This retrospective analysis included 1,177 infertile patients who underwent IVF/ICSI cycles and fresh embryo transfer from May 2017 to December 2022. The women were monitored until live birth after fresh embryo transfer. The study protocol was approved by the institutional human ethics committee.

The inclusion criteria were: (1) flexible antagonist protocol with fresh embryo transfer cycle, (2) basal follicle-stimulating hormone (bFSH) ≤ 15 IU/L, and (3) body mass index (BMI) <40kg/m².

The exclusion criteria were: (1) patients with chromosomal abnormalities, reproductive malformation, and a history of recurrent spontaneous abortion; (2) patients undergoing coasting to prevent ovarian hyperstimulation syndrome; (3) incomplete information; (4) uterine pathologies that might compromise pregnancy potential; (5) patients with elevated progesterone or LH level before the addition of GnRH-ant.

The premature LH surge refers to an endogenous LH peak that occurs before follicle maturation or human chorionic gonadotrophin (hCG) injection. It was defined as either maximum LH levels exceeding two and a half times the baseline LH level on day 2 of the same menstrual cycle (13), or the absolute value > 10 IU/mL (32–34).

Patients received IVF/ICSI treatment according to the flexible GnRH antagonist protocol. On the second day of the menstrual cycle, recombinant human follicle-stimulating hormone 150–300 U (Gonal-F; Merck, Lyon, France; Puregon, MSD, Boulogne, France) was injected as Gn. The Gn doses were determined based on patient age, body mass index (BMI), FSH, and AFC. A daily dose of 0.25/0.125 mg of GnRH-ant (Cetrotide, Merck, Lyon, France) was administered when the leading follicle size was 14 mm or the serum E2 level reached 300pg/mL or a premature luteinizing hormone surge was recognized.

Oocytes were then collected by follicular aspiration under ultrasound 34–36 h after triggering with GnRH-a (Triptoreline, Decapeptyl, Ipsen, France) or recombinant hCG (rhCG, Ovitrelle, Merck, Lyon, France). IVF or ICSI fertilization was selected according to semen conditions. Eighteen hours after fertilization, embryo development was monitored daily and graded based on the number and size of blastomeres, fragmentation rate, multinucleation, and early densification. High-quality embryos were defined as embryos with seven to ten blastomeres on the third day following oocyte retrieval (35).

All patients completed an IVF/ICSI cycle and then performed an ET cycle. On day 3 or day 5, one or two of the best embryos or blastocysts were selected and transferred using a soft Wallace catheter. In order to avoid multiple pregnancies and reduce the incidence of complications during pregnancy and the perinatal period, the number of embryos was selected according to the specific conditions of the patient and the embryo quality score, and one blastocyst was selected as far as possible. For luteal support in preparation for the fresh embryo transfer, 40mg of progesterone (20 mg/branch, Zhejiang Xianju Pharmaceutical Co., Ltd.) was injected daily and 30mg of oral dydrogesterone tablets (10 mg/tablet, Abbott Healthcare Products B.V.) or progesterone vaginal sustained-release gel (90 mg/dose, Crinone VR 8%, Merck, Sherano, Switzerland) were given daily. In addition, two bags of Chinese medicine, the Gushen Antai pills, were used daily.

Venous blood samples were collected on the second day of the menstrual cycle to the triggering day. Serum FSH, LH, E_2, and P levels were measured by means of the automated Elecsys Immunoanalyzer (Beckmann, America).

The live birth rate (LBR) was the primary outcome. The secondary outcomes included biochemical pregnancy rate (BPR), clinical pregnancy rate (CPR), number of oocytes, number of fertilization, number of blastomeres, number of embryos, and number of high-quality embryos/blastomeres.

Live birth was defined as live birth per embryo transfer in a fresh cycle.

Clinical pregnancy was defined as the presence of an intrauterine gestational sac with a fetal heartbeat detected on transvaginal ultrasonography after six weeks of gestation.

Biochemical pregnancy was defined as the presence of detectable ß-hCG in the blood and the value<10mIU/mL at approximately 14 days after transplantation.

LBR = number of cycles with live birth/number of transfer cycles × 100%

BPR = number of cycles with biochemical pregnancy/number of transfer cycles × 100%

CPR = number of cycles with clinical pregnancy/number of transfer cycles × 100%

Statistical analysis was conducted using SPSS version 26 and R4.2.1. In this study, P<0.05 was considered statistically significant. The Shapiro–Wilk test was used to assess data normality. Due to skewed distributions, quantitative variables were expressed as medians (interquartile range, range between the 25th and 75th percentiles), and the Mann–Whitney U test and Kruskal-Wallis test were performed. Qualitative variables were expressed as frequencies and analyzed using the chi-square test. Moreover, correlation analysis was performed using Spearman based on the R language. Binary logistic regression analysis was performed based on the following patient characteristics: female age, BMI, AFC, Gn dosage, GnRH-ant dosage, endometrial thickness, number of embryo transfers, high-quality embryos, blastocysts, and group PMLH. The crude odds ratios (OR) and adjusted OR were calculated with 95% confidence intervals (CI). In addition, curve fitting and threshold effect analysis were performed on group MLH0. Smooth curve fitting was carried out with the gam model to identify any non-linear relationship between the P levels and LBR. A piece-wise linear regression method was used to analyze the threshold effect between the P levels and LBR.

