The serum oestradiol/progesterone ratio on the day of OPU + 7, but not the day of OPU + 5, affects the rates of live birth in fresh blastocyst embryo transfer cycles

We conducted a hospital-based cohort study. This investigation was performed in accordance with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of Renmin Hospital, Hubei Medical University. Anonymous data were collected from the Reproductive Medicine Centre, Renmin Hospital, Hubei University of Medicine, between January 2016 and April 2021.

Patients who received the early-follicle-phase depot gonadotropin-releasing hormone agonist (GnRH-a) protocol were included. Patients were chosen if they satisfied all the following inclusion criteria: regular menstrual cycles ranging from 25 to 35 days; aged < 40 years; body mass index (BMI), 18–28 kg/m²; normal basal serum follicle-stimulating hormone (FSH) (< 10 mIU/ml) and anti-Müllerian hormone (AMH) (≥ 1.1 ng/ml) levels determined on days 2–3 of the cycle prior to controlled ovarian hyperstimulation (COH), and blastocyst transfer. The following were the exclusion criteria: patients with metabolic disorders, ovulatory dysfunction, pelvic tuberculosis, congenital uterine malformations, chromosomal abnormalities or single-gene disorders, cardiovascular diseases, or tumors. The patients were stratified into three groups based on E₂/P ratio tertiles on OPU + 7, namely, the low group (1.3–15.7 pg/ng), middle group (15.7–28.8 pg/ng), and high group (28.8–487.2 pg/ng). Followed-up was performed by communicating with the women by telephone until the pregnancy outcomes were known.

The patients received a single intramuscular injection of 3.75 mg long-acting triptorelin acetate (Decapeptyl; Ferring, SaintPrex, Switzerland) on day 2 or 3 of the cycle. After 30–42 days of downregulation, an ultrasound scan and serum concentration tests were performed, and the criteria were as follows: endometrial thickness ≤ 5 mm; follicles 5–7 mm; serum concentration of E₂ < 50 pg/ml; P < 1 ng/ml; and LH < 1 mIU/ml. Recombinant LH (Luveris; Merck Serono) (75 IU per day) was added in the mid- and late-follicular stages to promote follicular development when the serum LH level was below 1.2 mIU/ml. Then, the treatment followed by gonadotropin (Gn) stimulation, the doses of urinary human menopausal gonadotropin (HMG, Livzon Pharmaceutical, China) and recombinant FSH (Gonal-f, Merck Serono, Germany) were adjusted according to the growth trend of the follicles and serum hormone changes (150–450 IU per day). Recombinant hCG (Merck Serono, Italy) at a dose of 250 µg and urinary hCG (Livzon Pharmaceutical, China) at a dose of 1,000–2,000 IU were given to trigger oocyte maturation when two or more follicles reached preovulatory size (18–22 mm). We chose the trigger medication when multiple follicles were greater than 16 mm in size and according to the E₂ levels as follows: when there were more than 10 follicles and the E₂ level was greater than 2500 pg/ml, recombinant hCG was used alone; when there were more than 10 follicles or the E₂ level was less than 2500 pg/ml, both recombinant hCG and urinary hCG 1,000 IU were used; when there were less than 10 follicles or the E₂ level was less than 2500 pg/ml, both recombinant hCG and urinary hCG 2,000 IU were administered. Oocyte retrieval was done 36 h following the trigger. According to the standard insemination procedures used in the laboratory, all oocytes were inseminated using IVF or ICSI. Embryo scoring was conducted based on morphologic criteria; 6–8 cells with less than 20% fragmentation were considered to be good-quality embryos. On the fifth day after oocyte pick-up, ET was carried out with a soft catheter under transabdominal ultrasound guidance.

After oocyte retrieval, luteal-phase support was initiated and continued daily until 3 months of gestation with the daily application of 90 mg vaginal progesterone gel (Crinone; Merck Serono) and either 10 mg twice or three times daily oral dydrogesterone (Duphaston, Abbott, USA), 2 mg twice daily oestradiol valerate tablets (Progynova, Berlin, Germany), or 1 mg:10 mg daily vaginal oestradiol and dydrogesterone tablets (Femoston, Abbott, USA). The good-quality spare embryos were cryopreserved through a vitrification protocol. Fresh ET cancellation and freeze-all strategies were implemented in cases of high P concentrations on hCG day (> 2 ng/ml) or to prevent ovarian hyperstimulation syndrome (OHSS). We chose the luteal support medication according to the P and E₂ levels on OPU + 2: when the E₂ level was greater than 1000 pg/ml and the P level was greater than 100 ng/ml, Crinone was used alone; when the E₂ level was greater than 1000 pg/ml and the P level was 50–100 ng/ml, both Crinone and dydrogesterone were used; when the E₂ level was less than 1000 pg/ml and the P level was 50–100 ng/ml, Crinone and dydrogesterone plus Progynova were used; and when the E₂ level was less than 500 pg/ml and the P level was less than 50 ng/ml, Crinone, dydrogesterone, Progynova and Femoston were used.

