Percentile curves of serum estradiol levels during controlled ovarian stimulation in 905 cycles stimulated with recombinant FSH show that high estradiol is not detrimental to IVF outcome

Theocharis Papageorgiou, Juliette Guibert, François Goffinet, Catherine Patrat, Yvonne Fulla, Yvette Janssens and Jean-René Zorn1

Service Gynécologie-Obstétrique III, Clinique Universitaire Baudelocque, Hôpital Cochin, 123 bd Port-Royal, Paris 75014, France


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: High, normal and poor responders are usually defined by reference to subjectively selected estradiol (E2) levels at days 4–6 and the day of hCG administration (d-hCG). The purpose of this study was to use E2 percentile curves from day 5 until d-hCG to determine high, normal and poor responders, and to predict IVF outcome. METHODS: In this retrospective study, 762 patients underwent 905 cycles with a GnRH agonist/recombinant FSH short protocol. They were divided into three groups according to their age. Percentile E2 curves according to E2 levels were plotted. High responders were those patients with E2 levels above the 90th percentile, normal responders had E2 between the 10th and 90th percentiles, and poor responders had E2 below the 10th percentile. RESULTS: IVF outcome, expressed as number of oocytes, total embryos obtained and number of high grade embryos, was significantly better for patients with E2 above the 90th percentile at d-hCG for the three age groups and at day 5 for group A (<35 years). Pregnancy rates were higher for high responders, but the difference did not reach statistical significance. CONCLUSIONS: Percentile curves can be useful in controlled ovarian stimulation cycles to define high, normal and poor responders, and also to predict IVF outcome.

Key words: estradiol/hyperstimulation/ovarian stimulation/percentiles/rFSH


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Controlled ovarian stimulation (COS) for IVF cycles is usually monitored by serum estradiol (E2) levels and pelvic ultrasonography with two purposes: (i) to obtain an adequate number of mature oocytes, and (ii) to prevent the risk of severe ovarian hyperstimulation syndrome (OHSS).

In order to decide whether to proceed with an ongoing cycle and to define the so-called poor and high responders, Phelps et al. and Khalaf et al. used different cut-off E2 levels at day 4–6 and on the day of hCG administration (d-hCG) of the stimulation cycle. Poor responders, defined by E2 levels <75 pg/ml at day 4 (Phelps et al., 1998Go) or <50 pg/ml at day 6 (Khalaf et al., 2000Go), were associated with low pregnancy and high cancellation rates. Follicular aspiration in patients with E2 levels <500 pg/ml at d-hCG yields less than five mature oocytes (Sharara and McClamrock, 1999Go). On the other hand, high responders, defined as having E2 levels >3000 pg/ml at d-hCG, are more often subject to OHSS (Asch et al., 1991Go; Morris et al., 1995Go). However, the influence of high E2 levels on the outcome of IVF cycles is still controversial. Simon et al. suggested a detrimental effect of high E2 on endometrial receptivity (Simon et al., 1998Go), but others found no adverse consequences (Sharara and McClamrock, 1999Go; Levi et al., 2001Go).

The use of E2 levels to differentiate poor, normal and high responders, and to predict IVF outcome, has been the subject of debate. In most studies, cut-off levels were chosen arbitrarily and depend on various factors. Percentile curves offer a more objective definition of these three categories of patients. To our knowledge, only one study has used percentile curves to adjust the doses of gonadotrophins to the ovarian response and to evaluate IVF outcome in stimulation cycles (Forman et al., 1988Go). In our centre, since 1989 we have routinely based COS on percentile E2 curves for long and short GnRH agonist protocols combining triptorelin and hMG (J.-R.Zorn, unpublished data). Gonadotrophin dose and ovulation induction by hCG are adjusted according to the pattern of E2 levels on the percentile curve of the corresponding protocol.

