Reproductive BioMedicine Online (2014) 29, 299–304
w w w. s c i e n c e d i r e c t . c o m w w w. r b m o n l i n e . c o m
ARTICLE
GnRH agonist and GnRH antagonist protocols: comparison of outcomes among goodprognosis patients using national surveillance data Daniel Grow a, Jennifer F Kawwass b,c,*, Aniket D Kulkarni b, Tonji Durant b, Denise J Jamieson b,c, Maurizio Macaluso b a
Department of Obstetrics and Gynecology, Baystate Medical Center, Springfield, MA, USA; b Division of Reproductive Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA; c Division of Reproductive Endocrinology and Infertility, Department of Gynecology and Obstetrics, Emory University School of Medicine, Atlanta, GA, USA * Corresponding author.
E-mail address: [email protected] (JF Kawwass). Daniel Grow is a reproductive endocrinologist who uses safe, efficient assisted reproductive technology and promotes the added safety of single embryo transfer to maximize the concept of ‘one healthy baby at a time’. His laboratory focuses efforts on minimizing gamete and embryo stress through thoughtful techniques. His research focuses on the use of antral follicle counts, gentle embryo transfers, sperm-egg interaction, and reproductive surgery. He serves as Professor and Deputy Chair for the Department of Obstetrics and Gynecology at Tufts University School of Medicine, Baystate Health, where he practices, teaches residents, and facilitates the efficient delivery of women’s health services.
Abstract Implantation and live birth rates resulting from IVF cycles using gonadotropin-releasing hormone (GnRH) agonist and (GnRH) antagonist IVF protocols were compared among good-prognosis patients using the Centers for Disease Control and Prevention’s National Assisted Reproductive Technology Surveillance System 2009–2010 data (n = 203,302 fresh, autologous cycles). Bivariable and multivariable analyses were conducted between cycles to compare outcomes. Cycles were restricted as follows: age younger than 35 years, maximum FSH less than 10 mIU/mL, first assisted reproduction technology cycle and FSH dose less than 3601 IU. A subgroup analysis including only elective single embryo transfer was also carried out. Among good-prognosis patients, the GnRHagonist protocol was associated with a lower risk of cancellation before retrieval (4.3 versus 5.2%; P < 0.05) or transfer (5.5 versus 6.8%; P < 0.05), and a higher live birth rate per transfer (adjusted odds ratio [OR] 1.13, confidence interval [CI] 1.03 to 1.25) than the GnRH-antagonist group. Among the elective single embryo transfer group, the GnRH-agonist protocol was associated with a higher implantation rate (adjusted odds ratio [OR] 1.36, CI 1.08 to 1.73) and a higher live birth rate (adjusted OR 1.33, CI 1.07 to 1.66) compared with the GnRH-antagonist protocol. The GnRH-antagonist group had lower rates of ovarian hyperstimulation
http://dx.doi.org/10.1016/j.rbmo.2014.05.007 1472-6483/© 2014 Reproductive Healthcare Ltd. Published by Elsevier Ltd. All rights reserved.
300
D Grow et al.
syndrome. Among good-prognosis patients, agonist protocols decreased cancellation risk and increased odds of implantation and live birth. Antagonist protocols may confer decreased risk of hyperstimulation. © 2014 Reproductive Healthcare Ltd. Published by Elsevier Ltd. All rights reserved. KEYWORDS: agonist, antagonist, implantation rate, live birth rate, good prognosis
Introduction In assisted reproductive technology, GnRH analogues, agonists and antagonists, are widely used to prevent the endogenous LH surge and allow for well-timed oocyte retrieval (Al-Inany et al., 2006, 2011). Although GnRH agonists have been used for several decades (Cetel et al., 1983), antagonists have recently gained increased popularity, as their use provides more immediate pituitary suppression without an initial flare (shorter stimulation duration and fewer injections) and may confer decreased risk of ovarian hyperstimulation syndrome (OHSS) (Engmann et al., 2008). The effect of each protocol on implantation and live birth rates, however, remains controversial despite numerous studies, including randomized controlled trials (RCTs) involving heterogeneous, small study populations (Aboulghar et al., 2004; Albano et al., 2000; Barmat et al., 2005; Bodri et al., 2010; Marci et al., 2013). A 2006 Cochrane review including 27 RCTs found antagonists to be associated with a decreased live birth rate (odds ratio [OR] 0.82, 95% confidence interval [CI] 0.68 to 0.97 (Al-Inany et al., 2006). A 2011 Cochrane review including 45 RCTs, however, found no statistically significant difference in live birth rates between the agonist and antagonist groups (Al-Inany et al., 2011). The 2009–2010 Centers for Disease Control and Prevention National Assisted Reproductive Technology Surveillance System (NASS), was used to compare cycle cancellation rates, implantation rates, and live birth rates per transfer between cycles using a GnRH agonist and a GnRH antagonist protocol in good-prognosis patients (defined as women younger than 35 years, with a maximum FSH less than 10 mIU/mL, undergoing their first assisted reproductive technology cycle, with a total FSH dose between 1200 and 3600 IU). An analysis including only elective single embryo transfers transfers (SET), in which only one embryo was transferred and at least one supernumerary embryo was cryopreserved, was also carried out to compare outcomes within an even more homogenous good-prognosis group.
