Delayed initiation of GnRH antagonist in PCOS patients undergoing IVF: a retrospective cohort study of premature ovulation and embryological outcomes Authors: Nuri Peker, Ayşen Yücetürk, Esra Tığlı, Özge Karaosmanoğlu, Burak Elmas, İlke Özer Aslan, Bülent Tıraş Published: 2026 ABSTRACT BackgroundWe aimed to evaluate whether delayed initiation of gonadotropin hormone-releasing hormone (GnRH) antagonist treatment affects premature ovulation and the number of good-quality blastocysts.We retrospectively reviewed the medical records of polycystic ovary syndrome (PCOS) patients who underwent in vitro fertilization (IVF) treatment at our clinic between 2017 and 2025. Patients were divided into two groups: Group 1: Patients in whom GnRH antagonist injections were initiated when at least three dominant follicles measured ≥17 mm. Group 2: Patients in whom GnRH antagonist was initiated upon identification of a single dominant follicle reaching ≥14 mm. The groups were compared in terms of premature ovulation, estradiol (E2) level at GnRH antagonist initiation, total number of retrieved oocytes, number and proportion of metaphase II (MII) oocytes, fertilization rate, blastocyst number and rate, and good-quality blastocyst grades.ResultsA total of 89 patients were included in the study. Of these, 38 patients were assigned to Group 1 and 51 to Group 2. There were no statistically significant differences in age or body-mass index (BMI) between the groups. Serum E2 levels at GnRH antagonist initiation were significantly higher in Group 1. The “median-(Interquartile Range (IQR))” total oocyte count was 33 (27–41) in Group 1 and 30 (26–40) in Group 2. The median 2PN count was 19.5 (12.75–27) in Group 1 and 19.0 (15–22) in Group 2. There were no statistically significant differences between the groups in terms of MII oocyte or two pronuclei (2PN) count. Blastocyst outcomes were comparable between the groups; the median number of day 5–6 blastocysts was 7.0 (IQR: 5-11.5) in Group 1 and 8.0 (IQR: 5–11) in Group 2. Furthermore, the median number of Grade A blastocysts was 4.0 (IQR: 3-10) in Group 1 and 5.0 (IQR: 4-8) in Group 2, and there was no statistically significant difference between the groups regarding Grade‑A blastocyst counts. No premature ovulation occurred in either group.ConclusionIn this PCOS cohort, delayed GnRH antagonist initiation was not associated with impaired laboratory embryological outcomes or increased premature ovulation risk. Prospective randomized controlled trials are needed to confirm these preliminary findings. Background Gonadotropin-releasing hormone (GnRH) antagonists are used during in vitro fertilization (IVF) to prevent premature ovulation by suppressing pituitary gonadotropin release. There are two different protocols: the flexible and the fixed protocols [1, 2]. In the flexible protocol, a GnRH antagonist is initiated when a dominant follicle exceeds 14 mm in diameter and/or serum estradiol (E2) exceeds 300 pg/mL [1, 2]. On the other hand, in the fixed protocol, GnRH antagonist administration is initiated on the 5th or 6th day of stimulation, regardless of dominant follicle size or E2 concentration [1, 2]. Multiple studies have demonstrated that cumulative live birth rates do not differ significantly between fixed and flexible protocols across various patient groups [1, 2]. In addition to GnRH antagonist protocols, alternative approaches to prevent premature ovulation include oral progestins (medroxyprogesterone acetate, dydrogesterone, and micronized progesterone); the GnRH agonist protocols; and newer oral GnRH antagonist agents used to prevent premature ovulation [3]. In patients with PCOS, both injectable GnRH antagonists and oral progestins can be used to prevent premature ovulation. Oral progestins are an important alternative in polycystic ovary syndrome (PCOS) because fresh embryo transfer is often avoided and a freeze-all strategy is commonly applied. Besides, there are also novel approaches such as initiating GnRH antagonist treatment on the first day of stimulation or initiating GnRH antagonist based on luteinizing hormone (LH) levels during the controlled ovarian hyperstimulation (COH) cycle. A 2025 study reported that a day-1 antagonist protocol yielded a higher number of retrieved oocytes and improved embryo quality [4]. In our retrospective study, we reevaluated the criteria for initiating GnRH antagonist therapy in patients with PCOS and aimed to assess whether delayed GnRH antagonist initiation is associated with premature ovulation and reduced blastocyst quality. Methods Patients were identified through retrospective screening of GnRH antagonist cycles performed at our center between 2017 and 2025. The flow of participants through the study, including screening, exclusion, and group allocation, is shown in Fig. 1. The study was reviewed and approved by the local ethics committee and performed in accordance with the principles of the 1975 Declaration of Helsinki as revised in 2000 (Approval No:2025-18/665). Informed consent was obtained from all participants prior to inclusion. Patients were identified through retrospective screening of GnRH antagonist cycles performed at our center between 2017 and 2025. Fig. 1Flowchart of participants through the study. After setting n = 89, the reasons for exclusion from the remaining cycles included patients for whom the pregnancy outcome (clinical pregnancy, miscarriage, or live birth) could not be followed up after embryo transfer (patients from outside Turkey or outside Istanbul) and patients whose medical records lacked blood test results or folliculometry reports Flowchart of participants through the study. After setting n = 89, the reasons for exclusion from the remaining cycles included patients for whom the pregnancy outcome (clinical pregnancy, miscarriage, or live