrane collaborative review group on HIV infection and AID (2004). Results Literature search resulted in the retrieval of nine studies for analysis, which appear in Table II. The association between E 2 levels on the day of hcg administration and pregnancy achievement has not so far been examined in GnRH antagonist cycles. Figure 1. Study selection process. Sample size and statistical power of the studies identified All the studies included in the current review were retrospective. Two studies analysed, 100 patients, while three studies analysed. 500 IVF patients (median number of patients included: 152 patients). Overall, 3352 patients (3508 cycles) were considered in the current review (Table III). Power analysis was not performed in any of the studies identified. 2447

3 I.P.Kosmas, E.M.Kolibianakis and P.Devroey Table III. Studies evaluating the association between estradiol (E 2 ) levels on the day of hcg administration and pregnancy achievement Study (year) Country and period of enrolment Sample size (patients/cycles) Inclusion criteria Mean age of patients (years) Type of association between E 2 levels on the day of hcg administration and pregnancy achievement Mettler and Tavmergen (1989) Germany, /94 All IVF patients with oocyte retrieval Not mentioned Chenette et al. (1990) USA, /141 All IVF patients with oocyte retrieval Not mentioned Positive Dor et al. (1992) Israel, a /216 FSH,10 IU/LT, no male infertility, 35.7 (range 25 40) at least one oocyte retrieved Simon et al. (1995) Spain, /177 Normal and high responders to ovarian stimulation for IVF Normal responders: 33.1 ^ 3.8 b ; high responders: Negative Gelety and Buyalos (1995) Sharara and McClamrock (1999b) Ng et al. (2000) Papageorgiou et al. (2002) Chi-Huang Chen et al. (2003) USA, /50 Consecutive IVF patients with E pg/ml on the day of hcg or with E 2,3000 pg/ml USA, /106 FSH,11 IU/LT, no poor ovarian response Hong Kong, Inclusion criteria and characteristics of the patients analysed Cause of infertility was not taken into consideration among the inclusion criteria in the studies reviewed (Table III). The mean age of the patients included was not always mentioned in the studies reviewed (Table III). In addition, the mean FSH level prior to starting stimulation was only reported by Sharara and McClamrock (1999b). Type of stimulation used and criteria for hcg administration Type of ovarian stimulation used in the studies identified appears in Table IV. Most of the studies used leuprolide acetate (n ¼ 5) for pituitary down-regulation. In only one study ovarian stimulation was performed with recombinant FSH. hcg was used for triggering final oocyte maturation; however, the criteria employed for its administration varied considerably between authors and were not strict (Table IV). In the majority of the studies, i.m. progesterone was used for luteal support. Assessment of E 2 Assessment of E 2 was performed in most studies using radioimmunoassay methods. Only two studies have used a product from the same company for E 2 determination (Table IV). Changes in the quality of the studies included as a function of time Study design did not change with time, as all the studies identified were retrospective. However, the number of patients analysed appears to increase with time (Table III). In the first five studies performed from 1989 until 1995, 30.3 ^ 2.8 b 32.2 ^ 0.7 c Positive 33 ^ 4 b 1122/1122,40 years,,3 embryos transferred 33.1 ^ 3.2 c Negative France, /905,43 years, FSH.15 IU Not mentioned USA, /697 All IVF patients in their first cycle Not mentioned a not specified (assumed to be the same as the number of cycles) b SD. c SEM patients were included (678 cycles). In the last four studies performed after 1995, 2687 patients were analysed (2830 cycles). These represent 80.2% of the patients considered in this review. Classification of the studies according to the type of association between E 2 levels on the day of hcg administration and pregnancy achievement The majority of the studies (n ¼ 5; 1875 patients) do not support an association between E 2 levels on the day of hcg administration and pregnancy achievement (Table II). Two studies (191 patients) suggested that the higher the E 2 levels on the day of hcg administration, the higher the pregnancy rates achieved (Table II). On the contrary, two studies including a total of 1286 patients, suggested a detrimental role of high E 2 levels on the day of hcg administration for pregnancy achievement (Table II). After 1995, there have been no studies suggesting that higher E 2 levels are associated with higher pregnancy rates. Concomitantly, after 1995 the criteria used for hcg administration no longer include E 2 levels, in contrast to studies performed prior to Implantation rates were reported in only five of the nine studies included in this systematic review. Where reported, implantation rates showed the same association with estradiol levels on the day of hcg as pregnancy rates. Studies included in the current review Studies suggesting that pregnancy rates are positively associated with E 2 levels on the day of hcg administration Chenette et al. (1990) grouped 216 patients in centiles (thirds) according to E 2 levels on the day of hcg

