Comparing the clinical pregnancy outcomes of fresh embryo transfer in young patients with poor prognosis using different ovarian stimulation protocols Authors: Nan Jia, Ruolei Wang, Qiuyuan Li, Haoying Hao, Siyue Xu, Lei Jin, Yundong Mao, Juanzi Shi, Cuilian Zhang, Xiaoyan Liang, Shaodi Zhang Published: 2026 ABSTRACT AbstractObjectiveTo evaluate the clinical pregnancy outcomes of fresh embryo transfer at the cleavage stage utilizing gonadotropin-releasing hormone agonist (GnRH-a) and the antagonist protocols in young patients with poor prognosis.MethodsA retrospective cohort study was carried out on patients who underwent in vitro fertilization-embryo transfer (IVF-ET) between January 2016 and July 2024. Women under 35 years old with anti-Müllerian hormone (AMH) levels < 1.2 ng/ml in fresh cycles were included. A total of 1,732 cycles using the GnRH-a protocol and 3,168 cycles utilizing the antagonist protocol were investigated. The two groups were compared based on their baseline characteristics, clinical and embryological laboratory parameters, and outcome indicators. To compare the clinical results of the two stimulation protocols, we employed propensity score matching (PSM, 1:1) and multivariate logistic regression.ResultsAfter PSM (1:1) with a tolerance of 0.01, the number of cycles in the two groups was 1422 each.The GnRH-a group had a higher clinical pregnancy rate (55.01%, P = 0.003) and live birth rate (41.66%, P < 0.001) than the antagonist group (clinical pregnancy rate: 47.67%; live birth rate: 31.74%). Ectopic pregnancy, early miscarriage, preterm birth, and the number of live births did not differ significantly (P > 0.05). Multivariate logistic regression analysis revealed that the clinical pregnancy rate in the GnRH-a group (OR = 1.25,95%CI:1.02, 1.52) was 1.25 times higher than that in the antagonist group (P = 0.0297), and the live birth rate(OR = 1.38,95%CI:1.12, 1.71) was 1.38 times higher (P = 0.0025). Additionally, compared to the antagonist group (transfer cancellation rate: 45.71%; total embryo freezing: 31.58%), the GnRH-a group exhibited reduced rates of fresh cycle transfer cancellation (34.67%; P < 0.001) and total embryo freezing (22.01%; P < 0.001). The rates of oocyte retrieval cancellation, oocyte retrieval failure, and the lack of usable embryos did not differ significantly between the two groups (P > 0.05).ConclusionIn young patients with poor prognosis, the GnRH-a protocol for controlled ovarian stimulation in fresh embryo transfer cycles produces higher clinical pregnancy and live birth rates, as well as a lower fresh transfer cancellation rate, as compared to the antagonist protocol.Keywords: IVF-ET, Controlled ovarian stimulation, Young patients with poor prognosis, Clinical pregnancy rate, Live birth rate Background In the process of assisted reproduction with IVF-ET, an increasing number of guidelines and expert consensuses have selected a single, healthy live birth as the best treatment objective. Optimizing clinical outcomes for patients with poor ovarian response continues to be a major problem. Poor ovarian response is caused by multiple factors, and the affected population is diverse. Currently, our understanding of its pathophysiology is limited [1]. The POSEIDON group (Patient-Oriented Strategies Encompassing Individualized Oocyte Number) presented a new classification system in 2016 for patients with low ovarian reserve or poor response to exogenous gonadotropins [2, 3].Women under 35 years old with low ovarian reserve (AFC < 5, AMH < 1.2 ng/mL) are considered young patients with poor prognosis (POSEIDON group 3), which may have higher cycle cancellation rates, lower pregnancy rates per embryo transfer and cumulative pregnancy rates per initiated cycle than normal responders. As a result, establishing the best ovarian stimulation protocol to optimize clinical outcomes in this group is especially important. In 2020, the ESHRE and the Reproductive Endocrinology Guideline Group announced updated guidelines [4], recommending that both the GnRH-a protocol and the antagonist protocol can be utilized for patients with poor ovarian response. The latest 2026 guidelines [5] have not issued clear recommendations or denials for the application of these two protocols in patients with poor ovarian response. The antagonist protocol has been widely used due to its advantages, which include shorter stimulation duration, lower gonadotropin dosages, and reduced