Methods Study population and participants Data on 203,302 fresh, autologous IVF cycles initiated during 2009 and 2010 were taken from NASS (Sunderam et al., 2012), which receives mandatory cycle and demographic data on over 95% of all IVF cycles carried out in the USA. Collected data include patient demographics, medical and obstetric history, infertility diagnoses, detailed parameters of each assisted reproductive technology treatment cycle and, if applicable, the resultant pregnancy outcome. Additionally, each year, 7–10% of reporting clinics are randomly selected for data validation, with slightly greater
selection chances for larger clinics and clinics with a low cycle cancellation rate. During validation, a sample of assisted reproductive technology data reported by the clinics is compared with information recorded in medical records, and discrepancy rates are calculated. Overall, discrepancy rates for the variables evaluated in the present study were less than 5%, except for the diagnosis of infertility, which had higher discrepancy rates (up to 18%), mostly because of reporting of ‘other’ or ‘unexplained’ infertility instead of a specific cause. To reduce selection bias, we restricted our analysis to cycles of women using either a GnRH antagonist or long GnRH agonist protocol, with a good prognostic profile, defined as female age younger than 35 years, maximum baseline serum FSH less than 10 mIU/mL, no prior history of IVF, and total FSH dose between 1200 and 3600 IU. No clomiphene citrate cycles were included in the analysis. Women who underwent antagonist stimulation and received an agonist trigger were excluded from the comparison. For women with body mass index less than 15 or greater than 50 kg/m2, the values were set to missing. For the final model calculating the adjusted odds ratios, cycles with missing data for these criteria were excluded from the analysis. Ovarian hyperstimulation can be reported as ‘moderate,’ characterized by ascites, enlarged ovarian volume, and abdominal distension accompanied by nausea, vomiting, diarrhoea, or both, or as ‘severe’, characterized by moderate characteristics and also haemoconcentration, laboratory abnormalities, or clinical evidence of ascites, hydrothorax, or dyspnoea.
Statistical analysis Bivariable analyses were carried out to explore the relationship between stimulation protocol (GnRH agonist versus GnRH antagonist) and patient and cycle characteristics, including patient age, race, ethnicity, infertility diagnosis, obstetric history, and characteristics of the IVF cycle. The Pearson chisquared test for categorical variables and t-test for numeric variables were used to assess differences between agonist and antagonist cycles. Mixed-effect logistic models with clinic as the random effect were used to estimate crude odds ratios and adjusted odds ratios for implantation of at least one embryo and live birth per embryo transfer, comparing the two protocols. We chose to report adjusted odds ratios comparable with the 2006 and 2011 Cochrane Reviews (Al-Inany et al., 2006). Implantation rate was defined as the percentage of embryo transfer cycles resulting in implantation of at least one embryo. Race and ethnicity were excluded from the final model, owing to a large percentage of missing values within this field. To further reduce confounding and selection bias, a secondary analysis, including only cycles involving an elective SET (n = 2029 cycles) was carried out, as these cycles avoid
Agonist v Antagonist protocol: outcomes comparison among good prognosis patients comparisons between cycles in which different numbers of embryos were transferred and are likely to include goodprognosis patients with good-quality embryos. Elective SET was defined as the transfer of a single embryo with cryopreservation of at least one supernumerary embryo. Implantation of at least one embryo and live birth per embryo transfer were compared using the same multivariable model mentioned above to calculate odds ratios and adjusted odds ratios.
Ethical approval This study was approved by the Institutional Review Board (protocol number 2238.0 on 4 December 2013) of the Centers for Disease Control and Prevention; a waiver of informed consent was obtained.
Results Of the 203,302 fresh, autologous IVF cycles carried out in 2009 and 2010, 23,113 cycles met our good-prognosis inclusion criteria (Figure 1). A total of 16,001 first-time IVF cycles used a GnRH-gonist protocol and 7112 cycles used a GnRHantagonist protocol. The women in the GnRH agonist and antagonist groups were similar in age, had prior live births, prior pregnancy losses, and similar body mass index (Table 1). Infertility diagnosis differed slightly between groups. Mean total FSH dose was greater in the antagonist group. Compared with the GnRH antagonist group, the GnRH agonist group had a lower risk of cancellation before
retrieval (4.3 versus 5.2%; P < 0.05), and between retrieval and transfer (5.5 versus 6.8%; P < 0.05) (Table 1). The agonist group had 691 cancelled cycles (68% of which were due to low ovarian response, 17% of which were due to high ovarian response, and 14% were ‘withdrawal only for personal reasons’). The antagonist group had 367 cancelled cycles (76% due to low ovarian response, 8% due to high ovarian response, 14% ‘withdrawal only for personal reasons’, and the remaining 2% included inadequate endometrial response, concurrent illness, and inability to obtain a semen specimen). The mean number of oocytes retrieved was comparable for the two groups; 15.8 (SD 7.9) compared with 16.0 (SD 8.5) in the agonist and antagonist groups, respectively. More elective SET were carried out in the antagonist group (12.9 versus 10.9%; P < 0.05). The rate of moderate ovarian hyperstimulation was greater in the agonist group (2.1 versus 1.7%; P < 0.05), similar rates of severe hyperstimulation were reported between the two groups (0.7 versus 0.6%). Among those cycles complicated by hyperstimulation, 16.2% required hospitalization in the agonist group compared with 19.8% in the antagonist group. The implantation (57.6 versus 54.1%; P < 0.05) and live birth rates (52.6 versus 49.2%; P < 0.05) per transfer were significantly higher in the agonist compared with the antagonist group. The final mixed-effect multivariable logistic model (n = 18,021 cycles) adjusted for patient age, gravidity, assisted reproductive technology indication, FSH dose, body mass index, and number of embryos transferred. The multivariable model showed higher implantation (adjusted OR 1.14, CI 1.03 to 1.26) and live birth per transfer (adjusted OR 1.13, CI 1.03 to 1.25) in the agonist compared with antagonist group (Table 2). The results were similar for the subgroup analysis of the 2029 cycles involving elective SET; the implantation rate (adjusted OR 1.36, CI 1.08 to 1.73) and live birth
Fresh cycles, patient stimulation = 203,302
84,128
53,359
Agonist + FSH (n = 16,001)
691 cancellations
Agonist + FSH (n = 15,310)
Figure 1
Selection of good-prognosis patients.
