This study was approved by the Institutional Ethics Committee of the First Hospital of Zhengzhou University.
This was a single-centre retrospective cohort study. Patients who were undergoing IVF with gonadotropin and a GnRH agonist for COH were enrolled from August 2009 to February 2017. Patients were excluded from the study if they experienced one of the following conditions: (1) moderate or severe ovarian hyper-stimulation syndrome (OHSS) during COH; (2) pregnancy complications including gestational diabetes mellitus, hypertensive disorders and thyroid diseases; (3) vanishing twin syndrome, defined as a foetus with cardiac activity plus one or more gestational sacs or a foetus without cardiac activity ; or (4) loss to follow-up or loss of core data in our electronic database (e.g., E2 level on the trigger day).
In cases where a patient has given birth more than once, only the first live birth was included in the analysis during the study period. Ultimately, 3659 patients constituted our final study cohort. According to serum E2 levels on the day of hCG administration, the patients were categorized into six groups: group 1 (serum E2 levels ≤ 1000 pg/mL, n=230), group 2 (serum E2 levels 1001-2000 pg/mL, n=524), group 3 (serum E2 levels 2001-3000 pg/mL, n=783), group 4 (serum E2 levels 3001-4000 pg/mL, n=721), group 5 (serum E2 levels 4001-5000 pg/mL, n=548), and group 6 (serum E2 levels > 5000 pg/mL, n=852) .
Controlled hyper-stimulation induction and embryo transfer
All patients received one of the following four controlled ovarian stimulation (COS) regimens, which have been described previously : gonadotropin-releasing hormone (GnRH)-agonist short protocol, GnRH-antagonist, mild stimulation and progestin-primed ovarian stimulation (PPOS). The clinician selected the appropriate protocol for each patient on an individual basis according to the patient characteristics.
Pituitary suppression was achieved by injecting triptorelin acetate (Decapeptyl® 0.1 mg [Ferring, Germany] or Diphereline® 0.1/3.75 mg [Ipsen, France]) until the serum levels of E2, follicle-stimulating hormone (FSH) and luteinizing hormone (LH) were <30 mIU/mL, <5 mIU/mL and <5 mIU/mL, respectively. COS was initiated with several types of gonadotropin, namely, FSH (Gonal-F® 75 IU[Serono, Switzerland], Fostiman® 75 IU [IBSA, Switzerland],or Puregon® 50 IU [N. V. Organon, Netherlands]) or FSH combined with LH (hMG®, Livzon, China). In general, the initial dose of gonadotropin was determined according to the individual's age, BMI, basal follicle-stimulating hormone (bFSH) level, and response to previous stimulation cycles, as well as whether polycystic ovary syndrome (PCOS) was present.
After stimulation for at least 4 days, the dose was adjusted according to the ovarian response, which was assessed using ultrasound and serum E2 measurements. When more than 3 follicles reached 17 mm, hCG was injected, and oocytes were extracted 36-37 h later.
Serum hormone measurements
bFSH was measured on days 2-4 of the menstrual cycle and prior to the start of the IVF cycle. In each cycle, LH, E2 and P levels were measured during controlled ovulation induction (once every 2-4 days at the beginning, once every day in the late follicular stage, and once on the day of hCG injection), and the gonadotropin and gonadotropin levels of each group were compared. All blood samples were obtained in the fasting state, usually between 6:30 and 7:30 am. Throughout the study, we used a personal immune analyser (Roche Cobas e411; Roche Diagnostics, Mannheim, Germany) and the same analytical methods were used to measure all hormones.
With a standardized questionnaire telephone survey, specially trained nurses in our department conducted surveys of couples during each pregnancy to collect information on pregnancy complications, pregnancy date, place of birth, delivery mode, sex of the new-born, gestational age, birth weight, and neonatal diseases. In cases where attempts to contact the couple failed, the local family planning service was contacted to collect data.
The outcomes were birth weight indicators, including absolute birth weight and LBW. LBW was defined as birthweight <2500 g. Preterm birth (PTB) was defined as delivery before 37 weeks of gestation.
All continuous variables are expressed as means ±SDs and categorical variables as frequencies or percentages. To examine significant differences among groups, the Mann-Whitney and chi-square tests were used for continuous variables and categorical variables, respectively. Multiple logistic regression models were applied to examine the relationship between serum E2 levels and neonatal birthweight. The relationship between the serum E2 level and neonatal birthweight was also explored using smoothing plots. A two-piecewise linear regression model was applied to investigate the threshold effect according to the smoothing plot. To quantify the strength of the association, unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (Cis) were estimated and reported. In the multivariate adjusted logistic regression models, the following were considered confounders: maternal age (years); sterility classification; duration of infertility (years); maternal BMI (kg/m2); basal FSH (mIU/mL); basal E2 (pg/mL); basal LH (mIU/mL); previous IVF attempts; ovarian stimulation protocol; total hMG dose (IU); duration of stimulation (days); number of oocytes retrieved; number of viable embryos; number of embryos transferred; gestational age at delivery (weeks); and neonatal sex. Interaction and stratified analyses included maternal age (years), maternal BMI, previous IVF attempts, and number of embryos transferred. Receiver operator characteristics (ROC) curves were constructed to identify the E2 threshold, and the corresponding area under the curve (AUC) was calculated.
Data were analysed with the use of the statistical packages R (The R Foundation; http://www.r-project.org; version 3.4.3) and EmpowerStats (www.empowerstats.com; X&Y Solutions, Inc., Boston, MA).