4 Table IV. Ovarian stimulation characteristics in the studies included in the current review Study (year) Type of down-regulation Gonadotrophin regimen Criteria for hcg administration Day of embryo transfer Luteal support Method of estradiol assessment Primary outcome Type of association between E 2 levels on the day of hcg administration and pregnancy achievement Mettler and Tavmergen (1989) Chenette et al. (1990) Dor et al. (1992) Simon et al. (1995) Gelety and Buyalos (1995) Sharara and McClamrock (1996) Ng et al. (2000) Papageorgiou et al. (2002) Chi-Huang Chen et al. (2003) Triptorelin Leuprolide acetate Triptorelin, Leuprolide acetate, Leuprolide acetate, Leuprolide acetate, long/short protocol Buserelin, Triptorelin, short protocol Leuprolide acetate, Long protocol, hmg Long protocol, hmg E 2 $ pg/ml/ follicle $15 mm E 2 $ pg/ml per co-dominant follicle and leading follicle mm hmg E pg/ml and $2 follicles $16 mm FSH þ hmg E pg/ml and $2 follicles $19 mm Day 2 hcg Radioimmunoassays mono antibody technique/e 2 Biermann, Germany Day 2 Progesterone i.m. Radioimmunoassays/ Tandem, Hybritech, USA Day 2 Not mentioned Radioimmunoassays/ Diagnostic Products, USA Day 3 Micronized progesterone Immunoenzymatic assay/meia; Imx Abbot Scientific SA hmg E 2 $500 pg/ml and $2 follicles $18 mm Day 2 Progesterone i.m. Radioimmunoassays/ Pantex, USA FSH þ hmg $3 follicles $18 mm Day 3 Progesterone i.m. Radioimmunoassays/ Coat-a-Count Diagnostic Products Corp., USA hmg rfsh (follitropin a/b) Leading follicle $18 mm and $3 follicles.15 mm Day 2 hcg or progesterone (i.m. or vaginal) $2 follicles $15 mm Day 2 Micronized progesterone and hcg FSH $2 follicles $18 mm Day 3, Day 5 Progesterone i.m. Not mentioned/ Diagnostic Products, USA Immunoenzymatic assay/bayer Not mentioned/not mentioned Pregnancy rate a per oocyte retrieval Clinical pregnancy per oocyte retrieval Clinical pregnancy rate per oocyte retrieval Pregnancy rate per cycle started a Clinical pregnancy rate per embryo transfer Clinical pregnancy per oocyte retrieval Clinical pregnancy rate per embryo transfer Pregnancy rate a Clinical pregnancy per embryo transfer Positive Negative Positive Negative E2 and pregnancy achievement 2449 Downloaded from by guest on May 31, 2013

5 I.P.Kosmas, E.M.Kolibianakis and P.Devroey (,1722, ,.2777 pg/ml) and noted that significantly higher clinical pregnancy rate per oocyte retrieval was observed in the highest E 2 group compared with the other groups. Criteria used for hcg administration were based on E 2 levels and follicular development. Characteristics of the included patients in each of the E 2 groups analysed were not mentioned. Gelety and Buyalos (1995) compared 25 patients undergoing IVF with E 2 levels on the day of hcg.5000 pg/ml with 25 patients in whom E 2 levels were,3500 pg/ml (control group). The control group was matched for age, duration and type of infertility, and stimulation protocol. A higher clinical pregnancy rate per embryo transfer was observed in the high E 2 group. More embryos were transferred in the high compared with the low E 2 group, although the difference was marginally significant. Criteria used for hcg administration were based on E 2 levels and follicular development. Studies suggesting that pregnancy rates are not associated with E 2 levels on the day of hcg administration Mettler and Tavmergen (1989) divided 94 patients in three groups according to E 2 levels on day of hcg administration (,800, and.1500 pg/ml). No significant difference in pregnancy rates per oocyte retrieval was present between the groups compared. Characteristics of the patients included in the groups analysed were not mentioned. Criteria used for hcg administration included E 2 levels and follicular development. Dor et al. (1992) divided 216 patients into three groups according to E 2 levels on the day of hcg administration (, 500, , pg/ml). between E 2 levels on the day of hcg and clinical pregnancy