301
Age ≥ 35 years = 119,174 Prior assisted reproduction technology cycles = 30,769 Missing baseline FSH = 15,546 Baseline FSH ≥ 10 = 4869
32,944
FSH dose < 1200 or > 3600 or missing = 8396
24,548
Other protocols = 1435
23,113
Antagonist + FSH (n = 7112)
367 cancellations
Antagonist + FSH (n = 6745)
302
D Grow et al. Table 1 Population characteristics: GnRH agonist protocols, GnRH antagonist protocols, reported per autologous IVF cycle, 2009–2010. Characteristics
Age (years, mean ± SD) Prior pregnancy (gravidity) (%) Prior live birth (full-term or preterm) (%) Prior pregnancy loss (spontaneous abortions) (%) Infertility diagnosis (%) Tubal factor Ovulation disorders Endometriosis Male factor Other factors Unexplained Multiple factors (female) Multiple factors (male and female) FSH dose (mean ± SD) Body mass index (mean ± SD) Race White Others Missing Ethnicity Hispanic or Latino Non-Hispanic or Latino Missing Cancellation before retrieval (%) Retrievals not leading to transfer (%) Outcomes Number of cycles reaching embryo transfer Number of oocytes retrieved (mean ± SD)b Total number of embryos transferred (%) 1 2 ≥3 Cycles with implantation (%)c Live birthd for all cycles (%) Elective SET cyclesd n (%) Live birthd for eSET cycles (%) Moderate ovarian hyperstimulation n (%) Severe ovarian hyperstimulation n (%)
Stimulation protocol Agonist + FSH (n = 16,001)
Antagonist + FSH (n = 7112)
30.4 ± 2.7 36.0 16.2
30.5 ± 2.8 37.4a 17.2
19.6
19.9
9.4 11.8 5.1 28.6 7.7 15.3 6.2 15.9 2185.9 ± 625.8 25.4 ± 5.7 (n = 13,913)
7.8a 12.2 4.5 26.8 9.8 14.6 7.5 16.8 2230.0 ± 636.2a 25.3 ± 5.6 (n = 6189)
52.0 13.3 34.6
50.0a 12.9 37.0
6.4 73.5 20.1 4.32 5.5
7.7a 66.0 26.4 5.16a 6.8a
14,464
6289
15.8 ± 7.9
16.0 ± 8.5
15.40 77.05 7.56 8331 (57.6) 52.6 1489 (10.85) 53.93 342 (2.1) 118 (0.7)
17.43a 74.70 7.87 3403 (54.1)a 49.2a 767, 12.87a 46.15a 120, 1.7a 42 (0.)6
P < 0.05. Calculated per transfer where at least one embryo was transferred (denominator equivalent to “number of cycles reaching embryo transfer”). c “Cycles with implantation” – implantation calculated per cycle among cycles where at least one embryo was transferred. d Calculated per transfer where at least one embryo transferred and at least one embryo cryopreserved (denominators are 13,726 for agonist and 5960 for antagonist). a
b
Agonist v Antagonist protocol: outcomes comparison among good prognosis patients
303
Table 2 Odds of implantation and live birth per transfer among good-prognosis patients using agonist and antagonist protocols among completed fresh, autologous IVF cycles between 2009 and 2010.
All good prognosis completed cyclesa Implantation Live birth All cycles with elective SET Implantation Live birth
Odds ratio (confidence interval)
Adjusted odds ratio (confidence interval)
1.15 (1.04 to 1.27) 1.14 (1.04 to 1.26)
1.14 (1.03 to 1.26) 1.13 (1.03 to 1.25)
1.41 (1.13 to 1.76) 1.37 (1.11 to 1.68)
1.36 (1.08 to 1.73) 1.33 (1.07 to 1.66)
eSET, elective single embryo transfer; n = 2256 cycles for bivariable and 2029 cycles for multivariable analyses. Adjusted for patient age, gravidity, assisted reproductive technology indication, FSH dose and body mass index. a n = 20,753 cycles for bivariable analyses; 18, 021 cycles for multivariable analyses. ‘Good prognosis’ defined as woman’s age younger than 35 years, maximum baseline serum FSH less than 10 mIU/mL, no prior history of IVF, total FSH dose between 1200 and 3600 IU, and body mass index between 18 and 45 kg/m2. Adjusted for patient age, gravidity, ART indication, FSH dose, body mass index, number of embryos transferred.
rate (adjusted OR 1.33, CI 1.07 to 1.66) were higher in the agonist compared with the antagonist group.
Discussion Our study suggests that, among women who have a good prognosis as defined by age, fertility treatment history, ovarian reserve testing, and stimulation medication requirements, long GnRH-agonist protocols are associated with a lower cancellation rate and higher implantation and live birth compared with GnRH-antagonist protocols. These findings remained significant among the subgroup of good-prognosis patients undergoing elective SET. The antagonist protocols conferred a decreased risk of moderate but not severe ovarian hyperstimulation. The increased likelihood of cancellation before retrieval and before transfer among antagonist cycles may reflect an increased likelihood of ovarian cyst formation before stimulation medication start, over- or under-response to stimulation, or premature ovulation. The lower risk of hyperstimulation associated with antagonist protocols may be underestimated because those women for whom an antagonist protocol was selected may have been assigned to the antagonist group owing to their suspected risk of OHSS. Additionally, the differentiation between moderate and severe OHSS is a continuum, and classification may vary between clinics. Although the large sample size makes it statistically significant, the difference in mean total FSH dose between the agonist and antagonist groups was only 45 IU, and is likely not to be clinically significant. The increased likelihood of elective SET among antagonist cycles may reflect an attempt to decrease OHSS risk or clinic transfer tendency; nonetheless, implantation rate and live birth rate remained lower within the elective SET group for those who underwent an antagonist protocol after adjusting for patient age, gravidity, assisted reproductive technology indication, FSH dose, and body mass index. Previous studies including retrospective reviews and randomized controlled trials have reported conflicting results on
the effect of protocol choice on implantation and live birth (Al-Inany et al., 2006, 2011; Griesinger et al., 2005). A large number of multiple target tissues are affected by GnRH analogues, including the pituitary, ovary and endometrium. As a result, agonist and antagonists may alter the ovarian response, endometrial environment, and hormonal milieu at time of embryo transfer differently (Hsueh and Jones, 1981). The effect of protocol choice on implantation rate may reflect the possible deleterious effect of GnRH antagonists on the endometrium, and the possible increased competency oocytes exposed to GnRH agonists (Check et al., 2011; Ortmann et al., 2001; Schachter et al., 2008). The different effect on IVF outcomes between the two protocols is most likely to be subtle, and therefore difficult to capture without a large sampling of patients. Our study is strengthened by the large sample size and the ability to conduct subgroup analysis of good-prognosis patients undergoing elective SET. Previous RCTs were smaller (ranging from 20–730 participants), and either included a more diverse patient population or focused specifically on women with diminished ovarian reserve or polycystic ovary syndrome (Al-Inany et al., 2006, 2011). Additionally, our study reflects more current assisted reproductive technique, whereas previous studies span a time period during which treatment protocols and effectiveness may have evolved. Study limitations include dependence on data entered by individual clinics, inability to control for nuanced differences between stimulation protocols among different clinicians, such as oral contraceptive pill pretreatment, the subjective definition of ovarian hyperstimulation, and selection bias in protocol selection that accompanies a restrpospective analysis. We attempted to minimize this by selecting good-prognosis patients undergoing elective SET. It seems that, among good-prognosis patients, GnRHagonist protocols may confer decreased risk of cancellation and an increased odds of implantation and live birth; however, the GnRH-antagonist protocols may be associated with decreased risk of hyperstimulation. Protocol choice is a nuanced decision that incorporates patient reproductive treatment history, suspected ovarian response to medication, and
304 clinician and patient preference. The results of a single study may contribute additional evidence to the decision-making process, but should not drastically change clinical practice. Ideally, a large prospective randomized trial, although costly, would be the optimal method by which to gain further insight into the effect of protocol choice on pregnancy rates and outcomes.