In total, 1177 patients who received fresh embryo transfer with antagonist regimens from May 2017 to December 2022 were retrospectively analyzed. As displayed in Table 1 , there were 27 cases in group P1MLH1, 76 cases in group P1MLH0, 135 cases in group P0MLH1, and 939 cases in group P0MLH0. As illustrated in Table 1 , the age of participants in group P0MLH1 (35(31,39)) was significantly higher than group P0MLH0 (34(31,38), P=0.049*). The BMI of P0MLH0 (23.4(21.3,26.7)) was significantly higher than group P0MLH1 (23(20.75,25.3), P=0.048*). In addition, the AFC of group P1MLH1 24(18.5,28) was significantly higher than group P0MLH1 (16(8,24), P=0.001*) and group P0MLH0 (17(12,24), P=0.011*). The GnRH-ant dosage of group P1MLH0 (1(0.625,1.25)) was the highest (P=0.049*). The baseline FSH of group P0MLH1 (7.55(6.36,9.22)) was the highest (P=0.025*). The baseline LH of group P1MLH1 (5.01(3.38,6.15)) was the highest (P=0.025*). Furthermore, group P1MLH1 showed a significantly higher number of oocytes (11(8,15.5)) than group P0MLH1 (7(3,10.5), P=0.000*) and group P0MLH0 (8(5,11), P=0.007*). Oocyte numbers of group P1MLH0 (9(6,12.5)) was also higher than group P0MLH1 (7(3,10.5), P=0.001*) and group P0MLH0 (8(5,11), P=0.044*). The fertilization numbers of P1MLH1 (6(3.5,9), P=0.025*) and group P1MLH0 (5(4,8), P=0.004*) were higher than group P0MLH1 (4(2,7)). Embryo numbers of group P1MLH1(4(2,6)) were higher than group P0MLH1 (2(1,4), P=0.026*). The blastocyst number of group P1MLH1 (0(0,2)) was the highest (P=0.003*).

The relationship between each group PMLH and pregnancy outcomes was analyzed and compared. As displayed in Table 2 , BPR (P=0.034*), CPR (P=0.028*), and LBR (P=0.036*) were significantly different across the PMLH subgroups. In the pairwise comparison of group PMLH, the comparison between group P1MLH0 and group P0MLH0 showed a statistically significant difference (BPR: P=0.011*; CPR: P=0.005*; LBR: P=0.011*). The pairwise comparisons between other subgroups were no statistically significant.

Figure 1 displays the correlation analysis results between the indicators. A significant negative association was found between the maximum LH level and BMI (R=-0.13, P=0.000*), GnRH-ant days (R=-0.07, P=0.018*), GnRH-ant dosage (R=-0.11, P=0.000*), and the outcome of COH, including number of fertilization (NOF) (R=-0.08, P=0.005*), number of embryos (NOE) (R=-0.07, P=0.023*), number of high-quality embryo (NOH) (R=-0.07, P=0.012*). In contrast, a significant positive association was observed between the maximum LH level and female age (R=0.08, P=0.004*). In addition, the same trend was observed between the maximum LH and P levels on the hCG day, indicating a positive correlation (R=0.08, P=0.005*). The P level exhibited a negative correlation with BMI (R=-0.1, P=0.000*) and age (R=-0.11, P=0.000*), while demonstrating a positive association with AFC (R=0.18, P=0.000*) and COH outcome, including number of oocytes (NOC) (R=0.3, P=0.000*), NOF (R=0.23, P=0.000*), NOE (R=0.15, P=0.000*), number of blastocyst (NOB) (R=0.11, P=0.000*), NOH (R=0.09, P=0.002*).

Based on the correlation analysis results, the patients were further subgrouped by age and BMI. The participants were divided into quartiles of age ≤ 30, 30<age ≤ 34, 34<age ≤ 38, and 38<age. Moreover, the patients were subgrouped by BMI into BMI<18.5, 18.5 ≤BMI<24, and 24≤BMI based on guidelines for the prevention and management of overweight and obesity in the adult Chinese population (2004). High and low ovarian responses were defined as >18 oocytes retrieved and <6 oocytes retrieved, respectively. As shown in Figure 2 , higher maximum LH levels were associated with older age (P=0.006*), lower BMI (P=0.000*), and decreased ovarian response (P=0.046*). However, elevated P levels on hCG day were associated with younger individuals (P=0.001*), participants with lower BMI (P=0.006*), and higher ovarian response (P=0.000*), as illustrated in Figure 3 .

Furthermore, the association between transient premature LH surge and the above subgroups was explored. The statistically significant indicators are shown in Figures 4 , 5 . The patients with transiently premature LH surges had older age (P=0.045*), lower BMI (P=0.014*), lower GnRH-ant dosage (P=0.020*), and higher P levels (P=0.011*) on the hCG day. However, the increased P levels were mainly related to patients’ AFC (P=0.004*). Higher P levels were associated with a greater number of oocytes (P=0.000*), fertilization (P=0.002*), and number of embryos (P=0.046*).

Binary logistic regression analysis revealed that group P1MLH0 (OR =0.432 [0.218–0.854]; P =0.016*) had opposite influences on the live birth rate compared with group P0MLH0, as indicated in Table 3 . The same results were obtained in normal ovarian response (OR =0.49 [0.25–0.94]; P =0.03*), as indicated in Table 4 .

Figure 6 shows the fitted curves after adjustment for confounders between the P level on the hCG day and the live birth rate without transient premature LH surge. The results yielded a parabolic reverse-U curve as the P level increased. The curve initially increased before declining at the highest P level on the hCG day.

The results of the threshold effect analysis of the P levels and LBR are presented in Table 5 . The live birth rate showed a significant decrease when the P level was ≥1.1ng/ml.