Introduction
Vitrification technology has become indispensable not only for fertility preservation and selective single-embryo transfer but also for reducing the risk of ovarian hyperstimulation syndrome in in vitro fertilization (IVF). During frozen-thawed embryo transfer (FET), many centers prefer hormone replacement therapy (HRT), as it allows for controlled exposure to exogenous steroids and precise timing of embryo transfer. In artificial FET cycles, estradiol (E2) and progesterone are sequentially administered to align embryo transfer with the implantation window. A high dose of estrogen in the initial E2 phase suppresses ovulation, after which progesterone is given daily until the scheduled transfer. Adequate E2 priming stimulates proliferation of endometrial cells, optimizes progesterone receptor expression, and enhances endometrial receptivity.
With the rising number of FET cycles, research has focused on their outcomes. Some studies have shown that FET is associated with a higher risk of large for gestational age (LGA) and post-term neonates, as well as adverse perinatal outcomes such as pregnancy-induced hypertension and postpartum hemorrhage, compared with fresh embryo transfer [1-6]. Conversely, fresh transfer has been linked to higher rates of preterm birth and small for gestational age (SGA) infants, risks that are largely mitigated in FET. The underlying mechanisms remain debated. While some evidence suggests that embryo cryopreservation alters gene epigenetic regulation, potentially affecting placentation and fetal growth, a greater number of studies indicate that elevated E2 levels in fresh IVF cycles impair trophoblastic invasion, resulting in lower birth weights and shorter gestations compared with FET [7-12]. In FET cycles, the adverse effects of supraphysiologic E2 on implantation and placentation are thought to be reduced under more controlled endocrine conditions. However, the duration and dose of E2 priming in HRT-FET cycles may influence clinical and neonatal outcomes. Sekhon et al. [13] found that the duration of E2 supplementation before progesterone in artificial cycles did not affect frozen blastocyst transfer success but was inversely associated with gestational age. Although constant or increasing estrogen doses in HRT-FET cycles do not appear to alter live birth rates [14], E2 administration for more than 28 days has been linked to lower pregnancy rates and higher miscarriage rates [15].
Oral E2 is convenient, well-tolerated, and widely used in early protocols, but first-pass hepatic metabolism can be avoided via parenteral administration. Vaginal E2 administration, first reported in 1988 [16], can also be delivered through vaginal rings that provide steady systemic absorption. This route appears to produce higher circulating E2 levels, with even greater concentrations in endometrial tissue, compared with oral administration [17]. The objective of this retrospective study was to determine whether combined oral and vaginal E2 administration affects live birth rate, singleton birth weight, and gestational age compared with oral administration alone. This analysis specifically evaluated whether higher E2 priming from vaginal E2 influences FET clinical and neonatal outcomes.
Methods
1. Study design and patient population
This retrospective study was conducted at the Assisted Reproduction Medicine Department of Shanghai General Hospital, affiliated with Shanghai Jiao Tong University School of Medicine. It included 1,005 women who underwent their first FET between January 2015 and January 2018. Data were extracted from the center’s database. Inclusion criteria were maternal age <48 years and transfer of two day 2 or day 3 cleavage-stage embryos following HRT endometrial preparation. Patients were excluded if they met any of the following conditions: (1) use of cryopreserved or donor oocytes; (2) previous IVF or FET attempts; (3) concurrent HRT with sildenafil or growth hormone for endometrial development; (4) endometrial thickness <6 mm on the day progesterone administration began; (5) presence of endometrial polyps, submucosal myomas, or endometrial separation; (6) history of ≥4 induced abortions, uterine adhesions, or uterine malformations such as Müllerian anomalies (e.g., bicornuate or complete septate uterus); (7) E2 administration exceeding the prescribed dosage; or (8) E2 administration lasting >28 days. After applying these criteria, 792 women and 228 live birth singletons were included in the final analysis (Figure 1). The study protocol was approved by the Institutional Review Board and Ethics Committee of Shanghai General Hospital (2020KY016). Written informed consent was obtained from all patients.
2. IVF and laboratory protocols
Ovarian stimulation, oocyte retrieval, and fertilization procedures were performed as previously described [18]. For IVF, oocytes were inseminated in human tubal fluid supplemented with 10% serum substitute and approximately 300,000 progressively motile spermatozoa. For intracytoplasmic sperm injection, oocytes were placed in fertilization medium immediately after microinjection. Fertilization was assessed 18 hours later. Embryos were cultured in early cleavage medium until day 3, then transferred to multiblast medium. All cultures were maintained at 37 °C under 6% CO₂ and 5% O₂. Embryo development was evaluated on days 2, 3, 5, and 6 according to established guidelines [19]. Day 2 cleavage-stage embryos with ≥2 cells and <20% fragmentation and day 3 cleavage-stage embryos with ≥6 cells and <20% fragmentation were eligible for cryopreservation. Good-quality embryos were defined as follows: four to six cells with less than 10% fragmentation for day 2 embryos, seven to nine cells with less than 10% fragmentation for day 3 embryos.