The objectives of the present study were: (i) to determine the ovarian response in COS with recombinant (r)FSH in a short GnRH agonist protocol by using E2 percentile curves; (ii) to define low, normal and high responders; and (iii) to predict the IVF outcome by using percentile E2 serum levels at day 5 and d-hCG.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Patients
Using a computerized database, records of 2184 (COS) cycles for IVF/embryo transfer performed between January 1996 and October 2000 at the University IVF Center of Paris-Baudelocque, Cochin Hospital, Paris, were reviewed. During this period, 762 couples in an IVF programme for male, female, mixed or unexplained infertility underwent 905 conventional IVF or ICSI cycles using a GnRH agonist short (flare) protocol combining triptorelin and rFSH. Group A (women <35 years old) comprised 461 cycles, group B (35–38 years) 234 cycles and group C (>38 years) 210 cycles. There were 402 cycles of IVF and 503 cycles of ICSI. Exclusion criteria were age >43 years, day 3 FSH >15 IU/l or participation in other COS protocols. Due to the retrospective study design based on routine practice, consultation with the hospital ethics committee was not required.

Drugs and COS protocols
GnRH agonist
Triptorelin (Decapeptyl; Ipsen, Paris, France) was administered in a dose of 0.1 mg/day, starting 1–2 days after withdrawal bleeding.

rFSH
Follitropin-ß (Puregon; NV Organon, Paris, France) or follitropin-{alpha} (Gonal-F; Serono, Geneva, Switzerland) were used during this study. A total of 787 cycles were stimulated with Gonal-F and 118 with Puregon.

hCG
hCG (Gonadotrophine Chorionique Endo; Organon) was used at a dose of 5000 IU/ampoule. Patients received norethisterone (NE) (Primolut Nor; Schering, Paris, France) 10 mg daily for 8–16 days starting from day 3 of a natural cycle. Triptorelin, 0.1 mg daily, was given for 6–15 days starting 4 days after the last day of NE. At day 3 of triptorelin treatment, rFSH was started at a daily dose of 3–6 ampoules (75 IU) per day. Generally, the initial dose for patients in group A was 3 ampoules, for group B, 4 ampoules and for group C, 6 ampoules. From day 5 of the stimulation cycle, doses of rFSH were adjusted according to the number of follicles found on ultrasound and to the E2 levels. hCG 10 000 IU was injected when two or more follicles >15 mm were found. Oocyte retrieval was performed transvaginally under ultrasound guidance.

Embryology assessments
After liquefaction at 37°C, motile sperm were selected through a two-step density gradient (Puresperm; JCD, Lyon, France). The pellet was washed and resuspended in human tubal fluid (HTF) culture medium (Irvine scientific, CA, USA) supplemented with 7.5% autologous female decomplemented serum. For the IVF procedure, oocytes were inseminated with 10 000 motile sperm in microdroplets of 30 µl of HTF with 7.5% serum under mineral oil and kept under 5% CO2 at 37°C. For ICSI, the cumulus cells around the oocytes were removed with 80 IU/ml hyaluronidase, and the oocytes were checked for their nuclear status. Only oocytes in metaphase II were microinjected and kept in HTF + 7.5% serum under 5% CO2 at 37°C. The oocytes were checked 16–20 h after insemination or microinjection for evidence of fertilization. The oocytes were considered to be normally fertilized when two pronuclei were visible. Fertilized oocytes were kept in culture for an additional 24 h (HTF + 15% serum) and checked for cleavage. The embryos were evaluated according to the number of blastomas and the degree of fragmentation. Up to four embryos were transferred, and any remaining high grade embryos were cryopreserved.

Embryo transfer and luteal support
Two-day-old embryos were transferred using the Frydman catheter (CCD Paris, France). The number of embryos transferred was typically two, and occasionally up to four in patients >39 years old or when embryo quality was poor. Luteal support was administered in the form of vaginal progesterone (Utrogestan; Besins International, Paris, France) 800 mg per day for 15 days, and hCG 1500 IU was injected 48 h after the transfer in patients whose peak E2 levels were above the 90th percentile (>P90).