birth) could not be followed up after embryo transfer (patients from outside Turkey or outside Istanbul) and patients whose medical records lacked blood test results or folliculometry reports Inclusion criteria were female patients aged 20–45 years and patients who have at least 12 follicles in each ovary on ultrasound evaluation. Importantly, we included patients in whom GnRH antagonist treatment was initiated late, defined as initiation when at least three dominant follicles measured ≥ 17 mm while smaller follicles required additional stimulation. Patients with ≥ 5 follicles measuring ≥ 17 mm who underwent trigger the same night were also included. All patients had been diagnosed with PCOS according to the Rotterdam consensus criteria, requiring the presence of at least two of the following three features: oligo- and/or anovulation, clinical and/or biochemical signs of hyperandrogenism, and polycystic ovarian morphology on ultrasound. Exclusion criteria include: presence of systemic diseases (hypothyroidism, Cushing’s syndrome, hypertension, type 1 diabetes, etc.), recurrent IVF failure, recurrent implantation failure, and presence of azoospermia. Patients were divided into two groups:Group 1: GnRH antagonist injections were initiated when at least three dominant follicles reached ≥17 mm.The ≥17 mm threshold was selected because this represents the minimum follicle diameter at which the final oocyte maturation trigger is considered at our center. Initiating GnRH antagonist at this point therefore reflects the latest clinically permissible timepoint before triggering, allowing assessment of the maximum extent to which antagonist initiation can be safely delayed.Group 2: Patients in whom a GnRH antagonist was initiated upon identification of a single dominant follicle reaching ≥14 mm on transvaginal ultrasound, consistent with the flexible GnRH antagonist protocol applied at our center. Group 1: GnRH antagonist injections were initiated when at least three dominant follicles reached ≥17 mm.The ≥17 mm threshold was selected because this represents the minimum follicle diameter at which the final oocyte maturation trigger is considered at our center. Initiating GnRH antagonist at this point therefore reflects the latest clinically permissible timepoint before triggering, allowing assessment of the maximum extent to which antagonist initiation can be safely delayed. Group 1: GnRH antagonist injections were initiated when at least three dominant follicles reached ≥17 mm. The ≥17 mm threshold was selected because this represents the minimum follicle diameter at which the final oocyte maturation trigger is considered at our center. Initiating GnRH antagonist at this point therefore reflects the latest clinically permissible timepoint before triggering, allowing assessment of the maximum extent to which antagonist initiation can be safely delayed. Group 2: Patients in whom a GnRH antagonist was initiated upon identification of a single dominant follicle reaching ≥14 mm on transvaginal ultrasound, consistent with the flexible GnRH antagonist protocol applied at our center. Group 2: Patients in whom a GnRH antagonist was initiated upon identification of a single dominant follicle reaching ≥14 mm on transvaginal ultrasound, consistent with the flexible GnRH antagonist protocol applied at our center. All cycles included in this study utilized intracytoplasmic sperm injection (ICSI) as the fertilization method; no conventional IVF cycles were included. These two groups were compared in terms of premature ovulation, E2 level at initiation of GnRH antagonist injection, total number of retrieved oocytes, number and proportion of metaphase II oocytes, fertilization rate, and blastocyst number/rate and blastocyst grade. Furthermore, IVF success was also evaluated clinically, and additional assessments were performed regarding pregnancy, miscarriage, and live birth rates. COH protocol IVF treatment was conducted as follows: On the 2nd or 3rd day of the menstrual cycle, a transvaginal ultrasound is performed to assess the antral follicle count, and COH is initiated. We apply a flexible protocol for initiating the GnRH antagonist, and based on the ultrasound evaluation performed on day 2 of menstruation, patients were called for the first follow-up visit 4 or 5 days later. At this visit, follicle diameters are measured, and serum E2 levels are evaluated. The timing of the next follow-up is scheduled according to follicle sizes and E2 levels. During this process, when the dominant follicle reaches 14 mm or more, a GnRH antagonist injection is added to the treatment to prevent premature ovulation. Patients continue to be monitored at regular intervals, and when at least three follicles reach a diameter of 17 mm or more, final oocyte maturation is triggered using human chorionic gonadotropin (hCG), gonapeptyl, or a combination of gonapeptyl + hCG. Oocyte retrieval is performed 36 h after triggering, under sedation and with transvaginal ultrasound guidance. All patients received a combination of FSH (225–300 IU) and LH (75 IU) for ovarian stimulation. In PCOS patients, final oocyte maturation was triggered using a GnRH agonist. Although gonadotropin doses showed small variations across the study period (2017–2025), no statistically significant difference in stimulation parameters was identified between the groups (data not shown). Because of high patient volume, some patients were unable to attend scheduled follow-up visits, resulting in delays of 2–3 days; therefore, GnRH antagonist initiation was delayed in some cycles. Approximately 3 h after retrieval, oocytes were denuded using hyaluronidase, and mature metaphase II (MII) oocytes were fertilized by ICSI. Immature MI and GV oocytes are not fertilized. Fertilization assessment is performed 18–22 h after ICSI, and