risk of ovarian hyperstimulation syndrome (OHSS) [6]. However, because of its effects on follicular synchronization and endometrial receptivity [7], studies have shown that in patients with normal ovarian response, the live birth rate per fresh embryo transfer cycle using the antagonist protocol is lower than that with the long-acting GnRH-a protocol [8]. In individuals with poor ovarian response, a study [9] has demonstrated that the GnRH-a protocol can reduce cycle cancellation rates, improve transferable embryo availability, and increase clinical pregnancy rates. Consequently, the GnRH-a protocol may serve as a promising ovarian stimulation option for poor ovarian responders, especially among young patients with low ovarian reserve [10]. However, other researchers [11] have reported that adopting a delayed-start antagonist protocol can result in higher oocyte retrieval rates, MII oocyte rates, and clinical pregnancy rates compared to the GnRH-a protocol. Thus, the relative benefits and drawbacks of the GnRH-a and antagonist protocols remain debatable and warrant more exploration. Some researchers [12, 13] have claimed that AMH levels are related to ovarian stimulation outcomes such as oocyte yield and high-quality embryo rate, but cannot accurately predict clinical results. Patients with low AMH levels may yet have a high clinical pregnancy and live birth rates. We collected data from IVF-ET cycles at five large reproductive institutions in China, which included female patients under 35 years old with AMH < 1.2 ng/mL, according to the POSEIDON classification. In this study, we compared fresh embryo transfer outcomes following GnRH-a and antagonist protocols, with the intention of providing actionable clinical evidence to optimize ovarian stimulation plans for young patients with poor prognosis. Methods Study population This retrospective cohort study analyzed patients who received intracytoplasmic sperm injection (ICSI) or in vitro fertilization (IVF) at Henan Provincial People’s Hospital between January 2016 and July 2024, as well as at four other Chinese reproductive facilities between January 2016 and July 2020: Tongji Hospital, Tongji Medical College of Huazhong University of Science and Technology; The Sixth Affiliated Hospital of Sun Yat-sen University; Northwest Women’s and Children’s Hospital; and Jiangsu Provincial People’s Hospital. Included were female patients under 35 years with AMH levels < 1.2 ng/mL who received ovarian stimulation in fresh cycles using either a GnRH-a protocol or an antagonist protocol. This study exclusively incorporated cycles involving day 3 cleavage-stage embryo transfer following oocyte retrieval. Exclusion criteria Donor sperm, donor oocyte, or oocyte-freezing cycles; Untreated hydrosalpinx or intrauterine adhesions; Uterine anomalies that were either acquired or congenital, such as septate, arcuate, bicornuate, unicornuate, and didelphys uteruses; Repeated implantation failure, recurrent miscarriage, or adverse pregnancy history; Either partner’s chromosomal abnormalities or cycles of preimplantation genetic testing (PGT); Missing key information or cycles with no live birth results. The Institutional Review Board and Ethics Committees of the collaborating centers examined and approved this study: Henan Provincial People’s Hospital (Approval No:2024 Ethical Review NO.184, SYSZLL-2019110401); Tongji Hospital, Tongji Medical College of Huazhong University of Science and Technology (TJ-IRB 20210320); Jiangsu Provincial People’s Hospital (2020-SR-046); The Sixth Affiliated Hospital of Sun Yat-sen University (2020ZSLYEC-295); and Northwest Women’s and Children’s Hospital (2019013). The ethical committees waived the need for individual informed permission because the study was retrospective. To preserve personal information, all data were anonymised prior to analysis. The study adhered to the Declaration of Helsinki’s tenets. Ovarian stimulation Antagonist Protocol: Gonadotropin (Gn) was given at a dose of 200–300 U/day on days 2–3 of the menstrual cycle to start ovarian stimulation, with prompt adjustments based on hormone levels and follicular development. The antagonist (Cetrorelix acetate, 0.25 mg/vial, Merck Serono, Germany) was administered when the dominant follicle diameter reached ≥ 12 mm and continued until the trigger day (human chorionic gonadotropin, hCG 6,000–10,000 U). 