rate per oocyte retrieval could be shown. Criteria used for hcg administration included E 2 levels and follicular development. Characteristics of the included patients in the groups analysed were not mentioned. Sharara and McClamrock (1999b) analysed 106 patients in four groups according to E 2 levels on the day of hcg administration, using thousand interval increments (, 2000, , ,.4000 pg/ml). between E 2 levels on the day of hcg administration and clinical pregnancy per oocyte retrieval could be shown. Characteristics of the included patients in the groups analysed were not mentioned. Criteria used for hcg administration did not include E 2 levels and were based only on follicular development. Papageorgiou et al. (2002) divided 762 patients into three groups according to centile analysis of E 2 levels on the day of hcg administration (0 10th centile, 10 90th centile, th centile). No significant association between E 2 levels on the day of hcg administration and pregnancy rates could be shown. Criteria used for hcg administration did not include E 2 levels and were based only on follicular development. Characteristics of the included patients in the groups analysed were not mentioned. Chen et al. (2003) used centile analysis to divide 697 patients in three groups according to E 2 levels on day of hcg administration (, 1289, , pg/ml) could be shown between E 2 levels on the day hcg was administered and clinical pregnancy rate per embryo transfer. A subgroup analysis suggested that a positive association between E 2 levels and the probability of pregnancy was present only in cycles in which embryo transfer was performed on day 5 of in vitro culture. Criteria used for hcg administration did not include E 2 levels and were based exclusively on follicular development. Studies suggesting that pregnancy rates are negatively associated with E 2 levels on the day of hcg administration Simon et al. (1995) divided patients in eight groups according to E 2 levels on the day of hcg administration using 500 pg/ml increments (from 500 to.3500 pg/ml). A significantly lower pregnancy rate per cycle started was observed in the groups of patients with higher E 2 levels (E pg/ml) compared with those with lower E 2 levels (E 2,2500 pg/ml). Baseline characteristics did not differ between the groups compared. E 2 levels and follicular development were used to time hcg administration. Ng et al. (2000) divided patients into three groups according to E 2 levels on the day of hcg administration (,10 000, , pmol/l). A significantly lower clinical pregnancy rate per embryo transfer was observed in the group of patients with E 2 levels pmol/l compared with those with E 2 levels between and pmol/l. Similar numbers of embryos were transferred in the groups compared. Criteria used for hcg administration were based on follicular development. Discussion The purpose of the current review was to evaluate existing evidence regarding the association between E 2 levels on the day of hcg administration and pregnancy achievement in IVF cycles where gonadotrophin down-regulation was used. In theory, if there is a positive association between E 2 levels on the day of hcg administration and pregnancy rates, then low E 2 levels on that day should be considered as a bad prognostic factor for pregnancy. In this case, the cycle should either be cancelled or the follicular phase be prolonged in order to achieve higher E 2 levels, but avoiding the occurrence of ovarian hyperstimulation syndrome. Evaluation of the prolongation of follicular phase has so far been performed with conflicting results (Clark et al., 1991; Dimitry et al., 1991; Tan et al., 1992). Conversely, if there is a negative association between E 2 levels on the day of hcg administration and pregnancy rates, then low estradiol levels on that day should be considered as a good prognostic factor for pregnancy and there is a need to assess the value of maintaining low E 2 levels during ovarian stimulation, in order to increase the chance of conception (Simon et al., 1998). Finally, if there is a lack of association between E 2 levels on the day of hcg administration and pregnancy achievement, then E 2 assessment should not be incorporated in the criteria used for hcg administration and should only be