3. Frozen embryo transfer protocols
In subsequent cycles, patients received estrogen followed by progesterone for endometrial preparation before FET. All patients began with 6 mg of oral estrogen per day (E2 valerate from Bayer or estrodail from Femoston Abbott) starting on day 2 of menstruation, continued for 1 week. Patients who maintained this regimen were classified as the oral estradiol (OE) group. Others received additional vaginal E2 supplementation (2 mg of E2 [Femoston] per day) combined with the oral dose for 7 to 21 days; these were classified as the oral and vaginal estradiol (OVE) group (Figure 2). E2 treatment duration ranged from 10 to 28 days. Weekly transvaginal ultrasonography was performed to monitor endometrial development. Vaginal estrogen use was introduced mainly in 2017–2018 after its adoption in the department. Serum E2, luteinizing hormone, and progesterone levels were measured at each visit to detect premature ovulation. When FET timing was determined, daily progesterone administration was initiated, either intramuscularly or as vaginal Crinone from Merck, along with 40 mg of oral dydrogesterone per day and 6 mg OE per day. The choice between intramuscular progesterone (60 mg/day) and vaginal crinone (90 mg/day) was based on patient preference after counseling on the advantages and side effects of each route (e.g., vaginal itching and discharge vs. subcutaneous swelling from injections). For patients receiving two day 2 embryos, progesterone was started 2 days before FET; for two day 3 embryos, it was started 3 days prior. Embryo vitrification and thawing followed previously described protocols [20]. All embryo transfers were performed with the same 20 G flexible catheter under transabdominal ultrasonographic guidance. Post-transfer, the same daily estrogen and progesterone doses were maintained until a serum human chorionic gonadotropin test was performed 14 days later. If pregnancy was confirmed, hormone support continued until 12 weeks’ gestation.
4. Outcome measures and definitions
The primary outcome was live birth rate. Secondary outcomes included clinical pregnancy rate, infant birth weight, and rates of LGA, SGA, and preterm delivery. Live birth was defined as delivery of a viable infant after 28 gestational weeks. Clinical pregnancy was confirmed by ultrasonographic detection of a gestational sac or heartbeat. Gestational age was calculated from 14 days prior to embryo transfer. Preterm birth was defined as delivery between 28 and 37 gestational weeks. SGA was defined as birth weight <10th percentile and LGA as birth weight >90th percentile. Birth weight Z-scores were calculated after adjusting for gestational age and newborn sex using the formula: Z score=(χ–μ)/σ, where χ is the newborn birthweight, μ is the mean birthweight for the same sex and same gestational age in the reference group, and σ is the standard deviation of the reference group. The reference group comprised Chinese singleton newborns [21].
5. Statistical analysis
Baseline demographic characteristics, cycle details, and clinical and neonatal outcomes for patients and live birth singletons were compared using the Student t-test, Mann–Whitney U test, chi-square test, or Fisher exact test, as appropriate. The effect of E2 regimen on binary outcomes (live birth and clinical pregnancy) was assessed using multivariable logistic regression, adjusting for maternal age, body mass index (BMI), E2 treatment >21 days, transfer of at least one good-quality embryo, progesterone route, and endometrial thickness on the day progesterone was started. The impact of E2 regimen on preterm delivery, LGA, and SGA rates was evaluated with multivariable logistic regression, adjusting for the above covariates plus newborn sex. Multiple linear regression was used to assess the independent effect of E2 regimen on singleton birth weight. Results are reported as adjusted odds ratios (ORs) with 95% confidence intervals (95% CIs). All analyses were performed using SPSS Statistics ver. 26, with p-values <0.05 considered statistically significant.
Results
1. Clinical outcome
This analysis included 792 women and 228 live birth singletons. The overall clinical pregnancy rate was 45.1%, and the live birth rate was 36.4% (288 live births, including both singletons and twins). Of the 792 women, 324 received combined OVE group, and 468 received OE group. Baseline demographics and characteristics for the two E2 regimens are presented in Table 1. Among the 792 patients, no significant differences were observed between groups in maternal age, BMI, proportion of patients with at least one good-quality embryo transferred, progesterone route, circulating E2 level on day 14 after embryo transfer, or circulating progesterone (P) level on the day progesterone was initiated. However, the proportion of patients receiving E2 treatment for >21 days was higher in the OVE group than in the OE group. The OVE group also had higher serum E2 levels and thinner endometrial thickness on the day progesterone was started, as well as reduced endometrial thickness on the trigger day during controlled ovarian stimulation (COS). The higher serum E2 achieved via vaginal administration did not result in increased endometrial thickness.