embryos with confirmed fertilization (2PN) are monitored until day 3 or day 5. All embryos were cryopreserved for subsequent frozen–thawed embryo transfer. Premature ovulation was operationally defined as any of the following clinical events: cycle cancellation due to ovulation prior to scheduled oocyte retrieval, spontaneous follicular rupture identified on transvaginal ultrasound at the day of trigger or at oocyte pick-up, or unexpected absence of oocytes during follicular aspiration. Because LH and progesterone levels were not measured during stimulation, the absence of premature ovulation in this study refers exclusively to the absence of clinically evident events rather than biochemical confirmation of LH suppression. Blastocysts were graded according to the Gardner grading system. Grade A blastocysts were defined as those with a tightly packed inner cell mass (ICM) and a cohesive trophectoderm with many cells. Grade B blastocysts were defined as those with a loosely grouped ICM and a trophectoderm with few cells. Statistical analysis Statistical analyses were performed using GraphPad Prism (GraphPad Software v9.0.0, San Diego, CA, USA). Normality of continuous variables was assessed using the Shapiro–Wilk test. Data with normal distribution were expressed as mean ± standard deviation (SD) and compared using Welch-corrected independent samples t-test. Non-normally distributed variables were presented as median (range and interquartile range) and compared using the two-tailed Mann–Whitney U test. Within-group comparisons of Grade A and Grade B blastocyst counts which represent paired measurements from the same patient were performed using the Wilcoxon matched-pairs signed-rank test. Between-group comparisons were performed using the Mann-Whitney U test. Correlations between serum E2 levels and oocyte/embryo parameters were evaluated using Spearman’s rank correlation coefficient. All tests were two-tailed, and p < 0.05 was considered statistically significant. No a priori sample size calculation was performed given the retrospective design of the study. Post-hoc power analysis indicated that, based on the observed effect sizes for the primary outcomes (r = 0.02–0.12) and the available sample size (n = 89), the study had limited statistical power to detect small between-group differences. Results A total of 89 patients were included in the study; 38 were assigned to Group 1 and 51 to Group 2 (Fig. 1). Table 1 compares the two groups in terms of age and BMI. The mean age was 28.82 ± 3.80 years in Group 1 and 30.10 ± 3.84 years in Group 2, with no statistically significant difference between groups (t = 1.567, df = 80.37, p = 0.121; mean difference: 1.28 years, 95% CI: −0.35 to 2.91; η² = 0.030). The mean BMI was 25.69 ± 4.85 kg/m² in Group 1 and 26.66 ± 5.48 kg/m² in Group 2, again with no significant difference (t = 0.885, df = 84.34, p = 0.379; mean difference: 0.97 kg/m², 95% CI: −1.21 to 3.16; η² = 0.009). Effect sizes were small for both variables, confirming that age and BMI were balanced between groups. We compared two independent groups across two variables: treatment duration and GnRH agonist use duration (Table 2). Because none of the variables were normally distributed (Shapiro-Wilk, p < 0.05 for all), nonparametric tests and median-based descriptive statistics were used. Mann-Whitney U tests were conducted and indicated that treatment duration (u:572.50, p < 0.001) and GnRH antagonist duration (u:23.00, p < 0.001) were statistically significantly shorter in Group 1. Rank biserial correlation was used to quantify effect size, and treatment duration presented a moderate effect size (r = 0.41) whereas GnRH antagonist duration yielded an extremely large effect (r = 0.98). No significant differences were observed between the two groups in terms of the total FSH dose and antral follicle count were comparable between the two groups (Table 2). Table 1Comparison of age and BMI between the two groupsVariableGroup 1Group 2p value95% CIη²Age (years)28.82 ± 3.8030.10 ± 3.840.12091.28 (− 0.35 to 2.91)0.030BMI (kg/m²)25.69 ± 4.8526.66 ± 5.480.37880.97 (− 1.21 to 3.16)0.009p values were determined using Welch’s t-test. 95% CI: confidence interval of the mean difference (Group 2 − Group 1). η²: eta-squared effect size; values of 0.01, 0.06, and 0.14 indicate small, medium, and large effects, respectively*Data are expressed as mean ± standard deviation Table 1Comparison of age and BMI between the two groupsVariableGroup 1Group 2p value95% CIη²Age (years)28.82 ± 3.8030.10 ± 3.840.12091.28 (− 0.35 to 2.91)0.030BMI (kg/m²)25.69 ± 4.8526.66 ± 5.480.37880.97 (− 1.21 to 3.16)0.009p values were determined using Welch’s t-test. 95% CI: confidence interval of the mean difference (Group 2 − Group 1). η²: eta-squared effect size; values of 0.01, 0.06, and 0.14 indicate small, medium, and large effects, respectively*Data are expressed as mean ± standard deviation Comparison of age and BMI between the two groups p values were determined using Welch’s t-test. 95% CI: confidence interval of the mean difference (Group 2 − Group 1). η²: eta-squared effect size; values of 0.01, 0.06, and 0.14 indicate small, medium, and large effects, respectively *Data are expressed as mean ± standard deviation Table 2Comparison of controlled ovarian hyperstimulation (COH) parameters between group 1 and group 2VariableGroup 1Group 2Mann-Whitney UP ValueEffect size (r)COH duration (median (Q1-3)) (day)9.5 (9–10)10 (10–11)572.50P < 0.0010.41GnRH antagonist treatment (median (Q1-3)) (day)3 (2–3)5 (4–5)23.00P < 0.0010.98Total FSH dose (per cycle) IU2400 (2150–2700)2625 (2250–3300)735.5P = 0.05390.2Antral follicle count (Right ovary)20 (15–25)15 (15–20)1104.0P = 0.2470.12Antral follicle count (left ovary)20 (15–25)15 (15–20)1078.0P = 0.3510.10Continuous variables are presented as median (Q1-Q3). Effect size is reported as the r coefficient for the Mann-Whitney U test. Significant differences were observed between the two groups in terms of COH duration and GnRH antagonist injection duration. The effect size for GnRH antagonist injection duration was extremely large. Total FSH dose per cycle did not differ significantly. Likewise, right and left ovarian antral follicle count showed no statistically significant differences*Mann-Whitney U test is used to compare two groups Table 2Comparison of controlled ovarian hyperstimulation (COH) parameters between group 1 and group 2VariableGroup 1Group 2Mann-Whitney UP ValueEffect size (r)COH duration (median (Q1-3)) (day)9.5 (9–10)10 (10–11)572.50P < 0.0010.41GnRH antagonist treatment (median (Q1-3)) (day)3 (2–3)5 (4–5)23.00P < 0.0010.98Total FSH dose (per cycle) IU2400 (2150–2700)2625 (2250–3300)735.5P = 0.05390.2Antral follicle count (Right ovary)20 (15–25)15 (15–20)1104.0P = 0.2470.12Antral follicle count (left ovary)20 (15–25)15 (15–20)1078.0P = 0.3510.10Continuous variables are presented as median (Q1-Q3). Effect size is reported as the r coefficient for the Mann-Whitney U test. Significant differences were observed between the two groups in terms of COH duration and GnRH antagonist injection duration. The effect size for GnRH antagonist injection duration was extremely large. Total FSH dose per cycle did not differ significantly. Likewise, right and left ovarian antral follicle count showed no statistically significant differences*Mann-Whitney U test is used to compare two groups Comparison of controlled ovarian hyperstimulation (COH) parameters between group 1 and group 2 Continuous variables are presented as median (Q1-Q3). Effect size is reported as the r coefficient for the Mann-Whitney U test. Significant differences were observed between the two groups in terms of COH duration and GnRH antagonist injection duration. The effect size for GnRH antagonist injection duration was extremely large. Total FSH dose per cycle did not differ significantly. Likewise, right and left ovarian antral follicle count showed no statistically significant differences *Mann-Whitney U test is used to compare two groups Serum E2 levels measured on the day GnRH antagonist was initiated showed a statistically significant difference between the two groups; E2 levels in Group 1 were significantly higher than those in Group 2 (U = 229, p < 0.0001). The median (IQR) [min–max] E2 value was 2995 (IQR: 2533–3697) [1452–10623] pg/mL in Group 1 (n = 38) and 1497 (IQR: 1221–2264) [294–3923] pg/mL in Group 2 (n = 51) (Fig. 2). Fig. 2E2 levels on the day the GnRH antagonist was initiated E2 levels on the day the GnRH antagonist was initiated The distribution characteristics of the total oocyte count were evaluated for both groups using the Shapiro–Wilk test. As the assumption of normality was not met in Group 1 (W = 0.9271, p = 0.0163) or Group 2 (W = 0.9132, p = 0.0012), the non-parametric Mann–Whitney U test was used for between-group comparisons. Figure 3 compares the total oocyte number between the two groups. The median (IQR) [min–max] total oocyte count was 33 (27–41) [20–69] in Group 1 and 30 (26–40) [15–66] in Group 2. No statistically significant difference was observed between the groups regarding total oocyte count (U = 902, p = 0.581). Fig. 3The total oocyte count The total oocyte count Figure 4 demonstrates the MII oocyte count in both groups. The median (IQR) [min–max] MII count was 22 (15–29) [5–58] in Group 1 and 23 (18–30) [8–57] in Group 2. The Mann–Whitney U test showed no statistically significant difference in MII counts between the groups (U = 888, two-tailed p = 0.5045). The difference between medians was 1, and the Hodges–Lehmann estimate was also 1. Fig. 4The number of metaphase 2 oocytes The number of metaphase 2 oocytes The distribution of 2PN counts was first assessed for normality. According to the Shapiro–Wilk test, the distribution was normal in Group 1 (W = 0.9703, p = 0.4007), whereas the normality assumption was not met in Group 2 (W = 0.8633, p < 0.0001). The median (IQR) [min–max] 2PN count was 19.5 (12.75–27) [4–43] in Group 1 and 19.0 (15–22) [7–55] in Group 2. The Mann–Whitney U test showed no statistically significant difference in 2PN counts between the groups (U = 947.5, p = 0.8606). The difference between medians was − 0.5, while the Hodges–Lehmann estimate of the median difference was 0.0. These findings indicate that 2PN counts were similar between the two groups (Fig. 5). Fig. 5Comparison of 2PN embryo counts between two groups Comparison of 2PN embryo counts between two groups Figure 6 compares the number of blastocysts on days 5–6 between the two groups. The distribution was non-normal in both groups (Group 1: W = 0.9278, p = 0.0171; Group 2: W = 0.9122, p = 0.0011). The median (IQR) [min–max] number of day 5–6 blastocysts was 7.0 (5–11.5) [3–18] in Group 1 and 8.0 (5–11) [1–27] in Group 2. According to the Mann–Whitney U test, no statistically significant difference was observed between the groups (U = 967.5, p = 0.9917). The difference between medians was 1.0, while the Hodges–Lehmann estimate of the median difference was 0.0. These findings indicate no meaningful difference between the two groups in day 5–6 blastocyst development. Fig. 6Comparison of day 5–6 blastocyst count between two groups Comparison of day 5–6 blastocyst count between two groups The distribution of Grade A blastocyst counts did not meet the normality assumption (Group 1: W = 0.8797, p = 0.0007; Group 2: W = 0.8922, p = 0.0002). The median (IQR) [min–max] of Grade A blastocysts was 4.0 (3–10) [0–14] in Group 1 and 5.0 (4–8) [0–23] in Group 2. According to the Mann–Whitney U test, no statistically significant difference was observed between the groups regarding Grade A blastocyst counts (U = 838, p = 