6 E 2 and pregnancy achievement performed to assess the risk of ovarian hyperstimulation syndrome (D Angelo et al., 2004). No data from prospective studies are available at present regarding the role of E 2 levels on the day of hcg administration for the achievement of pregnancy. Although supraphysiological E 2 levels during ovarian stimulation for IVF represent one of the major deviations undergone by the female endocrine environment compared with the natural cycle, their significance for pregnancy achievement in IVF has only been assessed retrospectively. The majority of the studies included in the current review (7/9), among these the three largest trials including 2581 patients (76.9% of patients), do not support the view that there is a positive association between E 2 levels on the day of hcg administration and pregnancy achievement in IVF cycles, in which inhibition of premature LH surge is performed with GnRH agonists. The retrospective nature of the studies included in the current review might partially explain the controversy surrounding the results reported. However, this might also be due to additional confounding factors such as the small number of patients included in some studies, especially those published before 1995 and the use of different assays for the assessment of E 2 levels. Moreover, in three of the studies analysed, patients contributed more than one cycle for analysis. However, data were analysed as independent data by Student s t-test instead of using generalized linear models. This might artificially decrease P-values and yield confusing conclusions. Further confounding factors might lie in the existing differences in the inclusion/exclusion criteria used, especially as baseline patient characteristics were not always reported. Differences in the type of analogue or in the analogue protocol used and the day that embryo transfer took place, might also have affected the results observed. The study by Chen et al. (2003) provided some evidence that a differential association between E 2 levels on the day of hcg administration and pregnancy rate may be present, depending on the day that embryo transfer is carried out. This might require further investigation. No power calculations have been performed in the studies included in the present review. This could have provided information on the ability of the studies to show an association between E 2 levels on the day of hcg administration and pregnancy achievement, if such an association exists. In addition, analysis methods varied considerably between studies. Percentile analysis of E 2 levels on the day of hcg administration has been used to create groups of patients for comparison, but also arbitrary threshold E 2 levels have been employed for that purpose. Studies supporting the view that there is a positive or a negative association between E 2 levels and pregnancy achievement failed to converge to a similar threshold level of E 2 at which the probability of pregnancy is significantly altered. This might have been achieved if receiver operating curve characteristics had been used for analysis. However, identifying a threshold level of E 2 that predicts the probability of pregnancy might not be feasible, as an Figure 2. Studies evaluating association between E 2 levels on the day of hcg administration and pregnancy achievement according to the criteria used for triggering final oocyte maturation. 2451

7 I.P.Kosmas, E.M.Kolibianakis and P.Devroey Table V. Characteristics of a properly designed study evaluating the association between estradiol (E 2 ) levels on the day of hcg administration and pregnancy achievement Type of study Population Stimulation scheme Type of analogue Criteria for hcg administration Embryo transfer Luteal support Main outcome measure Data analysis Prospective Clearly defined inclusion and exclusion criteria Patients should contribute only one cycle in the study The same initiating dose of gonadotrophin should be used, preferably fixed, or adjusted according to follicular development only Either agonist or antagonist a Criteria should not include E 2 levels and should be strict (hcg is administered as soon as an arbitrary criterion for follicular development is reached) b All patients should have the same number of embryos transferred to ensure that E 2 groups analysed are comparable for implantation and pregnancy rates Similar for all patients Ongoing pregnancy rate Percentile analysis to define groups of E 2 to be used in the analysis objectively Receiver operating characteristics curves to assess the predictive value of E 2 on pregnancy achievement. Ultimately logistic regression should be performed: dependent variable achievement of ongoing pregnancy, independent variables: E 2 levels on the day of hcg, and other relevant covariates a A meta-analysis or a systematic review should be carried out separately for each analogue