Acknowledgements Dr. Donna Session participated in the review and critique of the manuscript; she did not receive any financial compensation.
References Aboulghar, M.A., Mansour, R.T., Serour, G.I., Al-Inany, H.G., Amin, Y.M., Aboulghar, M.M., 2004. Increasing the dose of human menopausal gonadotrophins on day of GnRH antagonist administration: randomized controlled trial. Reprod. Biomed. Online 8, 524–527. Al-Inany, H.G., Abou-Setta, A.M., Aboulghar, M., 2006. Gonadotrophinreleasing hormone antagonists for assisted conception. Cochrane Database Syst. Rev. (3), CD001750. Al-Inany, H.G., Youssef, M.A., Aboulghar, M., Broekmans, F., Sterrenburg, M., Smit, J., Abou-Setta, A.M., 2011. Gonadotrophinreleasing hormone antagonists for assisted reproductive technology. Cochrane Database Syst. Rev. (5), CD001750. Albano, C., Felberbaum, R.E., Smitz, J., Riethmuller-Winzen, H., Engel, J., Diedrich, K., Devroey, P., 2000. Ovarian stimulation with HMG: results of a prospective randomized phase III European study comparing the luteinizing hormone-releasing hormone (LHRH)-antagonist cetrorelix and the LHRH-agonist buserelin. European Cetrorelix Study Group. Hum. Reprod. 15, 526–531. Barmat, L.I., Chantilis, S.J., Hurst, B.S., Dickey, R.P., 2005. A randomized prospective trial comparing gonadotropin-releasing hormone (GnRH) antagonist/recombinant follicle-stimulating hormone (rFSH) versus GnRH-agonist/rFSH in women pretreated with oral contraceptives before in vitro fertilization. Fertil. Steril. 83, 321–330. Bodri, D., Sunkara, S.K., Coomarasamy, A., 2010. Gonadotropinreleasing hormone agonists versus antagonists for controlled ovarian hyperstimulation in oocyte donors: a systematic review and meta-analysis. Fertil. Steril. 95, 164–169.
D Grow et al. Cetel, N.S., Rivier, J., Vale, W., Yen, S.S., 1983. The dynamics of gonadotropin inhibition in women induced by an antagonistic analog of gonadotropin-releasing hormone. J. Clin. Endocrinol. Metab. 57, 62–65. Check, J.H., Cohen, R., Amui, J., Choe, J.K., Brasile, D., 2011. Evidence that the main adverse effect of ganirelix on pregnancy and implantation rates is on the embryo rather than the endometrium. Clin. Exp. Obstet. Gynecol. 38, 326–327. Engmann, L., Diluigi, A., Schmidt, D., Nulsen, J., Maier, D., Benadiva, C., 2008. The use of gonadotropin-releasing hormone (GnRH) agonist to induce oocyte maturation after cotreatment with GnRH antagonist in high-risk patients undergoing in vitro fertilization prevents the risk of ovarian hyperstimulation syndrome: a prospective randomized controlled study. Fertil. Steril. 89, 84–91. Griesinger, G., Felberbaum, R., Diedrich, K., 2005. GnRH antagonists in ovarian stimulation: a treatment regimen of clinicians’ second choice? Data from the German national IVF registry. Hum. Reprod. 20, 2373–2375. Hsueh, A.J., Jones, P.B., 1981. Extrapituitary actions of gonadotropinreleasing hormone. Endocr. Rev. 2, 437–461. Marci, R., Graziano, A., Lo Monte, G., Piva, I., Soave, I., Marra, E., Lisi, F., Moscarini, M., Caserta, D., 2013. GnRH antagonists in assisted reproductive techniques: a review on the Italian experience. Eur. Rev. Med. Pharmacol. Sci. 17, 853–873. Ortmann, O., Weiss, J.M., Diedrich, K., 2001. Embryo implantation and GnRH antagonists: ovarian actions of GnRH antagonists. Hum. Reprod. 16, 608–611. Schachter, M., Friedler, S., Ron-El, R., Zimmerman, A.L., Strassburger, D., Bern, O., Raziel, A., 2008. Can pregnancy rate be improved in gonadotropin-releasing hormone (GnRH) antagonist cycles by administering GnRH agonist before oocyte retrieval? A prospective, randomized study. Fertil. Steril. 90, 1087–1093. Sunderam, S., Kissin, D.M., Flowers, L., Anderson, J.E., Folger, S.G., Jamieson, D.J., Barfield, W.D., 2012. Assisted reproductive technology surveillance – United States, 2009. MMWR Surveill. Summ. 61, 1–23.
Authors’ roles: DG, AK, TD, JFK and MM were involved in the study design, analysis, manuscript drafting, and critical discussion.
Declaration: The authors report no financial or commercial conflicts of interest.