33–36 h later, oocyte retrieval was performed under transvaginal ultrasound supervision. GnRH-a Protocol: This comprised the long-acting protocol in the early follicular phase and the short-acting protocol in the mid-luteal phase. In the former, 3.75 mg of a long-acting GnRH agonist (Triptorelin/Diphereline, Ipsen Biotech, France; Leuprorelin/Beiyi, Lizhu Pharmaceutical Co., Shanghai, China) was given on menstruation days 2–4, and 28–30 days later, transvaginal ultrasound and hormone level monitoring were performed. In the latter, 0.1 mg of a short-acting GnRH agonist (Decapeptyl, Ferring Pharmaceuticals, Germany; Jinlei, Changchun Jinsai Pharmaceutical Co., China) was given once daily for 14–18 days beginning in the mid-luteal phase. Gn was administered to initiate stimulation based on the patient’s unique situation once downregulation requirements (FSH < 5 IU/L, LH < 5 IU/L, E₂ <50 ng/L, and follicle diameter < 5 mm) were satisfied. Based on serum hormone levels and follicular development, the Gn dosage was maintained or modified during stimulation. Oocyte retrieval was performed 36–37 h after intramuscular administration of 6,000–10,000 U of hCG on the trigger day. Embryo grading and transfer Depending on the patient’s condition, 1–2 cleavage-stage embryos were transferred on the third day following oocyte retrieval if there was no tendency for OHSS, no infection, and no other contraindications for fresh embryo transfer in the cycle, and if the endometrial thickness was ≥ 8 mm with a uniform echogenic pattern. An embryo with 7–9 uniformly sized blastomeres, regular morphology, homogenous cytoplasm, and a fragmentation rate of less than 10% was considered high-quality [14]. If at least one of the two embryos transferred was of high quality, the cycle was categorized as belonging to the high-quality embryo group. On the day of oocyte retrieval, luteal phase support started with 10 mg of oral dydrogesterone tablets (Duphaston, Abbott Laboratories, USA) taken twice a day along with 90 mg of 8% progesterone vaginal sustained-release gel (Crinone, Merck Serono, Germany). After embryo transfer, the luteal support dosage remained unchanged. 14 days after transfer, serum β-hCG was tested; luteal support was continued if β-hCG was ≥ 50 U/L. The drug was gradually lowered starting at 8 weeks of gestation and stopped at 10 weeks. A transvaginal ultrasound was carried out 28 days after embryo transfer. Clinical pregnancy was defined as the existence of a gestational sac either inside or outside the uterus. Early miscarriage was defined as a miscarriage that occurred before 12 weeks of gestation. A live birth was defined as a delivery at or after 28 weeks of gestation of a newborn exhibiting one of the following signs of life: breathing, heartbeat, umbilical cord pulsation, or voluntary muscle movement. A live birth cycle was defined as a cycle that resulted in a live birth. Observation indicators AMH was determined using the magnetic particle chemiluminescence method and a SMART 6500 fully automatic magnetic particle chemiluminescence immunoassay analyzer. The detection range for the comparable kit (Guangzhou Kangrun Biotechnology Co., Ltd.) was 0.01-30 ng/mL. The correlation coefficient (r) needed to be at least 0.9900, the intra-assay coefficient of variation (CV) < 8.0%, the inter-assay CV ≤ 10.0%, and the limit of detection (LOD) ≤ 0.01 ng/mL. In order to measure baseline hormones and AMH levels, fasting blood samples were taken from patients undergoing assisted reproductive treatment on days 2–4 of the menstrual cycle. Additionally, transvaginal ultrasound examination was performed to count the AFC, as detailed in the literature [15]. Age, body mass index (BMI), basal follicle-stimulating hormone (FSH), AMH, AFC, duration of infertility, type of infertility, sperm retrieval and fertilization methods, etc. were among the patients’ baseline indicators. Total embryo freezing cycles ( included OHSS, thin endometrium, increased serum progesterone concentrations, and personal preferences, etc.), canceled oocyte retrieval cycles, cycles with no recovered oocytes, and cycles with no usable embryos (i.e., cycles in which oocyte retrieval was carried out and oocytes were acquired, but ultimately no usable embryos were available) were all considered cancelled transfer cycles. The main outcome indicators were clinical pregnancy rate and live birth rate, and the supplementary outcome indicators included early miscarriage rate, ectopic pregnancy rate, and preterm birth rate. Clinical