End points and outcome measures
The main outcome for plotting rFSH E2 patterns for the three different age groups was E2 levels from day 5 to d-hCG. IVF outcomes were expressed as the number of oocytes, number of embryos obtained, number of high grade embryos, E2 peak levels and pregnancy rates (PRs) at day 5 and d-hCG. High, normal and poor responders were defined as those women with E2 levels at day 5 and d-hCG >P90, between 10th and 90th percentile (P10–P90) and below the 10th percentile (<P10) respectively. The P10 and P90 cut-offs were chosen based on previous reports for other fields of gynaecology–obstetrics (Lubchenco et al., 1963Go; Leroy and Lefort, 1971Go).

Laboratory analysis
E2 determination was performed using an immunoenzyme assay by sequential competitive technique on an Immuno-1 multiparameter analyser (Bayer reference ES2). The capture antibody used was a rabbit polyclonal antibody against E2 conjugated to fluorescein. The competitive marker was constituted by E2 conjugated to alkaline phosphatase. Separation was performed using a dispersed solid phase that is achieved by magnetic particles linked to monoclonal antibody anti-fluorescein. The signal was read on the spectrophotometer at 405 and 450 nm. E2 standard USP in a saline buffer bovine albumin with sodium azoture was used for calibration. The molar conversion factor was 3.67 (pmol/l = pg/mlx3.67). The analytical range was from 0 to 3600 pg/ml (0 to 13 200 pmol/l). Samples with concentration values exceeding the maximal calibrator can be diluted in the zero calibrator or in an albumin buffer. Analytical sensitivity was 5 pg/ml, corresponding to twice the SD of the zero calibrator. The functional limit of detection at the 20% of inter-assay coefficient of variation (CV) was near 10 pg/ml (36.7 pmol/l). Inter-assays imprecision show CV <20% for a high concentration of E2 (>300 pg/ml), and near 5% for a low concentration of E2 (<30 pg/ml). Blood samples were collected in tubes without anticoagulant and separated by centrifugation. Assays were run on the day of sampling with 50 µl of serum. The method remained unchanged during the study period.

Statistical analysis
Results are expressed as means ± SD. Statistical comparisons were made by using the {chi}2-test and analysis of variances when appropriate. Statistical significance was set at P < 0.05. Statview software (Abacus, F-4.5, 1999; Berkeley, CA, USA) was used for statistical analysis.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
E2 thresholds (P10/P90) at day 5 and d-hCG were as follows: group A, 51/466 and 1007/3444 pg/ml; group B, 44/463 and 851/3414 pg/ml; and group C, 50/486 and 868/3155 pg/ml. Relatively high P10 and P90 levels for group C compared with group B may be explained by the higher starting doses of rFSH used for patients >38 years old. The percentile curves of E2 levels from day 5 to d-hCG are presented in Figure 1Go. The general IVF outcomes of the three groups are presented in Table IGo. As expected, lower doses of rFSH were used in younger patients (group A), resulting in higher peak E2 values, a higher number of total and metaphase II oocytes retrieved and a greater number of total and high grade (grade I and II) embryos obtained.



View larger version (11K):
[in this window]
[in a new window]
 
Figure 1. Estradiol percentile curves for group A (<35 years), group B (35–38 years) and group C (>38 years).

 

View this table:
[in this window]
[in a new window]
 
Table I. Reproductive outcome of three ages groups stimulated with rFSH in GnRH agonist short protocol
 
IVF outcomes for high, normal and poor responders in groups A, B and C are presented in Table IIGo. There was no difference between the ages of poor, high and normal responders for all groups. This eliminates the possibility of bias for comparisons between different responders in the same group of patients. The number of oocytes, total embryos and high grade embryos was higher in >P90 and lower in <P10 responders in all age groups. Differences were statistically significant (P < 0.05) only in younger patients (group A) at day 5. For d-hCG results, statistical significance was reached for all groups. A possible interpretation for this is that between day 5 and d-hCG, the dose of rFSH is adjusted according to the ovarian response. Pregnancy rates were also higher in high responders and lower in poor responders, but differences did not reach statistical significance in any group. Three patients (0.3%) (two from group A and one from group B) were hospitalized for severe OHSS. Interestingly, the three patients belonged to the P10–P90 group. Seven patients (0.8%), four from the P10–P90 group and three from the >P90 group, were treated as out-patients for moderate OHSS.