Received 28 January 2014; refereed 15 May 2014; accepted 21 May 2014.
w w w. s c i e n c e d i r e c t . c o m w w w. r b m o n l i n e . c o m
ARTICLE
GnRH agonist and GnRH antagonist protocols: comparison of outcomes among goodprognosis patients using national surveillance data Daniel Grow a, Jennifer F Kawwass b,c,*, Aniket D Kulkarni b, Tonji Durant b, Denise J Jamieson b,c, Maurizio Macaluso b a
Department of Obstetrics and Gynecology, Baystate Medical Center, Springfield, MA, USA; b Division of Reproductive Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA; c Division of Reproductive Endocrinology and Infertility, Department of Gynecology and Obstetrics, Emory University School of Medicine, Atlanta, GA, USA * Corresponding author.
E-mail address: [email protected] (JF Kawwass). Daniel Grow is a reproductive endocrinologist who uses safe, efficient assisted reproductive technology and promotes the added safety of single embryo transfer to maximize the concept of ‘one healthy baby at a time’. His laboratory focuses efforts on minimizing gamete and embryo stress through thoughtful techniques. His research focuses on the use of antral follicle counts, gentle embryo transfers, sperm-egg interaction, and reproductive surgery. He serves as Professor and Deputy Chair for the Department of Obstetrics and Gynecology at Tufts University School of Medicine, Baystate Health, where he practices, teaches residents, and facilitates the efficient delivery of women’s health services.
Abstract Implantation and live birth rates resulting from IVF cycles using gonadotropin-releasing hormone (GnRH) agonist and (GnRH) antagonist IVF protocols were compared among good-prognosis patients using the Centers for Disease Control and Prevention’s National Assisted Reproductive Technology Surveillance System 2009–2010 data (n = 203,302 fresh, autologous cycles). Bivariable and multivariable analyses were conducted between cycles to compare outcomes. Cycles were restricted as follows: age younger than 35 years, maximum FSH less than 10 mIU/mL, first assisted reproduction technology cycle and FSH dose less than 3601 IU. A subgroup analysis including only elective single embryo transfer was also carried out. Among good-prognosis patients, the GnRHagonist protocol was associated with a lower risk of cancellation before retrieval (4.3 versus 5.2%; P < 0.05) or transfer (5.5 versus 6.8%; P < 0.05), and a higher live birth rate per transfer (adjusted odds ratio [OR] 1.13, confidence interval [CI] 1.03 to 1.25) than the GnRH-antagonist group. Among the elective single embryo transfer group, the GnRH-agonist protocol was associated with a higher implantation rate (adjusted odds ratio [OR] 1.36, CI 1.08 to 1.73) and a higher live birth rate (adjusted OR 1.33, CI 1.07 to 1.66) compared with the GnRH-antagonist protocol. The GnRH-antagonist group had lower rates of ovarian hyperstimulation
http://dx.doi.org/10.1016/j.rbmo.2014.05.007 1472-6483/© 2014 Reproductive Healthcare Ltd. Published by Elsevier Ltd. All rights reserved.
300
D Grow et al.
syndrome. Among good-prognosis patients, agonist protocols decreased cancellation risk and increased odds of implantation and live birth. Antagonist protocols may confer decreased risk of hyperstimulation. © 2014 Reproductive Healthcare Ltd. Published by Elsevier Ltd. All rights reserved. KEYWORDS: agonist, antagonist, implantation rate, live birth rate, good prognosis
Introduction In assisted reproductive technology, GnRH analogues, agonists and antagonists, are widely used to prevent the endogenous LH surge and allow for well-timed oocyte retrieval (Al-Inany et al., 2006, 2011). Although GnRH agonists have been used for several decades (Cetel et al., 1983), antagonists have recently gained increased popularity, as their use provides more immediate pituitary suppression without an initial flare (shorter stimulation duration and fewer injections) and may confer decreased risk of ovarian hyperstimulation syndrome (OHSS) (Engmann et al., 2008). The effect of each protocol on implantation and live birth rates, however, remains controversial despite numerous studies, including randomized controlled trials (RCTs) involving heterogeneous, small study populations (Aboulghar et al., 2004; Albano et al., 2000; Barmat et al., 2005; Bodri et al., 2010; Marci et al., 2013). A 2006 Cochrane review including 27 RCTs found antagonists to be associated with a decreased live birth rate (odds ratio [OR] 0.82, 95% confidence interval [CI] 0.68 to 0.97 (Al-Inany et al., 2006). A 2011 Cochrane review including 45 RCTs, however, found no statistically significant difference in live birth rates between the agonist and antagonist groups (Al-Inany et al., 2011). The 2009–2010 Centers for Disease Control and Prevention National Assisted Reproductive Technology Surveillance System (NASS), was used to compare cycle cancellation rates, implantation rates, and live birth rates per transfer between cycles using a GnRH agonist and a GnRH antagonist protocol in good-prognosis patients (defined as women younger than 35 years, with a maximum FSH less than 10 mIU/mL, undergoing their first assisted reproductive technology cycle, with a total FSH dose between 1200 and 3600 IU). An analysis including only elective single embryo transfers transfers (SET), in which only one embryo was transferred and at least one supernumerary embryo was cryopreserved, was also carried out to compare outcomes within an even more homogenous good-prognosis group.