0.2776). The difference between medians was 1.0, and the Hodges–Lehmann estimate of the central tendency difference was also 1.0. Grade B blastocyst counts did not follow a normal distribution in either group (Group 1: W = 0.7263, p < 0.0001; Group 2: W = 0.8509, p < 0.0001). The median (IQR) [min–max] of Grade B blastocysts was 2.0 (1–4) [0–17] in Group 1 and 1.0 (0–3) [0–9] in Group 2. No statistically significant difference was observed between the groups regarding Grade B blastocyst count (U = 855, p = 0.3388; r = 0.10; Hodges–Lehmann difference: 0, 95% CI: −1 to 0). Within-group comparisons showed that the number of Grade A blastocysts was significantly higher than the number of Grade B blastocysts in both Group 1 (Wilcoxon matched-pairs signed-rank test, p = 0.0052; median difference = − 2.5) and Group 2 (p < 0.0001; median difference = − 4.0), indicating that blastocyst quality was maintained in both groups regardless of when the antagonist was started. Table 3 summarizes the comparison of total oocyte count, MII, 2PN, day 5–6 blastocyst count, and blastocyst grades between the groups. Table 3Comparison of oocyte and embryo development parameters between groupsVariableGroup 1Group 2p valueEffect size (r)95% CITotal Oocyte Number33 (27–41) [20–69]30 (26–40) [15–66]0.5810.06−1 (− 6 to 3)MII Oocyte Number22 (15–29) [5–58]23 (18–30) [8–57]0.50450.071 (− 3 to 5)2PN19.5 (12.75–27) [4–43]19.0 (15–22) [7–55]0.86060.020 (− 5 to 4)Day 5–6 Blastocyst Number7.0 (5–11.5) [3–18]8.0 (5–11) [1–27]0.99170.0010 (− 2 to 2)Grade-A Blastocyst Number4.0 (3–10) [0–14]5.0 (4–8) [0–23]0.27760.121 (− 1 to 2)Grade-B Blastocyst Number2.0 (1–4) [0–17]1.0 (0–3) [0–9]0.33880.100 (− 1 to 0)Between-group comparisons were performed using the two-tailed Mann–Whitney U testA p value < 0.05 was considered statistically significant. Effect sizes were calculated as r = |Z|/√N (N = 89); values of 0.1, 0.3, and 0.5 indicate small, medium, and large effects, respectively. 95% CI: Hodges–Lehmann confidence interval of the median difference (Group 2 − Group 1)IQR interquartile range, MII metaphase II, 2PN two-pronuclear embryos*Data are presented as median (IQR) [minimum–maximum] Table 3Comparison of oocyte and embryo development parameters between groupsVariableGroup 1Group 2p valueEffect size (r)95% CITotal Oocyte Number33 (27–41) [20–69]30 (26–40) [15–66]0.5810.06−1 (− 6 to 3)MII Oocyte Number22 (15–29) [5–58]23 (18–30) [8–57]0.50450.071 (− 3 to 5)2PN19.5 (12.75–27) [4–43]19.0 (15–22) [7–55]0.86060.020 (− 5 to 4)Day 5–6 Blastocyst Number7.0 (5–11.5) [3–18]8.0 (5–11) [1–27]0.99170.0010 (− 2 to 2)Grade-A Blastocyst Number4.0 (3–10) [0–14]5.0 (4–8) [0–23]0.27760.121 (− 1 to 2)Grade-B Blastocyst Number2.0 (1–4) [0–17]1.0 (0–3) [0–9]0.33880.100 (− 1 to 0)Between-group comparisons were performed using the two-tailed Mann–Whitney U testA p value < 0.05 was considered statistically significant. Effect sizes were calculated as r = |Z|/√N (N = 89); values of 0.1, 0.3, and 0.5 indicate small, medium, and large effects, respectively. 95% CI: Hodges–Lehmann confidence interval of the median difference (Group 2 − Group 1)IQR interquartile range, MII metaphase II, 2PN two-pronuclear embryos*Data are presented as median (IQR) [minimum–maximum] Comparison of oocyte and embryo development parameters between groups Between-group comparisons were performed using the two-tailed Mann–Whitney U test A p value < 0.05 was considered statistically significant. Effect sizes were calculated as r = |Z|/√N (N = 89); values of 0.1, 0.3, and 0.5 indicate small, medium, and large effects, respectively. 95% CI: Hodges–Lehmann confidence interval of the median difference (Group 2 − Group 1) IQR interquartile range, MII metaphase II, 2PN two-pronuclear embryos *Data are presented as median (IQR) [minimum–maximum] Reproductive outcomes following frozen-thawed embryo transfer cycles were comparable between groups (Table 4). Clinical pregnancy rates were 86.8% (33/38) in Group 1 and 86.3% (44/51) in Group 2 (p = 1.00). Miscarriage rates per clinical pregnancy were 18.2% (6/33) in Group 1 and 15.9% (7/44) in Group 2 (p = 1.00). Live birth rates per patient were 57.9% (22/38) in Group 1 and 60.8% (31/51) in Group 2 (p = 0.829). Fisher’s exact test results (all p > 0.8) and Cohen’s h effect sizes (all < 0.07) indicated no statistically significant differences between the groups in any reproductive outcome. Table 4Reproductive outcomes between groupsOutcomeGroup 1 (n = 38)Group 2 (n = 51)p valueCohen’s hClinical pregnancy, n (%)33 (86.8)44 (86.3)1.00< 0.07Miscarriage, n (%)*6 (18.2)7 (15.9)1.00< 0.07Live birth, n (%)†22 (57.9)31 (60.8)0.829< 0.07Ongoing pregnancy, n (%)†5 (13.2)6 (11.8) - - Cumulative success rate, n (%)†‡27 (71.1)37 (72.5) - - *Miscarriage rates are calculated based on the number of clinical pregnancies (Group 1: n = 33; Group 2: n = 44)†Live birth, ongoing pregnancy, and cumulative success rates are calculated based on the total number of patients in each group‡Cumulative success rate = live birth + ongoing pregnancy. Between-group comparisons were performed using Fisher’s exact test. Effect sizes were calculated using Cohen’s h, where values of 0.2, 0.5, and 0.8 indicate small, medium, and large effects, respectively Table 4Reproductive outcomes between groupsOutcomeGroup 1 (n = 38)Group 2 (n = 51)p valueCohen’s hClinical pregnancy, n (%)33 (86.8)44 (86.3)1.00< 0.07Miscarriage, n (%)*6 (18.2)7 (15.9)1.00< 0.07Live birth, n (%)†22 (57.9)31 (60.8)0.829< 0.07Ongoing pregnancy, n (%)†5 (13.2)6 (11.8) - - Cumulative success rate, n (%)†‡27 (71.1)37 (72.5) - - *Miscarriage rates are calculated based on the number of clinical pregnancies (Group 1: n = 33; Group 2: n = 44)†Live birth, ongoing pregnancy, and cumulative success rates are calculated based on the total number of patients in each group‡Cumulative