View this table:
[in this window]
[in a new window]
 
Table II. IVF outcomes for group A (<35 years), group B (35–38 years) and group C (>38 years)
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The challenge of COS is to achieve a reasonable balance between two conflicting outcomes, namely obtaining the maximum number of mature oocytes and embryos without exposing patients to the risk of severe OHSS. It is useful to identify high responders early in order to try to prevent OHSS, and to identify poor responders to prevent cancellation. The difficulty is to properly define high, normal and poor responders to COS. This was the subject of a recent discussion (McDonough, 1999Go) which pointed out that most studies suffered from small sample sizes and nearly all used different definitions based upon E2 cut-off levels. In addition, determination of E2 values depends on many factors, including hormonal assay methodology. Therefore, cut-off levels expressed as absolute values of E2 in pg or pmole/l from one centre may not be used universally.

For poor responders identified at day 5, P10 E2 were ~50 pg/ml, which are similar to those previously described in IVF cycles stimulated with gonadotrophins, either with or without GnRH agonist. In the study by Phelps et al. E2 levels <75 pg/ml at day 4 yielded poor IVF outcomes (Phelps et al. 1998Go), and for Khalaf et al. E2 levels <50 pg/ml at day 6 had higher cancellation rates and poorer IVF results (Khalaf et al., 2000Go). Sharara and McClamrock defined poor responders as patients who had <500 pg/ml E2 at d-hCG (Sharara and McClamrock, 1999Go). In the present study, it is of interest that poor responders at day 5 had better IVF outcomes than poor responders identified later at d-hCG for all groups of patients. This can be explained by the fact that doses of rFSH were increased in cases of poor response at day 5, and this increase obviously improved the ovarian response. This difference was not observed for normal responders, probably because of the unchanged initial dose of rFSH in these patients.

Phelps et al. considered high responders to be patients with E2 levels >200 pg/ml at day 4 (Phelps et al., 1998Go). Interestingly, none of our patients with E2 >P90 at day 5 was hospitalized for OHSS. This may be explained by the reduction of rFSH dose for those patients after day 5. For our high responders identified at d-hCG, P90 E2 levels were similar to those of high responders defined by others (Simon et al., 1998Go; Sharara and McClamrock, 1999Go). Generally, in our study, patients with E2 levels >P90 had significantly better IVF outcomes (expressed as numbers of oocytes, total embryos and high-grade embryos) than normal responders (P10–P90) and, as expected, than poor responders. In the only study in which E2 percentile curves were used to predict IVF outcomes, Forman et al. showed lower PRs for patients with E2 >P90 (>2320 pg/ml) (Forman et al., 1988Go). Of note is that this study was published before the GnRH agonist era. Our data are consistent with those previously published by Sharara and McClamrock (Sharara and McClamrock, 1999Go).

The issue of deleterious effects of high E2 levels on uterine receptivity has been under debate for many years. High levels of steroid hormones induce morphological alterations of the endometrium (Garcia et al., 1984Go; Bladford et al., 1997Go) and alter endometrial E2/progesterone ratios, which is associated with impairment of endometrial receptivity (Gidley-Baird et al., 1986Go). In COS, a significant reduction of the nuclear receptors for progesterone and E2 has been demonstrated in both glands and stroma (Hadi et al., 1994Go). Several studies have shown a negative effect of high E2 levels on pregnancy and implantation rates (Testart et al., 1986Go; Pellicer et al., 1996Go; Simon et al., 1998Go). Moreover, Valbueña et al. found that a high E2 concentration affects embryonic adhesion (Valbueña et al., 2001Go). However, most of these studies used small sample sizes and different definitions of high responders, or dealt with in-vitro models. Our results are from a large cohort of human subjects are consistent with previously reported data (Chenette et al., 1990Go; Tarin et al., 1992Go; Loumaye et al., 1997Go; Sharara and McClamrock, 1999Go; Levi et al., 2001Go). According to our study, we did not find any deleterious effect of high E2 levels, and patients with E2 >P90 had better IVF outcomes in all group of patients. Our three cases of severe OHSS belonged to group A (<35 years) and were normal responders (d-hCG E2 = P10–P90). No cases of severe OHSS were detected in patients with E2 >P90 at d-hCG, in contrast to previously published data which reported incidences of 0.94 and 2.9% of severe OHSS (Asch et al., 1991Go; Jaffe et al., 1993Go).