Methods Study population and participants Data on 203,302 fresh, autologous IVF cycles initiated during 2009 and 2010 were taken from NASS (Sunderam et al., 2012), which receives mandatory cycle and demographic data on over 95% of all IVF cycles carried out in the USA. Collected data include patient demographics, medical and obstetric history, infertility diagnoses, detailed parameters of each assisted reproductive technology treatment cycle and, if applicable, the resultant pregnancy outcome. Additionally, each year, 7–10% of reporting clinics are randomly selected for data validation, with slightly greater
selection chances for larger clinics and clinics with a low cycle cancellation rate. During validation, a sample of assisted reproductive technology data reported by the clinics is compared with information recorded in medical records, and discrepancy rates are calculated. Overall, discrepancy rates for the variables evaluated in the present study were less than 5%, except for the diagnosis of infertility, which had higher discrepancy rates (up to 18%), mostly because of reporting of ‘other’ or ‘unexplained’ infertility instead of a specific cause. To reduce selection bias, we restricted our analysis to cycles of women using either a GnRH antagonist or long GnRH agonist protocol, with a good prognostic profile, defined as female age younger than 35 years, maximum baseline serum FSH less than 10 mIU/mL, no prior history of IVF, and total FSH dose between 1200 and 3600 IU. No clomiphene citrate cycles were included in the analysis. Women who underwent antagonist stimulation and received an agonist trigger were excluded from the comparison. For women with body mass index less than 15 or greater than 50 kg/m2, the values were set to missing. For the final model calculating the adjusted odds ratios, cycles with missing data for these criteria were excluded from the analysis. Ovarian hyperstimulation can be reported as ‘moderate,’ characterized by ascites, enlarged ovarian volume, and abdominal distension accompanied by nausea, vomiting, diarrhoea, or both, or as ‘severe’, characterized by moderate characteristics and also haemoconcentration, laboratory abnormalities, or clinical evidence of ascites, hydrothorax, or dyspnoea.
Statistical analysis Bivariable analyses were carried out to explore the relationship between stimulation protocol (GnRH agonist versus GnRH antagonist) and patient and cycle characteristics, including patient age, race, ethnicity, infertility diagnosis, obstetric history, and characteristics of the IVF cycle. The Pearson chisquared test for categorical variables and t-test for numeric variables were used to assess differences between agonist and antagonist cycles. Mixed-effect logistic models with clinic as the random effect were used to estimate crude odds ratios and adjusted odds ratios for implantation of at least one embryo and live birth per embryo transfer, comparing the two protocols. We chose to report adjusted odds ratios comparable with the 2006 and 2011 Cochrane Reviews (Al-Inany et al., 2006). Implantation rate was defined as the percentage of embryo transfer cycles resulting in implantation of at least one embryo. Race and ethnicity were excluded from the final model, owing to a large percentage of missing values within this field. To further reduce confounding and selection bias, a secondary analysis, including only cycles involving an elective SET (n = 2029 cycles) was carried out, as these cycles avoid
Agonist v Antagonist protocol: outcomes comparison among good prognosis patients comparisons between cycles in which different numbers of embryos were transferred and are likely to include goodprognosis patients with good-quality embryos. Elective SET was defined as the transfer of a single embryo with cryopreservation of at least one supernumerary embryo. Implantation of at least one embryo and live birth per embryo transfer were compared using the same multivariable model mentioned above to calculate odds ratios and adjusted odds ratios.
Ethical approval This study was approved by the Institutional Review Board (protocol number 2238.0 on 4 December 2013) of the Centers for Disease Control and Prevention; a waiver of informed consent was obtained.
Results Of the 203,302 fresh, autologous IVF cycles carried out in 2009 and 2010, 23,113 cycles met our good-prognosis inclusion criteria (Figure 1). A total of 16,001 first-time IVF cycles used a GnRH-gonist protocol and 7112 cycles used a GnRHantagonist protocol. The women in the GnRH agonist and antagonist groups were similar in age, had prior live births, prior pregnancy losses, and similar body mass index (Table 1). Infertility diagnosis differed slightly between groups. Mean total FSH dose was greater in the antagonist group. Compared with the GnRH antagonist group, the GnRH agonist group had a lower risk of cancellation before
retrieval (4.3 versus 5.2%; P < 0.05), and between retrieval and transfer (5.5 versus 6.8%; P < 0.05) (Table 1). The agonist group had 691 cancelled cycles (68% of which were due to low ovarian response, 17% of which were due to high ovarian response, and 14% were ‘withdrawal only for personal reasons’). The antagonist group had 367 cancelled cycles (76% due to low ovarian response, 8% due to high ovarian response, 14% ‘withdrawal only for personal reasons’, and the remaining 2% included inadequate endometrial response, concurrent illness, and inability to obtain a semen specimen). The mean number of oocytes retrieved was comparable for the two groups; 15.8 (SD 7.9) compared with 16.0 (SD 8.5) in the agonist and antagonist groups, respectively. More elective SET were carried out in the antagonist group (12.9 versus 10.9%; P < 0.05). The rate of moderate ovarian hyperstimulation was greater in the agonist group (2.1 versus 1.7%; P < 0.05), similar rates of severe hyperstimulation were reported between the two groups (0.7 versus 0.6%). Among those cycles complicated by hyperstimulation, 16.2% required hospitalization in the agonist group compared with 19.8% in the antagonist group. The implantation (57.6 versus 54.1%; P < 0.05) and live birth rates (52.6 versus 49.2%; P < 0.05) per transfer were significantly higher in the agonist compared with the antagonist group. The final mixed-effect multivariable logistic model (n = 18,021 cycles) adjusted for patient age, gravidity, assisted reproductive technology indication, FSH dose, body mass index, and number of embryos transferred. The multivariable model showed higher implantation (adjusted OR 1.14, CI 1.03 to 1.26) and live birth per transfer (adjusted OR 1.13, CI 1.03 to 1.25) in the agonist compared with antagonist group (Table 2). The results were similar for the subgroup analysis of the 2029 cycles involving elective SET; the implantation rate (adjusted OR 1.36, CI 1.08 to 1.73) and live birth
Fresh cycles, patient stimulation = 203,302
84,128
53,359
Agonist + FSH (n = 16,001)
691 cancellations
Agonist + FSH (n = 15,310)
Figure 1
Selection of good-prognosis patients.