pregnancy rate = the number of clinical pregnancy cycles /the number of embryo transfer cycles × 100%; Preterm birth rate = the number of preterm birth (gestational age < 37 weeks) cycles / the number of live birth cycles × 100%; Live birth rate = the number of live birth cycles /the number of embryo transfer cycles × 100%. Statistical analysis Statistics are presented using means ± standard deviation, median (25th and 75th percentiles) [M (Q1, Q3)] or number (%), depending on the situation.The independent sample t-test, Mann-Whitney nonparametric test, χ² test, Fisher’s exact test or Multivariate logistic regression analysis were used to compare the groups. Because this was a non-interventional trial, PSM (1:1) was carried out with a matching tolerance of 0.01 to account for potential confounding factors related to exposure and intended outcomes. Standardized mean differences(SMD) were calculated to evaluate the balance of confounders between treatment groups before and after matching. SMD values ranging from − 0.1 to 0.1 indicated well-balanced treatment groups. All analyses were conducted using EmpowerStats, an R-based program.The threshold for statistical significance was fixed at P < 0.05. Results Study population This study contained a total of 4,900 oocyte retrieval cycles, with 3,168 in the antagonist protocol group (1,593 transfer cycles, accounting for 50.28%) and 1,732 in the GnRH-a protocol group (1,118 transfer cycles, accounting for 64.55%). To avoid the influence of baseline differences, PSM was calculated using female age, AMH, AFC, and baseline FSH. The matching ratio was 1:1, with a tolerance of 0.01. Prior to PSM, the two groups differed in several of baseline factors. After matching, the two groups had 1422 cycles each, and the standardized differences for all variables were less than 0.1, indicating very minor differences (Table 1). Table 1Baseline characteristics before and after PSMOvarian Stimulation Protocolbefore matchingafter matchingAntagonist ProtocolGnRH-a ProtocolStandardize diff.P-valueAntagonist ProtocolGnRH-a ProtocolStandardize diff.P-valueNumber of cycles3168173214221422Female age (years)30.52 ± 2.7130.35 ± 2.850.060.09730.47 ± 2.7730.47 ± 2.7400.951AMH (ng/ml)0.72 ± 0.310.86 ± 0.270.49< 0.0010.84 ± 0.260.84 ± 0.270.010.889AFC5.00 (3.00–7.00)7.00 (5.00–10.00)0.72< 0.0017.18 ± 3.307.34 ± 3.200.050.166Baseline FSH (mIU/ml)8.56 (6.76–11.15)7.55 (6.19–9.33)0.37< 0.0018.40 ± 3.538.25 ± 3.380.040.234Male age (years)31.97 ± 3.8231.87 ± 4.280.030.3932.03 ± 4.0531.99 ± 4.200.010.813Female BMI (kg/m2)22.49 ± 3.4622.51 ± 3.510.010.85922.70 ± 3.5422.45 ± 3.490.070.067Male BMI (kg/m2)24.98 ± 3.5625.13 ± 3.700.040.29125.05 ± 3.5425.16 ± 3.840.030.536Duration of infertility (years)3.00 (2.00–5.00)3.00 (2.00–5.00)0.030.3083.00 (2.00–5.00)3.00 (2.00–5.00)0.030.401Type of infertility (%)0.080.0070.070.062Primary infertility1947 (61.81%)1001 (57.89%)865 (61.04%)818 (57.61%)Secondary infertility1203 (38.19%)728 (42.11%)552 (38.96%)602 (42.39%)Method of sperm retrieval (%)0.040.1830.010.855Ejaculation2869 (95.95%)1580 (95.12%)1293 (95.07%)1296 (95.22%)Surgical retrieval121 (4.05%)81 (4.88%)67 (4.93%)65 (4.78%)Fertilization method (%)0.010.7290.040.332IVF2231 (74.82%)1238 (74.35%)982 (71.99%)1006 (73.65%)ICSI751 (25.18%)427 (25.65%)382 (28.01%)360 (26.35%) Baseline characteristics before and after PSM Fresh cycle embryo transfer cancellation Before and after PSM, the GnRH-a protocol group had a significantly reduced cancellation rate (before: 35.45%; after: 34.67%) for fresh cycle embryo transfer compared to the antagonist protocol group (before: 49.72%; after: 45.71%) (P < 0.001). Following matching, the GnRH-a protocol group had a lower total embryo freezing rate (22.01%) than the antagonist protocol group(31.58%); nevertheless, the rates of cancelation before oocyte retrieval, no oocytes retrieved, and no useable embryos did not differ (P > 0.05). Refer to Table 2 for more details. Table 2Fresh cycle embryo transfer cancellation before and after PSMNumber of cyclesbefore matchingafter matchingAntagonist ProtocolGnRH-a ProtocolStandardize diff.P-valueAntagonist ProtocolGnRH-a ProtocolStandardize diff.P-valueTransfer cancellation (%)1575 (49.72%)614 (35.45%)0.29< 0.001650 (45.71%)493 (34.67%)0.23< 0.001Cancellation before oocyte retrieval (%)86 (2.71%)37 (2.14%)0.040.21629 (2.04%)32 (2.25%)0.010.698No oocytes retrieved (%)60 (1.89%)18 (1.04%)0.070.02213 (0.91%)15 (1.05%)0.010.704No usable embryos (%)470 (14.84%)156 (9.01%)0.18< 0.001159 (11.18%)133 (9.35%)0.060.108Total embryo freezing (%)959 (30.27%)403 (23.27%)0.16< 0.001449 (31.58%)313 (22.01%)0.22< 0.001 Fresh cycle embryo transfer cancellation before and after PSM Treatment-related characteristics Numerical variations in some treatment-related features persisted between the two cohorts despite matching being done to reduce the differences. Clinical indicators The antagonist protocol group had lower E₂ levels on HCG-day (median 1105.50), total Gn dosage (2426.01 ± 837.74), Gn duration (9.01 ± 1.90) and endometrial thickness on trigger day (10.52 ± 2.44) compared to the GnRH-a protocol group (HCG-day E₂ median: 1303.50; total Gn dosage median: 2913.44 ± 918.50; Gn duration: 11.02 ± 2.32, endometrial thickness on trigger day: 11.35 ± 2.68), with statistically significant differences (P < 0.05). Laboratory indicators The antagonist protocol group retrieved fewer oocytes (median: 5.00) and had a lower proportion of cycles with two embryos transferred (58.94%) compared to the GnRH-a protocol group (total oocytes retrieved median: 6.00; proportion of cycles with two embryos transferred: 69.00%) (P < 0.05). The number of available D3 embryos and embryo grading were not significantly different between the two groups (P > 0.05). See Table 3 for more information. Table 3Treatment-related characteristics before and after propensity-score matchingbefore matchingafter matchingAntagonist ProtocolGnRH-a ProtocolStandardize diff.P-valueAntagonist ProtocolGnRH-a ProtocolStandardize diff.P-valueNumber of cycles3168173214221422E₂ on HCG day (pg/ml)965.40 (562.50–1485.00)1322.00 (821.00-2001.00)0.42 (0.36, 0.48)< 0.0011105.50 (706.10-1685.50)1303.50 (804.20-1964.25)0.20 (0.13, 0.28)< 0.001Endometrial thickness on trigger day (mm)10.46 ± 2.5111.36 ± 2.730.34 (0.28, 0.40)< 0.00110.52 ± 2.4411.35 ± 2.680.33 (0.25, 0.40)< 0.001Total Gn dosage (IU)2406.96 ± 870.712874.71 ± 930.790.52 (0.46, 0.58)< 0.0012426.01 ± 837.74)2913.44 ± 918.500.55 (0.48, 0.63)< 0.001Gn duration (days)9.05 ± 1.9311.01 ± 2.340.91 (0.85, 0.98)< 0.0019.01 ± 1.9011.02 ± 2.320.95(0.87,1.03)< 0.001Oocytes retrieved4.00 (2.00–6.00)6.00 (4.00–9.00)0.55 (0.49, 0.61)< 0.0015.00 (3.00–7.00)6.00 (4.00–8.00)0.25 (0.18, 0.33)< 0.001Available D3 embryos2.00 (1.00–4.00)3.00 (2.00–5.00)0.34 (0.28, 0.40)< 0.0013.00 (2.00–4.00)3.00 (2.00–5.00)0.10 (0.02, 0.17)0.06Number of embryos transferred0.27< 0.0010.21< 0.0011695 (43.63%)344 (30.77%)317 (41.06%)288 (31.00%)2898 (56.37%)774 (69.23%)455 (58.94%)641 (69.00%)Embryo grading (%)0.010.7410.020.657High-quality embryos977 (77.23%)824 (77.81%)482 (78.63%)685 (77.66%)Non-high-quality embryos288 (22.77%)235 (22.19%)131 (21.37%)197 (22.34%) Treatment-related characteristics before and after propensity-score matching Clinical pregnancy outcomes Table 4 shows matched data, indicating that the clinical pregnancy rate (55.01%) and live birth rate (41.66%) in the GnRH-a protocol group were higher than those in the antagonist protocol group (clinical pregnancy rate: 47.67%; live birth rate: 31.74%), with statistically significant differences (P < 0.05). There were no significant differences between the two groups in terms of ectopic pregnancy, early miscarriage, preterm birth, or number of live births (P > 0.05). Table 4Clinical Outcomes in the PSM CohortOvarian Stimulation ProtocolAntagonist ProtocolGnRH-a ProtocolStandardize diff.P-valueNumber of cycles772929Clinical pregnancy rate368 (47.67%)511 (55.01%)0.150.003Ectopic pregnancy rate4 (1.09%)12 (2.35%)0.100.168Early miscarriage rate39 (10.60%)65 (12.72%)0.070.336Live birth rate245 (31.74%)387 (41.66%)0.21< 0.001Preterm birth rate40 (16.33%)68 (17.57%)0.030.685Number of live births0.000.9931190 (77.55%)300 (77.52%)255 (22.45%)87 (22.48%) Clinical Outcomes in the PSM Cohort Female age, AMH, AFC, baseline FSH, and so on were used as adjustment variables in multivariate logistic regression analysis. The results showed that the GnRH-a protocol group had a higher clinical pregnancy rate (OR = 1.25, 95%CI:1.02, 1.52; P = 0.0297) and live birth rate (OR = 1.38, 95%CI:1.12, 1.71; P = 0.0025) compared to the antagonist protocol group (Table 5). Table 5Multivariable logistic regression of Clinical Outcomes in the PSM CohortNon-adjustedAdjustOR(95%CI)P-valueOR(95%CI)P-valueClinical pregnancyAntagonist protocolreferencereferenceGnRH-a protocol1.34 (1.11, 1.63)0.00261.25 (1.02, 1.52)0.0297Live