An important difference between our study and prior reports (Simon et al., 1998Go; Sharara and McClamrock, 1999Go; Khalaf et al., 2000Go) is the stimulation regimen. We cannot exclude the possibility that the type of gonadotrophin influences oocyte and embryo quality. Moreover, high responders with rFSH may behave differently. We used both types of rFSH (follitropin-{alpha} and -ß) and we do believe that this has affected our results, as previous reports have shown that there are no differences in the biological and clinical activity of the two rFSH hormone preparations (Brinsden et al., 2000Go; Harlin et al., 2002Go).

In conclusion, we believe that percentile curves can help practitioners to follow stimulation cycles and to adjust the dose of the stimulation regimen according to the ovarian response. Percentile curves have been used successfully in other fields of gynaecology–obstetrics, such as fetal ultrasonography, and can be useful in COS monitoring. Our high responders (E2 >P90) had better IVF outcomes than normal and poor responders in all age groups. More studies with larger sample sizes using the same protocols have to be performed for further confirmation of our results.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The authors kindly acknowledge the assistance and advice of Dr Yukio Sonoda.


    Notes
 
1 To whom correspondence should be addressed. E-mail: tour-med{at}wanadoo.fr Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Asch, R.H., Li, H.P., Balmaceda, J.P., Weckestein, L.N. and Stone, S.C. (1991) Severe ovarian hyperstimulation syndrome in assisted reproductive technologies: definition of high risk groups. Hum. Reprod., 6, 1395–1399.[Abstract]

Bladford, A.J., Najmabadi, S. and Paulson, R.J. (1997) Ultrastructural characteristics of the luteal phase endometrium in donors undergoing controlled ovarian hyperstimulation. Fertil. Steril., 67, 625–630.[ISI][Medline]

Brinsden, P., Akagbosu, F., Gibbons, L.M., Lancaster, S., Gourdon, D., Pharm, D., Engrand, P. and Loumaye, E. (2000) A comparison of the efficacy and tolerability of two recombinant human follicle-stimulating hormone preparations in patients undergoing in vitro fertilization–embryo transfer. Fertil. Steril., 73, 114–116.[ISI][Medline]

Chenette, P.E., Sauer, M.V. and Paulson, R.J. (1990) Very high serum E2 levels are not detrimental to clinical outcome if in vitro fertilization. Fertil. Steril., 54, 858–863.[ISI][Medline]

Forman, R., Fries, N., Testart, J., Belaisch-Allart, J., Hazout, A. and Frydman, R. (1988) Evidence for an adverse effect of elevated serum E2 concentration on embryo implantation. Fertil. Steril., 49, 118–122.[ISI][Medline]

Garcia, J.E., Acosta, A.A., Hsiu, J.G. and Jones, H.W. (1984) Advanced endometrial maturation after ovulation induction with human menopausal gonadotrophin/human chorionic gonadotrophin for in vitro fertilization. Fertil. Steril., 41, 31–37.[ISI][Medline]

Gidley-Baird, A.A., O’Neil, C., Sinosich, M.J., Porter, R.N., Pike, I.L. and Saunders, D.M. (1986) Failure of implantation in human in vitro fertilization and embryo transfer patients: the effects of altered progesterone/estrogen rations in humans and mice. Fertil. Steril., 45, 69–75.[ISI][Medline]

Hadi, F.H., Chantler, E., Anderson, E., Nicholson, R., McClelland, R.A. and Seif, M.W. (1994) Ovulation induction and endometrial steroid receptors. Hum. Reprod., 9, 2405–2410.[Abstract]

Harlin, J., Aanesen, A., Csemiczky, G., Wramsby, H. and Fried, G. (2002) Delivery rates following IVF treatment, using two recombinant FSH preparations for ovarian stimulation. Hum. Reprod., 17, 304–309.[Abstract/Free Full Text]