301
Age ≥ 35 years = 119,174 Prior assisted reproduction technology cycles = 30,769 Missing baseline FSH = 15,546 Baseline FSH ≥ 10 = 4869
32,944
FSH dose < 1200 or > 3600 or missing = 8396
24,548
Other protocols = 1435
23,113
Antagonist + FSH (n = 7112)
367 cancellations
Antagonist + FSH (n = 6745)
302
D Grow et al. Table 1 Population characteristics: GnRH agonist protocols, GnRH antagonist protocols, reported per autologous IVF cycle, 2009–2010. Characteristics
Age (years, mean ± SD) Prior pregnancy (gravidity) (%) Prior live birth (full-term or preterm) (%) Prior pregnancy loss (spontaneous abortions) (%) Infertility diagnosis (%) Tubal factor Ovulation disorders Endometriosis Male factor Other factors Unexplained Multiple factors (female) Multiple factors (male and female) FSH dose (mean ± SD) Body mass index (mean ± SD) Race White Others Missing Ethnicity Hispanic or Latino Non-Hispanic or Latino Missing Cancellation before retrieval (%) Retrievals not leading to transfer (%) Outcomes Number of cycles reaching embryo transfer Number of oocytes retrieved (mean ± SD)b Total number of embryos transferred (%) 1 2 ≥3 Cycles with implantation (%)c Live birthd for all cycles (%) Elective SET cyclesd n (%) Live birthd for eSET cycles (%) Moderate ovarian hyperstimulation n (%) Severe ovarian hyperstimulation n (%)
Stimulation protocol Agonist + FSH (n = 16,001)
Antagonist + FSH (n = 7112)
30.4 ± 2.7 36.0 16.2
30.5 ± 2.8 37.4a 17.2
19.6
19.9
9.4 11.8 5.1 28.6 7.7 15.3 6.2 15.9 2185.9 ± 625.8 25.4 ± 5.7 (n = 13,913)
7.8a 12.2 4.5 26.8 9.8 14.6 7.5 16.8 2230.0 ± 636.2a 25.3 ± 5.6 (n = 6189)
52.0 13.3 34.6
50.0a 12.9 37.0
6.4 73.5 20.1 4.32 5.5
7.7a 66.0 26.4 5.16a 6.8a
14,464
6289
15.8 ± 7.9
16.0 ± 8.5
15.40 77.05 7.56 8331 (57.6) 52.6 1489 (10.85) 53.93 342 (2.1) 118 (0.7)
17.43a 74.70 7.87 3403 (54.1)a 49.2a 767, 12.87a 46.15a 120, 1.7a 42 (0.)6
P < 0.05. Calculated per transfer where at least one embryo was transferred (denominator equivalent to “number of cycles reaching embryo transfer”). c “Cycles with implantation” – implantation calculated per cycle among cycles where at least one embryo was transferred. d Calculated per transfer where at least one embryo transferred and at least one embryo cryopreserved (denominators are 13,726 for agonist and 5960 for antagonist). a
b
Agonist v Antagonist protocol: outcomes comparison among good prognosis patients
303
Table 2 Odds of implantation and live birth per transfer among good-prognosis patients using agonist and antagonist protocols among completed fresh, autologous IVF cycles between 2009 and 2010.
All good prognosis completed cyclesa Implantation Live birth All cycles with elective SET Implantation Live birth
Odds ratio (confidence interval)
Adjusted odds ratio (confidence interval)
1.15 (1.04 to 1.27) 1.14 (1.04 to 1.26)
1.14 (1.03 to 1.26) 1.13 (1.03 to 1.25)
1.41 (1.13 to 1.76) 1.37 (1.11 to 1.68)
1.36 (1.08 to 1.73) 1.33 (1.07 to 1.66)
eSET, elective single embryo transfer; n = 2256 cycles for bivariable and 2029 cycles for multivariable analyses. Adjusted for patient age, gravidity, assisted reproductive technology indication, FSH dose and body mass index. a n = 20,753 cycles for bivariable analyses; 18, 021 cycles for multivariable analyses. ‘Good prognosis’ defined as woman’s age younger than 35 years, maximum baseline serum FSH less than 10 mIU/mL, no prior history of IVF, total FSH dose between 1200 and 3600 IU, and body mass index between 18 and 45 kg/m2. Adjusted for patient age, gravidity, ART indication, FSH dose, body mass index, number of embryos transferred.
rate (adjusted OR 1.33, CI 1.07 to 1.66) were higher in the agonist compared with the antagonist group.
Discussion Our study suggests that, among women who have a good prognosis as defined by age, fertility treatment history, ovarian reserve testing, and stimulation medication requirements, long GnRH-agonist protocols are associated with a lower cancellation rate and higher implantation and live birth compared with GnRH-antagonist protocols. These findings remained significant among the subgroup of good-prognosis patients undergoing elective SET. The antagonist protocols conferred a decreased risk of moderate but not severe ovarian hyperstimulation. The increased likelihood of cancellation before retrieval and before transfer among antagonist cycles may reflect an increased likelihood of ovarian cyst formation before stimulation medication start, over- or under-response to stimulation, or premature ovulation. The lower risk of hyperstimulation associated with antagonist protocols may be underestimated because those women for whom an antagonist protocol was selected may have been assigned to the antagonist group owing to their suspected risk of OHSS. Additionally, the differentiation between moderate and severe OHSS is a continuum, and classification may vary between clinics. Although the large sample size makes it statistically significant, the difference in mean total FSH dose between the agonist and antagonist groups was only 45 IU, and is likely not to be clinically significant. The increased likelihood of elective SET among antagonist cycles may reflect an attempt to decrease OHSS risk or clinic transfer tendency; nonetheless, implantation rate and live birth rate remained lower within the elective SET group for those who underwent an antagonist protocol after adjusting for patient age, gravidity, assisted reproductive technology indication, FSH dose, and body mass index. Previous studies including retrospective reviews and randomized controlled trials have reported conflicting results on
the effect of protocol choice on implantation and live birth (Al-Inany et al., 2006, 2011; Griesinger et al., 2005). A large number of multiple target tissues are affected by GnRH analogues, including the pituitary, ovary and endometrium. As a result, agonist and antagonists may alter the ovarian response, endometrial environment, and hormonal milieu at time of embryo transfer differently (Hsueh and Jones, 1981). The effect of protocol choice on implantation rate may reflect the possible deleterious effect of GnRH antagonists on the endometrium, and the possible increased competency oocytes exposed to GnRH agonists (Check et al., 2011; Ortmann et al., 2001; Schachter et al., 2008). The different effect on IVF outcomes between the two protocols is most likely to be subtle, and therefore difficult to capture without a large sampling of patients. Our study is strengthened by the large sample size and the ability to conduct subgroup analysis of good-prognosis patients undergoing elective SET. Previous RCTs were smaller (ranging from 20–730 participants), and either included a more diverse patient population or focused specifically on women with diminished ovarian reserve or polycystic ovary syndrome (Al-Inany et al., 2006, 2011). Additionally, our study reflects more current assisted reproductive technique, whereas previous studies span a time period during which treatment protocols and effectiveness may have evolved. Study limitations include dependence on data entered by individual clinics, inability to control for nuanced differences between stimulation protocols among different clinicians, such as oral contraceptive pill pretreatment, the subjective definition of ovarian hyperstimulation, and selection bias in protocol selection that accompanies a restrpospective analysis. We attempted to minimize this by selecting good-prognosis patients undergoing elective SET. It seems that, among good-prognosis patients, GnRHagonist protocols may confer decreased risk of cancellation and an increased odds of implantation and live birth; however, the GnRH-antagonist protocols may be associated with decreased risk of hyperstimulation. Protocol choice is a nuanced decision that incorporates patient reproductive treatment history, suspected ovarian response to medication, and
304 clinician and patient preference. The results of a single study may contribute additional evidence to the decision-making process, but should not drastically change clinical practice. Ideally, a large prospective randomized trial, although costly, would be the optimal method by which to gain further insight into the effect of protocol choice on pregnancy rates and outcomes.