birthAntagonist protocolreferencereferenceGnRH-a protocol1.54 (1.26, 1.88)< 0.00011.38 (1.12, 1.71)0.0025Adjusted variables: female age, AMH, AFC, FSH, male age, duration of infertility, fertilization method, endometrial thickness on trigger day, number of embryos transferred Multivariable logistic regression of Clinical Outcomes in the PSM Cohort Adjusted variables: female age, AMH, AFC, FSH, male age, duration of infertility, fertilization method, endometrial thickness on trigger day, number of embryos transferred Fitted curves of clinical outcomes Female age was considered as a continuous variable, with fitted curves for clinical pregnancy rate and live birth rate (Figs. 1 and 2). These curves further demonstrated that the GnRH-a protocol was associated with higher clinical pregnancy and live birth rates. Fitted curve of clinical pregnancy rate Fitted Curve of live birth Rate The antagonist protocol was shown by the red curve in the picture, whereas the GnRH-a protocol was represented by the blue curve. Discussion In this multicenter, large-sample retrospective analysis of 4,900 cycles in young patients with poor prognosis, we evaluated the clinical efficacy of the antagonist protocol and the GnRH-a protocol in IVF-ET cycles, with clinical pregnancy rate and live birth rate in fresh embryo transfer cycles serving as the main outcome indicators. Due to the inherent limitations of retrospective studies, clinicians may have been more inclined to choose the GnRH-a protocol for patients with relatively better ovarian reserve. As a result, the two groups exhibited distinct baseline markers. To address this, baseline characteristics were balanced using PSM (1:1 with a matching tolerance of 0.01). Both before and after matching, the results showed that the clinical pregnancy rate and live birth rate in the GnRH-a protocol group were significantly higher than those in the antagonist protocol group, while the early miscarriage rate, ectopic pregnancy rate, preterm birth rate, and number of live births did not differ significantly. A Multivariate logistic regression analysis after matching also revealed that adopting the GnRH-a protocol resulted in higher clinical pregnancy and live birth rates among young patients with poor prognosis. AFC and AMH are widely known for their high sensitivity and specificity in predicting poor ovarian response [16]. The menstrual cycle has no effect on AMH measurement, and the correlation between AMH levels and the number of oocytes retrieved remains stable even if the sampling and stimulation are separated by up to a year [17], showing its relative stability and convenience. Thus, some researchers [18] believe that AMH can be a useful biomarker for predicting poor ovarian response in IVF cycles, outperforming AFC and FSH. In addition, the accuracy of AFC counting may be influenced by multiple factors such as training and workflow variances among reproductive centers, ultrasonography equipment resolution, and operator experience.For this reason, we included only AMH as the classification indicator of ovarian reserve function, namely, patients under 35 years old with AMH < 1.2 ng/mL. In the final data, the number of cycles in the antagonist protocol group (3,168) was larger than that in the GnRH-a protocol group (1,732), indicating that the antagonist protocol was more frequently chosen for ovarian stimulation in young patients with poor prognosis. Moreover, the fresh cycle transfer cancellation rate was higher in the antagonist protocol group compared to the GnRH-a protocol group before and after matching. Based on the matched data, we can conclude that the main variation in the rate of transplantation cancellation between the two groups was related to the rate of total embryo freezing, which inevitably increased treatment duration and cost for patients. The risk of OHSS following fresh embryo transfer is quite low in young patients with poor prognosis since the number of oocytes extracted and estradiol levels on the trigger day are not high. Furthermore, the GnRH-a protocol had shown better clinical pregnancy and live birth rates in fresh cycles. Taken together, it was suggested that the GnRH-a protocol for fresh embryo transfer may confer better cost-effectiveness in young patients with poor prognosis. Since the development of GnRH agonists and antagonists in the 1980s and 1990s, they have been widely applied in assisted reproductive technology (ART). The GnRH-a