This was a multi-center randomized parallel-group trial comparing three different protocols that incorporated different modes of GnRH analogue administration. The study complied with the Declaration of Helsinki and was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-1012-114-345), and each participating hospital. This study has been registered at ClinicalTrials.gov (No. NCT01402336). Patients were enrolled from the following four university-based or university-affiliated hospitals, which submitted case report forms without rule violations: Seoul National University Hospital, CHA Gangnam Medical Center, CHA Bundang Medical Center, and Cheil General Hospital and Women's Healthcare Center.

Eligible participants had PCOS as defined by the Rotterdam criteria [12,13]. All women therefore had at least two of the three following characteristics: (1) oligo- or anovulation, (2) clinical and/or biochemical signs of hyperandrogenism, and (3) polycystic ovaries with the exclusion of other possible etiologies. Further inclusion criteria included: age 20–40 years; presence of both ovaries without ovarian tumors; a normal uterine cavity as assessed through an ultrasound scan, hysterosalpingograms, or hysteroscopy; normal renal, liver and hematological indices; normal thyroid function; and normal prolactin levels. The exclusion criteria were an extrauterine pregnancy or abortion in the past 3 months; abnormal gynecological bleeding of undetermined origin; a congenital uterine anomaly; congenital adrenal hyperplasia; androgen-secreting tumors; Cushing syndrome; any contraindication for recombinant follicle-stimulating hormone (rFSH), GnRH analogue, or human chorionic gonadotropin (hCG) administration; and severe male factor infertility. Eligible subjects were identified prospectively from women who planned to start an in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI) treatment cycle. Written informed consent was obtained from each woman who wished to participate. After confirming patient eligibility, a blood test was performed to obtain the baseline female reproductive hormone profile, and height and weight were measured. Patients were randomized 1:1:1 to the following three groups using a web-based system: the day 1 antagonist group, the day 6 antagonist group, or the long agonist group. The web-based system was provided by the Medical Research Collaborating Center of Seoul National University Hospital (SNUH; https://mrcc.snuh.org). Stratified randomization was used (stratification according to the institution, block randomization with a randomly selected block size). The investigator was blind to the randomization process, including the size of each block. Neither patients nor doctors were blinded to the treatment assigned. Any patients who withdrew their consent continued with their routine IVF treatment and were excluded from the analysis. All serious adverse events were submitted to the Clinical Trials Unit, Seoul National University Hospital, and the trial was evaluated periodically by an independent data and safety monitoring board.

The schematic representation of each protocol is shown in Figure 1. All subjects received oral contraceptive pills (OCPs) for 21 days starting on menstrual day 5 in the cycle prior to the treatment cycle. An OCP containing ethinyl E₂ (0.03 mg) and cyproterone acetate (2 mg) (Diane-35; Bayer AG, Leverkusen, Germany) was recommended, but other OCPs were permitted. A starting dose of 150 IU of rFSH (300 IU, 450 IU, 900 IU; Gonal-F, Merck Serono, London, UK) was recommended, but alterations were allowed if necessary (Table 1). GnRH analogues were administered as described below.

Recombinant hCG (250 µg; Ovidrel, Merck Serono) or urinary hCG (10,000 IU; IVF-C, LG Life Sciences, Iksan, Korea) was administered when at least two follicles reached 17 mm in diameter. Oocyte retrieval was performed 36–38 hours following hCG administration, and fertilization was carried out using IVF, ICSI or half ICSI following the investigator's decision. For luteal support, daily vaginal progesterone (Crinone, Merck Serono) was started on the day of oocyte retrieval and was continued up to 8 weeks. When severe OHSS was anticipated, embryo transfer was waived, and all embryos were cryopreserved. Since many different criteria exist for performing cryopreservation of all embryos for OHSS prevention, and the purpose of this study was not to evaluate the OHSS prevention strategy, the decision was left to the clinicians. Embryo transfer was performed 2–5 days after oocyte retrieval, and a serum hCG test was performed 2 weeks after fertilization to confirm pregnancy. The gestational sac and fetal heartbeat were identified using transvaginal ultrasound in patients weekly thereafter.

The morphological assessment of day 3 embryos was performed according to the following criteria: grade 1, blastomeres of equal size, no cytoplasmic fragments; grade 2, blastomeres of equal size, minor cytoplasmic fragments or blebs; grade 3, blastomeres of distinctly unequal size, few or no cytoplasmic fragments; grade 4, blastomeres of equal or unequal size, significant cytoplasmic fragmentation; grade 5, few blastomeres of any size, severe or complete fragmentation. Good-quality embryos were defined as grade 1 or grade 2 embryos. Blastocysts were evaluated on day 5 using a scoring system based on the expansion state of the blastocyst and the consistency of the inner cell mass and trophectoderm cells. Good-quality embryos were defined as those with a score of 3BB or higher.