Acknowledgements Dr. Donna Session participated in the review and critique of the manuscript; she did not receive any financial compensation.
References Aboulghar, M.A., Mansour, R.T., Serour, G.I., Al-Inany, H.G., Amin, Y.M., Aboulghar, M.M., 2004. Increasing the dose of human menopausal gonadotrophins on day of GnRH antagonist administration: randomized controlled trial. Reprod. Biomed. Online 8, 524–527. Al-Inany, H.G., Abou-Setta, A.M., Aboulghar, M., 2006. Gonadotrophinreleasing hormone antagonists for assisted conception. Cochrane Database Syst. Rev. (3), CD001750. Al-Inany, H.G., Youssef, M.A., Aboulghar, M., Broekmans, F., Sterrenburg, M., Smit, J., Abou-Setta, A.M., 2011. Gonadotrophinreleasing hormone antagonists for assisted reproductive technology. Cochrane Database Syst. Rev. (5), CD001750. Albano, C., Felberbaum, R.E., Smitz, J., Riethmuller-Winzen, H., Engel, J., Diedrich, K., Devroey, P., 2000. Ovarian stimulation with HMG: results of a prospective randomized phase III European study comparing the luteinizing hormone-releasing hormone (LHRH)-antagonist cetrorelix and the LHRH-agonist buserelin. European Cetrorelix Study Group. Hum. Reprod. 15, 526–531. Barmat, L.I., Chantilis, S.J., Hurst, B.S., Dickey, R.P., 2005. A randomized prospective trial comparing gonadotropin-releasing hormone (GnRH) antagonist/recombinant follicle-stimulating hormone (rFSH) versus GnRH-agonist/rFSH in women pretreated with oral contraceptives before in vitro fertilization. Fertil. Steril. 83, 321–330. Bodri, D., Sunkara, S.K., Coomarasamy, A., 2010. Gonadotropinreleasing hormone agonists versus antagonists for controlled ovarian hyperstimulation in oocyte donors: a systematic review and meta-analysis. Fertil. Steril. 95, 164–169.
D Grow et al. Cetel, N.S., Rivier, J., Vale, W., Yen, S.S., 1983. The dynamics of gonadotropin inhibition in women induced by an antagonistic analog of gonadotropin-releasing hormone. J. Clin. Endocrinol. Metab. 57, 62–65. Check, J.H., Cohen, R., Amui, J., Choe, J.K., Brasile, D., 2011. Evidence that the main adverse effect of ganirelix on pregnancy and implantation rates is on the embryo rather than the endometrium. Clin. Exp. Obstet. Gynecol. 38, 326–327. Engmann, L., Diluigi, A., Schmidt, D., Nulsen, J., Maier, D., Benadiva, C., 2008. The use of gonadotropin-releasing hormone (GnRH) agonist to induce oocyte maturation after cotreatment with GnRH antagonist in high-risk patients undergoing in vitro fertilization prevents the risk of ovarian hyperstimulation syndrome: a prospective randomized controlled study. Fertil. Steril. 89, 84–91. Griesinger, G., Felberbaum, R., Diedrich, K., 2005. GnRH antagonists in ovarian stimulation: a treatment regimen of clinicians’ second choice? Data from the German national IVF registry. Hum. Reprod. 20, 2373–2375. Hsueh, A.J., Jones, P.B., 1981. Extrapituitary actions of gonadotropinreleasing hormone. Endocr. Rev. 2, 437–461. Marci, R., Graziano, A., Lo Monte, G., Piva, I., Soave, I., Marra, E., Lisi, F., Moscarini, M., Caserta, D., 2013. GnRH antagonists in assisted reproductive techniques: a review on the Italian experience. Eur. Rev. Med. Pharmacol. Sci. 17, 853–873. Ortmann, O., Weiss, J.M., Diedrich, K., 2001. Embryo implantation and GnRH antagonists: ovarian actions of GnRH antagonists. Hum. Reprod. 16, 608–611. Schachter, M., Friedler, S., Ron-El, R., Zimmerman, A.L., Strassburger, D., Bern, O., Raziel, A., 2008. Can pregnancy rate be improved in gonadotropin-releasing hormone (GnRH) antagonist cycles by administering GnRH agonist before oocyte retrieval? A prospective, randomized study. Fertil. Steril. 90, 1087–1093. Sunderam, S., Kissin, D.M., Flowers, L., Anderson, J.E., Folger, S.G., Jamieson, D.J., Barfield, W.D., 2012. Assisted reproductive technology surveillance – United States, 2009. MMWR Surveill. Summ. 61, 1–23.
Authors’ roles: DG, AK, TD, JFK and MM were involved in the study design, analysis, manuscript drafting, and critical discussion.
Declaration: The authors report no financial or commercial conflicts of interest.
Received 28 January 2014; refereed 15 May 2014; accepted 21 May 2014.