Jaffe, S.B., Jaffe, L.H. and Jewelewicz, R. (1993) Incidence of severe ovarian hyperstimulation syndrome with extremely elevated serum estrogen levels. Gynecol. Obstet. Invest., 35, 222–227.[ISI][Medline]

Khalaf, Y., Taylor, A. and Braude, P. (2000) Low serum E2 concentrations after five days of controlled ovarian hyperstimulation for in vitro fertilization are associated with poor outcome. Fertil. Steril., 74, 63–66.[ISI][Medline]

Leroy, B. and Lefort, F. (1971) A propos du poids et de la taille des nouveaus-nés à la naissance. Rev. Fr. Gynecol., 66, 391–395.

Levi, L.J., Drews, M.R., Bergh, P.A., Miller, B.T. and Scott, R.T. (2001) Controlled ovarian hyperstimulation does not adversely affect endometrial receptivity in in-vitro fertilization cycles. Fertil. Steril., 76, 670–674.[ISI][Medline]

Loumaye, E., Engrand, P., Howles, C.M. and O’Dea, L. (1997) Assessment of the role of serum luteinizing hormone and E2 response to follicle-stimulating hormone on in vitro fertilization treatment outcome. Fertil. Steril., 67, 889–899.[ISI][Medline]

Lubchenco, L.O., Hansman, C., Dressler, M. and Boyd, E. (1963) Intrauterine growth as estimated from liveborn birthweight data at 24 to 42 weeks of gestation. Pediatrics, 32, 793–800.[Abstract]

McDonough, P. (1999) low and high responders—at what level of serum E2 do things start to get fuzzy? Fertil. Steril., 71, 584–586.[ISI]

Morris, R.S., Paulson, R.J., Sauer, M.V. and Lobo, R.A. (1995) Predictive value of serum E2 concentrations and oocyte number in severe hyperstimulation syndrome. Hum. Reprod., 10, 811–814.[Abstract]

Pellicer, A., Valbueña, D., Cano, F., Remohí, J. and Simón, C. (1996) Lower implantation rates in high responders: evidence for an altered endocrine milieu during the preimplantation period. Fertil. Steril., 65, 1190–1195.[ISI][Medline]

Phelps, J.Y., Levine, A.S., Hockman, T.N., Zacur, H.A., Wallach, E.E. and Hinton, E.L. (1998) Day 4 E2 levels predict pregnancy success in women undergoing controlled ovarian hyperstimulation for IVF. Fertil. Steril., 69, 1015–1019.[ISI][Medline]

Sharara, F.I. and McClamrock, H.D. (1999) High E2 levels and high oocyte yield are not detrimental to in vitro fertilization outcome. Fertil. Steril., 72, 401–405.[ISI][Medline]

Simón, C., Garcia Velasco, J.J., Valbueña, D., Peinado, J.A., Moreno, C., Remohí, J. and Pellicer, A. (1998) Increasing uterine receptivity by decreasing E2 levels during the preimplantation period in high responders with the use of a follicle-stimulating hormone step-down regimen. Fertil. Steril., 70, 234–239.[ISI][Medline]

Tarín, J.J., Sampaio, M.C., Calatayud, C., Castellvi, R.M., Bonilla-Musoles, F. and Pellicer, A. (1992) Relativity of the concept ‘high responder to gonadotrophins’. Hum. Reprod., 7, 19–22.[Abstract]

Testart, J., Belaisch-Allart, J. and Frydman, R. (1986) Relationship between embryo transfer results and ovarian response and in vitro fertilization rates: analysis of 186 human pregnancies. Fertil. Steril., 45, 237–243.[ISI][Medline]

Valbueña., D., Martin, J., de Pablo, J.L., Remohí, J., Pellicer, A. and Simón, C. (2001) Increasing levels of estradiol are deleterious to embryonic implantation because they directly affect the embryo. Fertil. Steril., 76, 962–968.[ISI][Medline]

Submitted on December 5, 2001; resubmitted on March 20, 2002; accepted on July 17, 2002.