and type of protocol used. b At present the optimal moment for administered hcg has not been established, but in antagonist cycles probably should be no later than the first day when $3 follicles of $17 mm diameter are present (Kolibianakis et al., 2004). association between E 2 levels on the day of hcg administration and pregnancy achievement may not be present at all. Instead pregnancy rates might be associated to exposure to hormones (both E 2 and progesterone) during the follicular phase and not to hormone levels on a certain day (Kolibianakis et al., 2003). From the nine included studies, only one reported progesterone levels in the groups of patients analysed according to estradiol levels on the day of hcg (Simon et al., 1995). In that study it was mentioned that the ratio of estradiol/progesterone on the day of hcg is not important in predicting implantation. No study performed after 1995 has reported a positive association between E 2 levels and pregnancy rates. This might be related to an increase in the sample size in the more recently performed studies, which has resulted in increased accuracy of the results reported. However, it might also be explained by the concomitant change observed in the criteria used for hcg administration. All four studies performed after 1995 have not taken into consideration E 2 levels for administering hcg. The type of association between E 2 levels and pregnancy outcome according to criteria used for hcg administration appears in Figure 2. It is difficult to understand how accurately the association between E 2 levels and pregnancy rates can be evaluated when E 2 levels are incorporated in the criteria used for hcg administration. Obviously this results in the alteration of the parameter of interest (E 2 ) for which the association with pregnancy achievement needs to be assessed. In order to evaluate the association between E 2 levels on the day of hcg administration and pregnancy rates, criteria used for triggering final oocyte maturation in IVF should only be based on follicular development. This is true for studies that were performed after 1995, which suggest that there is no positive association between E 2 levels on the day of hcg administration and pregnancy achievement. It should be noted that the criteria used for administering hcg in the above studies were not strict. hcg was administered when $x number of follicles of $ y diameter were present on ultrasound. This means that hcg was given either on the 2452 day the above criterion was met or later. Consequently, two patients in the same study may have been treated differently. The true association between E 2 levels on the day of hcg administration and pregnancy rates can probably be evaluated accurately if criteria for triggering final oocyte maturation do not take into consideration E 2 levels and are strict. In this case, hcg is administered as soon as a threshold of follicular development is reached, regardless of E 2 levels (Clark et al., 1991; Kolibianakis et al., 2004). In conclusion, currently there is no high-quality evidence to support or deny the value of E 2 determination on the day of hcg administration for pregnancy achievement in IVF cycles, where pituitary down-regulation is performed with GnRH agonists. Existing retrospective studies suggest that there is no positive association. However, in order to arrive at recommendations for clinical practice there is a need to perform well-designed prospective studies in both agonist and antagonist cycles (Table V). Acknowledgements This work is supported from Funds for Scientific Research, Flanders. Excluded studies Forman R, Fries N, Testart J, Belaisch-Allart J, Hazout A and Frydman R (1988) Evidence for an adverse effect of elevated serum E 2 concentrations on embryo implantation. Fertil Steril 49, Gratton RJ, Nisker JA, Daniel S, Toth S, Gunter J, Kaplan BR, Tummon IS and Yuzpe AA (1993) An aggressive philosophy in controlled ovarian stimulation cycles increases pregnancy rates. Hum Reprod 8, Levi AJ, Drews MR, Bergh PA, Miller BT and Scott RT Jr (2001) Controlled ovarian hyperstimulation does not adversely affect endometrial receptivity in vitro fertilization cycles. Fertil Steril 76, Lindheim SR, Cohen MA, Chang PL and Sauer MV (1999) Serum progesterone before and after human chorionic gonadotropin injection depends on the E 2 response to ovarian hyperstimulation during in vitro fertilization embryo transfer cycles. J Assist Reprod Genet 16, Pellicer A, Valbuena D, Cano F, Remohi J and Simon C (1996) Lower implantation rates in high responders: evidence for an altered endocrine milieu during the preimplantation period. Fertil Steril 65, Sharara FI and McClamrock HD (1999a) Ratio of oestradiol concentration on the day of human chorionic gonadotrophin administration to mid-luteal