success rate = live birth + ongoing pregnancy. Between-group comparisons were performed using Fisher’s exact test. Effect sizes were calculated using Cohen’s h, where values of 0.2, 0.5, and 0.8 indicate small, medium, and large effects, respectively Reproductive outcomes between groups *Miscarriage rates are calculated based on the number of clinical pregnancies (Group 1: n = 33; Group 2: n = 44) †Live birth, ongoing pregnancy, and cumulative success rates are calculated based on the total number of patients in each group ‡Cumulative success rate = live birth + ongoing pregnancy. Between-group comparisons were performed using Fisher’s exact test. Effect sizes were calculated using Cohen’s h, where values of 0.2, 0.5, and 0.8 indicate small, medium, and large effects, respectively The relationship between E2 (pg/mL) levels and oocyte/embryo development parameters was evaluated using Spearman’s rank correlation analysis, considering the distributional characteristics and measurement levels of the variables. Table 5 presents the correlation between E2 level and total oocyte count, MII number, 2PN number, day 5–6 blastocyst number and blastocyst grades. All analyses were conducted two-tailed, and a total of 89 observation pairs were included. A statistically significant, moderate positive correlation was found between E2 levels and total oocyte count (r number = 0.360, 95% CI: 0.158–0.533; p = 0.0005). Similarly, E2 levels showed a weak-to-moderate, statistically significant positive correlation with the number of MII oocytes (r = 0.231, 95% CI: 0.018–0.425; p = 0.029). A weak but statistically significant positive correlation was also observed between E2 levels and the number of 2PN embryos (r = 0.229, 95% CI: 0.016–0.423; p = 0.031). In contrast, no statistically significant relationship was found between E2 levels and the number of day 5–6 blastocysts (r = 0.191, 95% CI: −0.025–0.389; p = 0.073). Likewise, E2 levels were not significantly correlated with the number of Grade-A blastocyst (r = 0.187, 95% CI: −0.029–0.386; p = 0.080). No significant correlation was observed between E2 levels and Grade-B blastocyst count (r = − 0.017, 95% CI: −0.230–0.198; p = 0.875). Table 5Association between serum estradiol (E₂) level at GnRH antagonist initiation and oocyte/embryo development parametersVariableSpearman r%95 CIp value aTotal Oocyte Number0.36010.1580 to 0.5332 0.0005*** MII Oocyte Number0.23130.01799 to 0.4245 0.0292* 2PN0.22910.01569 to 0.4226 0.0308* Day 5–6 Blastocyst Number0.1907-0.02456 to 0.38900.0735Grade-A Blastocyst Number0.1867-0.02865 to 0.38550.0798Grade-B Blastocyst Number-0.01690-0.2303 to 0.19800.8751Data are presented as correlation coefficient (r) with 95% confidence intervals (CI)A p value < 0.05 was considered statistically significant. p < 0.05; *** p < 0.001MII Metaphase II, 2PN Two-pronuclear embryosaCorrelations were analyzed using Spearman’s rank correlation coefficient (two-tailed) Table 5Association between serum estradiol (E₂) level at GnRH antagonist initiation and oocyte/embryo development parametersVariableSpearman r%95 CIp value aTotal Oocyte Number0.36010.1580 to 0.5332 0.0005*** MII Oocyte Number0.23130.01799 to 0.4245 0.0292* 2PN0.22910.01569 to 0.4226 0.0308* Day 5–6 Blastocyst Number0.1907-0.02456 to 0.38900.0735Grade-A Blastocyst Number0.1867-0.02865 to 0.38550.0798Grade-B Blastocyst Number-0.01690-0.2303 to 0.19800.8751Data are presented as correlation coefficient (r) with 95% confidence intervals (CI)A p value < 0.05 was considered statistically significant. p < 0.05; *** p < 0.001MII Metaphase II, 2PN Two-pronuclear embryosaCorrelations were analyzed using Spearman’s rank correlation coefficient (two-tailed) Association between serum estradiol (E₂) level at GnRH antagonist initiation and oocyte/embryo development parameters Data are presented as correlation coefficient (r) with 95% confidence intervals (CI) A p value < 0.05 was considered statistically significant. p < 0.05; *** p < 0.001 MII Metaphase II, 2PN Two-pronuclear embryos aCorrelations were analyzed using Spearman’s rank correlation coefficient (two-tailed) Discussion There are two main protocols used to prevent premature ovulation in patients with PCOS; either initiation of a GnRH antagonist (using either a fixed or flexible protocol) or the administration of an oral progestin. In literature, there are numerous studies comparing these two protocols in terms of premature ovulation and including good-quality blastocyst outcomes, with no significant differences reported [5, 6]. In a comparative study by Xiao et al. 157 patients diagnosed with PCOS were divided into two groups according to the COH protocol: one group received the GnRH antagonist protocol, and the other group received the progestin-primed ovarian stimulation (PPOS) protocol. There were no statistically significant differences between the two protocols in terms of fertilization rate, cleavage-stage embryo rate, or the number of good-quality blastocysts. Even more important, there were no significant differences regarding clinical pregnancy rates, implantation rates, or ongoing pregnancy rates [5]. In another study, Huang et al. compared patients who received either PPOS or GnRH antagonist protocol in terms of premature ovulation and clinical results and reported similar results by Xiao et al. In the study, total oocyte number, MII oocyte rate, 2PN and blastocyst rate were comparable between two groups [6] and there were no differences between two groups according to the clinical pregnancy, ongoing pregnancy rate and live birth rates [6]. Premature LH surge was not observed in either group [6]. Ghosh Dastidar et al., compared a novel method, a day-1 antagonist protocol in which the GnRH antagonist was initiated on the first day of stimulation, with flexible day5/6 antagonist protocol