We measured serum P and E₂ levels, which represent luteal function, on OPU + 2, OPU + 5 and OPU + 7 using commercially available automated electrochemiluminescence immunoassays (UniCel® DxI 800 Access Immunoassy System, Beckman Coulter, USA and Access® Progesterone Calibrators, Access® Sensitive Estradiol Assay, Beckman Coulter, USA). Skilled technicians carried out all measurements in accordance with the manufacturer's instructions. P had a detection threshold of 0.1 ng/ml, and the in-house inter- and intra-assay coefficients of variation were 10 and 10%, respectively. E₂ had a detection limit of 15.0 pg/ml, and the in-house inter- and intra-assay coefficients of variation were 10 and 10%, respectively.

The outcome measures for patients with clinical and nonclinical pregnancy are presented first. Based on raw data on E₂ and P levels throughout the early and mid-luteal stages individually, the E₂/P ratio groups were identified.

In this study, the primary outcome was the LBR. The secondary outcomes were moderate or severe OHSS, hCG positivity, clinical pregnancy, ectopic pregnancy, pregnancy loss and preterm birth rates. Moderate or severe OHSS was diagnosed in women who fulfilled more than one of the following criteria: clinical ascites, hydrothorax, or dyspnoea (exertional or at rest) [9]. Biochemical pregnancy was defined as hCG > 10 mIU/ml 14 days after ET. Clinical pregnancy was defined as an intrauterine gestational sac identified by ultrasonography 30 days after ET. Early pregnancy loss was defined as spontaneous pregnancy loss before 12 weeks. Live birth was considered when a living fetus was born after 28 weeks of pregnancy.

The statistical packages R (The R Foundation; http://​www.​r-project.​org; version 3.6.1), EmpowerStats (http://​www.​empowerstats.​com) and SPSS 22.0 (IBM, Armonk, NY, USA) were utilized for all analyses. One-way analysis of variance or the Kruskal–Wallis test was used to examine the differences among groups, and continuous variables are shown as the mean with standard deviation or the median with interquartile range. Categorical variables were quantified as percentage-based figures and compared using either the Fisher's exact test or the Pearson chi-square test. Statistical significance was accepted as a two-sided P value < 0.05. Graphs were generated by using GraphPad Prism version 8.0 (GraphPad Software).

A multivariable logistic regression analysis was performed to assess significant relationships between the E₂/P ratio on OPU + 7 and pregnancy outcomes. The variables that indicated significance in the univariate analysis at P < 0.10 or more and those that might have an influence on live birth were included in the multivariable model. The GraphPad program was used to generate a spline curve by plotting the trends between pregnancy outcomes and various hormone levels.

Smooth curve fitting models were created using EmpowerStats software and R-project (version 3.6.1) in order to further analyze the substantial relationships between the E₂/P ratio on OPU + 7 and pregnancy outcomes. A two-piecewise linear regression model was also used to assess the threshold effect of the influencing factors on live birth using a smoothing function curve. The inflection point is obtained by recursive algorithm. Additionally, the one-line linear regression model and the two-piecewise linear regression model were also compared using a log-likelihood ratio test, and odds ratios (ORs) and 95% confidence intervals (CIs) for the threshold turning points of the independent influencing factors were computed before and after. The relationship was then further examined using model observation data, and ultimately, the chance of a live birth was properly examined using data obtained before and after each independent influencing factor's threshold inflection point.

The mean (± SD) patient age in this study population was 29.6 ± 3.5 years (range 18–41). After ET, 2,257 cycles (1,879 conventional IVF and 378 ICSI) produced a total of 1,606 clinical pregnancies, for a clinical pregnancy rate of 71.2%. Stratified by diagnostic classification, 1,483 patients had pelvic and tubal diseases (65.7%), 90 had endometriosis (4.0%), 470 had male factor infertility (20.8%), and 214 had unexplained infertility (9.5%) (Table 1).

2,257 women who were utilizing the depot GnRH-a regimen and were split into two groups based on clinical pregnancy made up the eligible cohort. There were significant differences in BMI, AFC, trigger dosage, the moderate or severe OHSS rate, number of transferred embryos, E₂ and P on hCG day, P and the E₂/P ratio on OPU + 2, and P and E₂/P ratio on OPU + 7 between the two groups (P < 0.05) (Supplementary Table 1), and the difference in the E₂/P ratio on OPU + 7 was especially notable (32.3 pg/ng vs. 25.2 pg/ng, P < 0.001) (Fig. 1). There was no significant difference in female age, AMH, infertility duration, infertility type, infertility factors, fertilization method, dosage and duration of Gn, endometrial thickness on hCG day, number of oocytes retrieved, the good-quality embryo rate or luteal support medication between the two groups (P > 0.05) (Supplementary Table 1).