The primary endpoint of this study was the number of oocytes retrieved. The secondary endpoints were the clinical pregnancy rate and the occurrence of moderate-to-severe OHSS. The classification by Golan and Weissman [14] was used to assess the severity of OHSS.

We set the baseline number of oocytes retrieved in the GnRH agonist long protocol for PCOS patients to be 13.2±5.8 based on the study by Bahceci et al. [15] and the SNUH data (unpublished), and the detectable difference between groups at 4.9, assuming an α level of 0.025. The α level was halved since we compared the early antagonist group with two other groups. It was calculated that a sample size of 80 patients was required in each group to achieve a 0.80 power. Such a sample size (80 patients per group) is not easily achieved even in a multicenter study, when studying a small proportion of the population, such as PCOS patients who undergo IVF. The previous two studies on this topic reported 49 patients [10] and 78 patients [11] and we initiated this study because we thought we could provide useful clinical information, even with small number of subjects.

The results of this preliminary report are presented as median (range), and nonparametric tests were used for statistical analysis. Continuous variables, including the primary endpoint (number of oocytes retrieved) were compared using the Kruskal-Wallis test. The comparison of the frequency of features, including the secondary endpoints (the clinical pregnancy rate and the occurrence of moderate-to-severe OHSS) were performed using the chi-square test. All statistical analyses were performed using IBM SPSS ver. 22.0 (IBM Corp., Armonk, NY, USA). The α level of significance was set at 0.05.

Between February 2011 and December 2016, a total of 36 infertile women with PCOS were enrolled in this study, after the exclusion of seven women who discontinued participation after the initial agreement (Figure 2). One cycle in the early antagonist group was cancelled due to poor ovarian response, leaving a total of 35 cycles eligible for OHSS surveillance. No oocyte was retrieved in one patient in the conventional antagonist group. The cryopreservation of all embryos without embryo transfer was performed in a total of eight patients; seven of these procedures were performed to prevent OHSS, and one was due to hemoperitoneum. Embryo transfer was performed in the remaining 26 cycles. Age, body mass index, and base-line hormone levels were comparable in all three groups (Table 1). The characteristics of the cycles are shown in Table 2, and the IVF outcomes are shown in Table 3.

The number of oocytes retrieved was greatest in the long agonist group (n=19), followed by the early antagonist group (n=16) and the conventional antagonist group (n=12). However, the difference was not statistically significant (p=0.111). The number of mature oocytes retrieved also followed a similar order, and likewise no statistically significant difference was observed among the early antagonist (n=10.5), conventional antagonist (n=8), and long agonist (n=13) groups (p=0.112).

The clinical pregnancy rate was greatest in the early antagonist group (50.0%), followed by the long agonist group (22.2%) and the conventional antagonist group (11.1%), although this difference was not statistically significant (p=0.180).

The incidence of moderate-to-severe OHSS was lowest in the early antagonist group (7.7%), followed by the conventional antagonist group (18.2%) and the long agonist group (27.3%). However, this trend was not statistically significant (p=0.463). The incidence of moderate-to-severe OHSS among patients at high risk for OHSS (E₂ concentration on hCG day >2,000 pg/mL) was lowest in the early antagonist group (12.5%), in contrast to the conventional antagonist group (40%) and the long agonist group (50.0%), but this difference was also not statistically significant (p=0.324).

The duration of stimulation was similar among the three groups. A higher starting dose of FSH seemed to be preferred in the long agonist group, but there was no statistically significant difference among the three groups. The total dose of gonadotropin was greatest in the early antagonist group (median, 1,931 IU), followed by the long agonist group (median, 1,800 IU) and conventional antagonist group (median, 1,350 IU), but the difference was not statistically significant. Although the E₂ concentration on day 6 seemed to be much higher in the early antagonist group than in the conventional antagonist group, the difference was not statistically significant. The E₂ concentration on the day of hCG administration was highest in the early antagonist group (median, 3,795 pg/mL), and was lowest in the conventional antagonist group (median, 969 pg/mL), but the difference was not statistically significant. The number of large follicles on the hCG day was slightly higher in the early antagonist group than in the conventional antagonist group, but this difference did not reach statistical significance. There was no difference in oocyte maturity among the three groups (p=0.800). The number of cycles that underwent embryo cryopreservation for OHSS prevention was greatest in the early antagonist group (n=4). The reason for this may have been the high E₂ concentration on the day of hCG although the difference from other groups was not statistically significant. However, COS was not more intense in the patients with OHSS in the early antagonist group since the median total dose of gonadotropin was not significantly different between patients with OHSS (2,006.25 IU) and patients without OHSS (1,650 IU, p=0.708).