and antagonist protocols remain two of the most important approaches for controlled ovarian stimulation. Numerous studies have compared these two protocols in various patient populations, examining clinical pregnancy rates, live birth rates, and safety outcomes; nevertheless, the results have been inconsistent. Huang et al. [10] compared the efficacy of GnRH-a and antagonist protocols in fresh embryo transfer cycles among patients with low AMH (< 1.2 ng/mL) across different age groups. They found that the GnRH-a protocol was associated with a lower transfer cancellation rate, higher implantation rate, and higher live birth rate in patients under 35 years of age, while the outcomes were similar between the two protocols in patients aged 35 years and above. Similarly, Li et al. [19] found that, when compared to the antagonist and mild stimulation protocols, the long-acting GnRH-a protocol improved cumulative clinical pregnancy and cumulative live birth rates in POSEIDON group 3 patients while decreasing cycle cancellation rates and abnormal endometrial rates. These findings are consistent with our results, which suggest that in young patients with poor prognosis, the GnRH-a protocol improves clinical outcomes over the antagonist protocol.This could be attributed to factors such as asynchronous follicular development, decreased MII oocyte rate, and poor endometrial receptivity after antagonist stimulation [7, 20]. Recent studies [21, 22] have demonstrated that in patients with normal ovarian reserve, the antagonist protocol can minimize the incidence of OHSS compared to the long GnRH-a protocol, without reducing the live birth rate. Similarly, Zhu et al. [23] reported similar results in a research that did not differentiate patients based on ovarian reserve function. A 2017 meta-analysis [6] included 26 studies on patients with normal ovarian function and 6 studies on poor ovarian responders found that, in women with normal ovarian function, the antagonist protocol was associated with significantly lower ongoing pregnancy rates and a reduced incidence of OHSS compared with the long-acting agonist protocol. Among patients with diminished ovarian reserve, pregnancy rates were similar between the antagonist and long GnRH-a protocols (RR: 0.87, 95% CI: 0.65–1.17). Therefore, researchers have proposed that the antagonist protocol can be considered the standard treatment for patients with decreased ovarian reserve. However, investigations evaluating the clinical effects of the two ovarian stimulation protocols in young patients with poor prognosis remain inconclusive. Two trials published in 2025, one involving 158 cycles [24] and the other involving 153 cycles [25] of poor ovarian responders, found that the clinical pregnancy rate, ongoing pregnancy rate, and early miscarriage rate were similar between the GnRH-a and antagonist protocols. Nonetheless, the antagonist protocol was associated with shorter stimulation duration and reduced pharmaceutical administration, resulting in greater cost-effectiveness. These research’ findings differ from ours for several reasons: Differences in stimulation protocol classification: In our study, both the long-acting protocol initiated in the early follicular phase and the short-acting protocol initiated in the luteal phase were classified as the GnRH-a protocol, which was then compared to the flexible antagonist protocol. 2) Different definitions of poor ovarian response or young poor prognosis: Previous studies used the Bologna criteria or POSEIDON categorization inconsistently. Given the variety and unpredictability of AFC measurements among centers in our multicenter trial, we defined the study population solely on the basis of female age and AMH levels. 3)Differences in sample size and methodological control: With a relatively large sample size in this multicenter study, we adopted PSM and multivariate regression analysis to control for confounding factors, which may help strengthen the objectivity and stability of our findings. Differences in stimulation protocol classification: In our study, both the long-acting protocol initiated in the early follicular phase and the short-acting protocol initiated in the luteal phase were classified as the GnRH-a protocol, which was then compared to the flexible antagonist protocol. 