8 E 2 and pregnancy achievement oestradiol concentration is predictive of in-vitro fertilization outcome. Hum Reprod 14, Simon C, Garcia Velasco JJ, Valbuena D, Peinado JA, Moreno C, Remohi J and Pellicer A (1998) Increasing uterine receptivity by decreasing E 2 levels during the preimplantation period in high responders with the use of a follicle-stimulating hormone step-down regimen. Fertil Steril 70, References Chen CH, Zhang X, Barnes R, Confino E, Milad M, Puscheck E and Kazer RR (2003) Relationship between peak serum E 2 levels and treatment outcome in in-vitro fertilization cycles after embryo transfer on day 3 or day 5. Fertil Steril 80, Chenette PE, Sauer MV and Paulson RJ (1990) Very high serum E 2 levels are not detrimental to clinical outcome of in vitro fertilization. Fertil Steril 54, Clark L, Stanger J and Brinsmead M (1991) Prolonged follicle stimulation decreases pregnancy rates after in vitro fertilization. Fertil Steril 55, Cochrane collaborative review group on HIV infection and AIDS (2004) D Angelo A, Davies R, Salah E, Nix BA and Amso NN (2004) Value of the serum E 2 level for preventing ovarian hyperstimulation syndrome: a retrospective case control study. Fertil Steril 81, Dimitry ES, Oskarsson T, Conaghan J, Margara R and Winston RM (1991) Beneficial effects of a 24 h delay in human chorionic gonadotropin administration during in-vitro fertilization treatment cycles. Hum Reprod 6, Dor J, Seidman DS, Ben-Shlomo I, Levran D, Karasik A and Mashiach S (1992) The prognostic importance of the number of oocytes retrieved and E 2 levels in poor and normal responders in in vitro fertilization (IVF) treatment. J Assist Reprod Genet 9, Evers JL (2000) Publication bias in reproductive research. Hum Reprod 15, Gelety TJ and Buyalos RP (1995) The influence of supraphysiologic E 2 levels on human nidation. J Assist Reprod Genet 12, Kolibianakis EM, Albano C, Kahn J, Camus M, Tournaye H, Van Steirteghem AC and Devroey P (2003) Exposure to high levels of luteinizing hormone and E 2 in the early follicular phase of gonadotropin-releasing hormone antagonist cycles is associated with a reduced chance of pregnancy. Fertil Steril 79, Kolibianakis EM, Albano C, Camus M, Tournaye H, Van Steirteghem AC and Devroey P (2004) Prolongation of the follicular phase in in vitro fertilization results in a lower ongoing pregnancy rate in cycles stimulated with recombinant follicle-stimulating hormone and gonadotropin-releasing hormone antagonists. Fertil Steril 82, Mettler L and Tavmergen EN (1989) Significance of E 2 values in IVF ET under a combined GnRH analogue desensitization and simultaneous gonadotropin stimulation for the outcome of pregnancies. Hum Reprod 4 (Suppl 8), Papageorgiou T, Guibert J, Goffinet F, Patrat C, Fulla Y, Janssens Y and Zorn JR (2002) Percentile curves of serum E 2 levels during controlled ovarian stimulation in 905 cycles stimulated with recombinant FSH show that high E 2 is not detrimental to IVF outcome. Hum Reprod 17, Sharara FI and McClamrock HD (1999b) High E 2 levels and high oocyte yield are not detrimental to in vitro fertilization outcome. Fertil Steril 72, Simon C, Cano F, Valbuena D, Remohi J and Pellicer A (1995) Clinical evidence for a detrimental effect on uterine receptivity of high serum E 2 concentrations in high and normal responder patients. Hum Reprod 10, Simon C, Garcia Velasco JJ, Valbuena D, Peinado JA, Moreno C, Remohi J and Pellicer A (1998) Increasing uterine receptivity by decreasing E 2 levels during the preimplantation period in high responders with the use of a follicle-stimulating hormone step-down regimen. Fertil Steril 70, Tan SL, Balen A, el Hussein E, Mills C, Campbell S, Yovich J and Jacobs HS (1992) A prospective randomized study of the optimum timing of human chorionic gonadotropin administration after pituitary desensitization in in vitro fertilization. Fertil Steril 57, Valbuena D, Martin J, de Pablo JL, Remohi J, Pellicer A and Simon C (2001) Increasing levels of E 2 are deleterious to embryonic implantation because they directly affect the embryo. Fertil Steril 76, Yu Ng EH, Yeung WS, Yee Lan Lau E, So WW and Ho PC (2000) High serum E 2 concentrations in fresh IVF cycles do not impair implantation and pregnancy rates in subsequent frozen-thawed embryo transfer cycles. Hum Reprod 15, Submitted on May 26, 2004; accepted on July 23,