in patients with PCOS and reported significantly improved laboratory outcomes in day-1 antagonist group [4]. In the day-1 antagonist group, the total oocyte number and the number of top-quality embryos were found to be significantly higher, however the cumulative clinical pregnancy rate per cycle and the pregnancy rate per embryo transfer did not show a statistically significant difference [4]. In the literature, no other published study exists about day-1 antagonist protocol; however, there are a few articles in which the initiation of GnRH antagonist is based on serum LH levels. There are only a limited number of studies focusing on patients in whom GnRH antagonist treatment is initiated during the late phase of ovarian stimulation. Hamdine et al. conducted a multicenter study comparing IVF outcomes between patients who were administered GnRH antagonist injections on the second day of ovarian stimulation and those who initiated treatment on the sixth day of stimulation. Total oocyte count, fertilization rate, embryo number, and embryo quality were comparable. Although the difference was not statistically significant, live birth rates and cumulative live birth rates were higher in the group in which GnRH antagonist was initiated on day 6 or later. In addition, no difference was observed in the incidence of ovarian hyperstimulation syndrome (OHSS) between the groups [7]. In a different study, the flexible GnRH antagonist protocol was compared with a novel flexible protocol adjusted according to serum LH levels during ovarian stimulation in PCOS patients. Serum LH levels were measured during COS, when LH was ≤ 4 IU/L, GnRH antagonist treatment was withheld. When LH exceeded 4 IU/L, GnRH antagonist therapy was initiated, with dosing adjusted according to LH concentration, GnRH antagonist therapy was started and varying doses depending on the LH concentration was administered [8]. Similarly, Jing Wang et al. published a study and evaluated the effects of LH levels on IVF outcomes in patients with PCOS who were administered GnRH antagonist protocol [9]. The patients were classified into three groups regarding the basal LH (bLH) levels, LH levels on hCG trigger day (hLH) and the hLH/bLH ratio. There was no significant difference in cumulative pregnancy rates among the three groups [9]. In a study with Tannus et al., the implantation rate, clinical pregnancy and live birth rates were compared between two groups, Group 1: patients with PCOS who received GnRH antagonist injection before day-6 of the ovarian stimulation and Group 2; patients with PCOS who were administered GnRH antagonist after day-6 of the ovarian stimulation [10]. Implantation, clinical pregnancy, and ongoing pregnancy rates were higher at group 2, although these differences were not statistically significant [10]. The effect of high basal LH levels and LH elevations during COH cycle on oocyte and embryo quality remains controversial however, recent studies suggest that neither elevated basal LH nor excessively increased LH levels during COH adversely affect oocyte or embryo quality [11–14]. Additionally, studies have suggested that elevated LH levels during COH may not have a linear relationship with premature ovulation [14]. In conclusion, it is considered that neither elevated basal LH levels nor increases in LH levels during COH adversely affect oocyte or embryo quality. At the start of the antagonist protocol, serum E₂ levels were significantly higher in both groups, exceeding the usual flexible protocol threshold of 300 pg/mL. This finding reflects the typical rapid follicular growth and strong ovarian response seen in women with PCOS. In these patients, E₂ levels often exceed standard thresholds before the antagonist is used. Importantly, the high E₂ levels observed in both groups did not negatively affect embryo development. This suggests that increased estradiol exposure at the start of the antagonist protocol might be a natural part of managing PCOS, rather than a factor that can be changed. These conclusions are further supported by the correlation analysis. Group 1 had significantly higher serum E₂ levels at antagonist initiation, but E₂ only moderately positively correlated with total oocyte yield and weakly correlated with MII oocyte and 2PN counts. Crucially, neither Grade-A nor Grade-B blastocysts demonstrated a significant correlation with E₂, nor were E₂ levels significantly correlated with blastocyst number or quality. In accordance with the inter-group analyses, the observed dissociation between E₂ concentrations and blastocyst outcomes suggests that the markedly elevated estradiol exposure observed in the delayed initiation cohort did not adversely affect embryonic development. These findings support the hypothesis that the quality of oocytes or embryos in patients with PCOS is not diminished by delaying antagonist initiation until a minimum of three follicles attain a diameter of ≥ 17 mm. On the other hand, another concern in patients with PCOS is premature ovulation [14, 15]. Both GnRH antagonist protocol and PPOS are effective in preventing premature LH surge and premature ovulation. Jabarpour et al., compared GnRH antagonist protocol with PPOS and reported that LH levels were slightly higher in the patients treated with the GnRH antagonist protocol. However, this difference was not statistically significant, and both protocols were reported to be successful in preventing an LH surge [15]. To the best of our knowledge, this is the first study to compare outcomes when GnRH antagonist initiation is delayed until ≥ 3 follicles reach ≥ 17 mm in PCOS patients undergoing IVF. In this study, GnRH antagonist treatment was initiated in patients who had at least three follicles measuring 17 mm or greater, and these patients were compared with those treated using the standard