According to Fig. 1, the level of P in the clinical pregnancy group was lower than that in the nonclinical pregnancy group (201.9 ng/ml vs. 213.1 ng/ml, P = 0.001), and the E₂/P ratio was higher than that in the nonclinical pregnancy group on OPU + 2 (8.4 pg/ng vs. 8.0 pg/ng, P = 0.003). The levels of E₂ and P and the E₂/P ratio on OPU + 5 were all higher than those on OPU + 2. On OPU + 7, the levels of E₂ and P were lower than those on OPU + 5, but the E₂/P ratio was higher than that on OPU + 5. Our study also revealed that the E₂/P ratio on OPU + 7 in the clinical pregnancy group was higher than that in the nonclinical pregnancy group (32.3 pg/ng vs. 25.2 pg/ng), and the difference was statistically significant (P < 0.001).

The E₂/P ratio on OPU + 7 was used to classify all patients into three groups. BMI, AMH, AFC, and infertility factors among the three groups were significantly different. (P < 0.05) (Table 1). Female age, infertility duration, infertility type and fertilization method did not significantly differ among the groups. (P > 0.05) (Table 1).

The ovarian stimulation characteristics and embryological outcomes of the three groups are also presented in Table 1. There were significant differences in the dosage and duration of Gn, number of oocytes retrieved, good-quality embryo rate, blastocyst formation rate, number of transferred embryos, trigger dosage and luteal support medication among the three groups (P < 0.05). There were no significant differences in endometrial thickness on hCG day or the fertilization rate (P > 0.05).

The pregnancy outcomes, stratified into three groups by the serum E₂/P ratio tertiles on OPU + 7, are presented in Table 2. There was no significant difference in the ectopic pregnancy rate, early, mid- or late-term pregnancy loss rate, preterm birth rate or number of fetuses delivered by one-way analysis of variance or the Kruskal–Wallis test (P > 0.05), but there were significant differences in the positive hCG rate, clinical pregnancy rate and LBR (P < 0.01) (Table 2). The LBRs in the three groups were 56.4%, 61.8% and 69.1% (P < 0.001), respectively. The positive hCG rate, clinical pregnancy rate and LBR all appeared to be significantly higher in the high-ratio group than in the low-ratio group (82.5% vs. 73.4%, 76.4% vs. 66.8%, 69.1% vs. 56.4%, P < 0.001).

To accounting for potential confounders, multivariable regression analysis was used. After controlling for female age, BMI, AMH, AFC, infertility duration, infertility type, infertility factors, fertilization method, administration on trigger day, luteal support, number of transferred embryos and the moderate or severe OHSS rate, the E₂/P ratio on OPU + 7 was positively associated with positive hCG (adjusted OR = 1.01; 95% CI, 1.01–1.02; P < 0.0001), clinical pregnancy (adjusted OR = 1.01; 95% CI, 1.00–1.01; P = 0.0067) and live birth (adjusted OR = 1.01; 95% CI, 1.00–1.01; P < 0.001) (Table 3). Furthermore, there were no significant differences in the ectopic pregnancy rate, early, mid- or late-term pregnancy loss rate, preterm birth rate or number of fetuses delivered after multivariable regression analysis (P > 0.05).

A nonlinear association between the E₂/P ratio and pregnancy outcome was revealed by the adjusted smooth curve fit. Specifically, the E₂/P ratio on OPU + 7 had a positive correlation with clinical pregnancy and live birth. Further threshold impact research was needed, nevertheless, as these variables did not have a simple linear relationship (Fig. 2). Thus, threshold saturation effect analysis of the association between the E₂/P ratio on OPU + 7 and clinical pregnancy or live birth was performed. The logarithmic likelihood ratio test showed that the E₂/P ratio on OPU + 7 had a curvilinear association with clinical pregnancy and live birth and that there were two separate points (K₁ = 78.09, K₂ = 76.97) (P < 0.05). When the E₂/P ratio was < 78.09 pg/ng (K₁ < 78.09), it was positively correlated with clinical pregnancy (OR = 1.01, 95% CI: 1.00–1.02, P < 0.001). Conversely, there was no correlation between the E₂/P ratio on OPU + 7 and clinical pregnancy rate when the E₂/P ratio was > 78.09 pg/ng (K₁ > 78.09) (OR = 1.00, 95% CI: 1.00–1.01, P = 0.94). Additionally, when the E₂/P ratio was < 76.97 pg/ng (K₂ < 76.97), it was positively correlated with live birth (OR = 1.01, 95% CI: 1.01–1.02, P < 0.001). Conversely, there was no correlation between the E₂/P ratio on OPU + 7 and LBR when the E₂/P ratio was > 76.97 pg/ng (K₁ > 76.97) (OR = 1.00, 95% CI: 1.00–1.00, P = 0.89) (Supplementary Table 2).