2) Different definitions of poor ovarian response or young poor prognosis: Previous studies used the Bologna criteria or POSEIDON categorization inconsistently. Given the variety and unpredictability of AFC measurements among centers in our multicenter trial, we defined the study population solely on the basis of female age and AMH levels. 3)Differences in sample size and methodological control: With a relatively large sample size in this multicenter study, we adopted PSM and multivariate regression analysis to control for confounding factors, which may help strengthen the objectivity and stability of our findings. The strengths of the present study include its multicenter design and relatively large sample size. Clinical pregnancy rate and live birth rate of fresh embryo transfer cycles were adopted as the primary outcome measures. Furthermore, PSM was conducted to balance key baseline characteristics, and multiple statistical approaches were applied to improve the stability and credibility of our analytical results. Nevertheless, several limitations inherent to the multi-center retrospective design should be acknowledged in this study. First, the patient enrollment period varied marginally across participating centers. Second, relevant data regarding cumulative live birth rate and time to live birth were not available in our database. These two indicators are essential for more comprehensive assessment of the efficacy of different ovarian stimulation protocols. In view of the above deficiencies, further well-designed studies with larger sample sizes and long-term follow-up are warranted, to incorporate key endpoints including cumulative live birth rate and time to live delivery. In addition, prospective randomized controlled trials are still required to further validate and compare the clinical performance of different ovarian stimulation strategies among young patients with suboptimal prognostic features. Conclusions In summary, for patients under 35 years old with AMH ≤ 1.2 ng/mL, the GnRH-a protocol yielded a higher clinical pregnancy rate, live birth rate, and lower cycle cancellation rate compared to the antagonist protocol. Collectively, the GnRH-a protocol appears to be a viable option for fresh embryo transfer in this clinical subgroup. Acknowledgements The authors express their gratitude to all of the employees at the five Reproductive Medicine Centers for their assistance. Abbreviations a Gonadotropin-releasing hormone agonist in Vitro fertilization-embryo transfer anti-Müllerian hormone Propensity score matching European Society of Human Reproduction and Embryology Patient-Oriented Strategies Encompassing Individualized Oocyte Number Antral follicle count Ovarian hyperstimulation syndrome Intracytoplasmic sperm injection In vitro fertilization Preimplantation genetic testing Gonadotropin Standardized mean differences human chorionic gonadotropin Body mass index Basal follicle-stimulating hormone Authors’ contributions SD Z, CL Z and XY L supervised the entire study, including procedures, conception, design, and conclusion, as well as participated in paper edits. LJ, JZ S, YD M, and NJ were in charge of collecting data. QY L, HY H and SY X were responsible for searching the literature and organizing concepts. NJ and RL W contributed to the data analysis and wrote the manuscript. All authors contributed to the article and approved the version that was submitted. Funding This study was funded by Fertility Research Program of Young and Middle-aged Physicians. IN 2024 (BJHPA-2024-SHZHYXZHQNYJ-010). Data availability All data supporting the findings of this study are available within the paper. Declarations Ethics approval and consent to participate The ethics committees of Henan Provincial People’s Hospital, Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology, Jiangsu Provincial Hospital, Northwest Women’s and Children’s Hospital, and Sixth Affiliated Hospital of Sun Yat-sen University evaluated and approved the studies involving human subjects. All participating couples provided written informed consent for ART treatment. The study conformed to the principles of the Declaration of Helsinki. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Nan Jia and Ruolei Wang share first authorship and have made equal contributions to this work. Contributor Information Cuilian Zhang, Email: luckyzcl@qq.com. Xiaoyan Liang, Email: liangxy2@mail.sysu.edu.cn. Shaodi Zhang, Email: zhangshd@126.com. References Associated Data Data Availability Statement All data supporting the findings of this study are available within the paper.