flexible protocol. Our results are consistent with previous studies showing that delayed initiation of GnRH antagonist treatment does not negatively affect IVF success, particularly, MII oocyte rate, blastocyst rate and good quality blastocyst rate. Additionally, no premature ovulation was observed in either group. Moreover, delayed start of GnRH antagonist approach is more cost-effective compared with the standard flexible protocol. With this study, we are taking the next logical step and raising the question of whether the antagonist protocol should be revised specifically for patients diagnosed with PCOS. By modifying the flexible protocol, we question whether the initiation of GnRH antagonist treatment in PCOS patients could be shifted from a dominant follicle diameter of ≥ 14 mm to ≥ 17 mm. Thus, these findings suggest that the necessity and timing of antagonist administration in carefully monitored PCOS patients may warrant further investigation. One of the primary limitations of our study is its retrospective design and the relatively small number of patients. The second limitation is that LH and progesterone levels were not measured during follow-up. Routine monitoring of LH and progesterone levels during stimulation is not a standard procedure at our institution, and the retrospective nature of the study prevented the acquisition of these data. Consequently, the assertion that premature ovulation did not transpire is grounded in clinical and embryological observations, rather than biochemical confirmation. Additionally, Anti-Müllerian Hormone (AMH) was not regularly evaluated at our center, especially for cycles preceding 2020, which restricted its use as a covariate in this retrospective analysis. AMH is a good predictor of oocyte yield, but many studies have shown that it does not predict fertilization rate, blastocyst formation, or live birth rate on its own when age and oocyte number are considered. Thus, not having it is unlikely to have had a significant effect on the main embryological outcomes reported here. Nevertheless, subsequent prospective studies ought to include AMH as a fundamental variable. Although this study spans eight years (2017–2025), the core laboratory procedures, including oocyte denudation, ICSI, embryo culture, and blastocyst grading, were performed at a single center with consistent methodology throughout the study period, minimizing the potential impact of temporal variation on embryological outcomes. A further limitation is that OHSS incidence following oocyte pick-up could not be systematically monitored. OHSS is a clinically important outcome in PCOS studies, as this patient population is at inherently elevated risk due to high ovarian response. Our clinic receives patients from all regions of the country as well as from neighboring countries; therefore, patients return to their hometowns after stimulation, making it impossible to reliably record the frequency and severity of OHSS. Notably, the freeze-all strategy applied in both groups may partially mitigate this limitation, as embryo cryopreservation eliminates the additional OHSS risk associated with fresh embryo transfer. Nevertheless, the absence of systematic OHSS data should be considered when interpreting the safety profile of delayed antagonist initiation. The last limitation is the non-randomized group allocation. Patients in Group 1 were not intentionally assigned to delayed initiation as part of a predefined protocol; rather, delays arose from logistical factors inherent to serving a geographically diverse patient population traveling from across Türkiye and neighboring countries. Importantly, Group 1 represents a highly selected subset of approximately 30,000 cycles screened over 9 years, only those in whom at least three follicles had already reached ≥ 17 mm at the delayed visit and in whom the treating physician independently approved continuation of stimulation were included. This stringent selection, while introducing observational bias, also suggests that these patients were carefully evaluated rather than passively neglected. It is important to emphasize that delayed antagonist initiation in Group 1 did not represent a pre-planned clinical protocol; rather, it was an unintended consequence of missed follow-up visits in a high-volume center serving a geographically dispersed patient population. In none of these cases was the delay a deliberate clinical decision at the outset of stimulation. This real-world scenario may reflect the conditions under which such clinical decisions are routinely made in high-volume IVF centers. Conclusion In conclusion, within this specific PCOS cohort, the delayed initiation of GnRH antagonists defined as initiation when a minimum of three follicles had a diameter of ≥ 17 mm, was not associated with diminished oocyte yield, inadequate embryo quality, or an elevated risk of premature ovulation. Furthermore, no statistically significant differences were detected between groups in terms of pregnancy, miscarriage and live birth rates. Besides, the significant reduction in total treatment duration and the duration of GnRH antagonist injections may have beneficial effect both in terms of cost effectiveness and the reduced number of injections applied during treatment. These findings suggest that COH can be sustained in this patient demographic without adversely affecting laboratory outcomes. Notably, this observation emerged from a real-world clinical scenario rather than a predefined protocol, lending ecological validity to the findings. Nonetheless, these results have to be considered preliminary because of the study’s limited sample size and retrospective design. Prospective randomized controlled trials are needed to confirm these findings before this approach can be widely recommended in clinical practice.