No significant differences in live birth rate (crude OR, 1.229; 95% CI, 0.917 to 1.649) or clinical pregnancy rate (crude OR, 1.260; 95% CI, 0.948 to 1.675) were found between the OVE and OE groups (Table 2). After adjusting for maternal age, BMI, E2 treatment >21 days, transfer of at least one good-quality embryo, progesterone route, and endometrial thickness on the day progesterone was started, the E2 regimen was not associated with a significant change in the odds of achieving live birth (adjusted OR, 1.327; 95% CI, 0.982 to 1.794; p=0.066) or clinical pregnancy (adjusted OR, 1.278; 95% CI, 0.937 to 1.743; p=0.121) (Table 2). Endometrial thickness on the day progesterone was started positively influenced the clinical pregnancy rate (adjusted OR, 1.184; 95% CI, 1.026 to 1.365; p=0.021) but did not significantly affect the live birth rate (adjusted OR, 1.119; 95% CI, 0.968 to 1.295; p=0.130) (Figures 3 and 4). Maternal age and the transfer of at least one good-quality embryo were independent factors positively associated with both live birth rate and clinical pregnancy rate (Figures 3 and 4).
2. Neonatal outcomes
To further evaluate the impact of E2 regimen on birth weight and gestational age, a subgroup analysis was performed in 228 live birth singletons from the overall cohort. These singletons were categorized into OVE and OE groups based on maternal E2 regimen. Baseline demographic and cycle characteristics are summarized in Table 3. No significant differences were observed between groups in maternal age, BMI, proportion of patients with at least one good-quality embryo transferred, or progesterone route. Consistent with the findings in the larger cohort, the OVE group had a higher proportion of patients with longer E2 treatment duration, thinner endometrium, and higher serum E2 levels on the day progesterone was started compared to the OE group (Table 3).
Neonatal outcomes stratified by the E2 regimen are also presented in Table 3. There were no significant differences between groups in preterm delivery rate, mean birth weight, or Z-scores. Since gestational age at delivery is a major determinant of birth weight, a subgroup analysis was conducted for term deliveries. The preterm subgroup was not analyzed for LGA and SGA rates due to insufficient sample size. Among term singletons, no significant differences were found between the OVE and OE groups in newborn gender distribution, mean birth weight, LGA rate, or SGA rate (Table 3).
In multivariate analyses (Table 4), the E2 regimen was not significantly associated with preterm delivery (adjusted OR, 0.969; 95% CI, 0.292 to 3.214; p=0.959), LGA (adjusted OR, 1.165; 95% CI, 0.545 to 2.490; p=0.694), or SGA in term deliveries (adjusted OR, 0.569; 95% CI, 0.096 to 3.369; p=0.535) after adjusting for maternal age, BMI, transfer of at least one good-quality embryo, E2 treatment >21 days, endometrial thickness on the day progesterone was started, progesterone route, and newborn gender. Due to the small sample size, multivariate analysis of LGA and SGA rates for preterm singletons was not performed. After controlling for multiple potential confounders, the E2 regimen was not significantly correlated with infant birth weight (β=–30.962; standard error=68.723; p=0.653) (Table 5).
Discussion
Previous reports have demonstrated superior IVF outcomes with the freeze-all strategy and elective embryo cryopreservation [22-24]. COS has been suggested to negatively affect endometrial receptivity [25,26]. The hyperestrogenic milieu during ovarian stimulation can disrupt embryo–endometrium synchrony and impair endometrial receptivity, contributing to lower implantation rates [27]. Perinatal outcomes, including low birth weight in term singletons, have been correlated with the supraphysiologic estrogen levels generated during COS [7-11]. Excessive E2 priming from the recruitment and maturation of multiple follicles can alter gene expression patterns, such as growth factor receptor bound protein 10 (Grb10) gene and GATA binding protein 3 (GATA3), and induce epigenetic changes, including DNA methylation and histone modifications, during embryonic and fetal development [28-31]. Additionally, ovarian stimulation has been associated with alterations in the immune environment, such as increased natural killer cell activity in oocytes compared with natural cycles [32]. Consequently, ovarian stimulation may not benefit implantation, placentation, or fetal development. While the freeze-all approach and FET improve endometrial synchrony, they are not without drawbacks. Adverse perinatal outcomes, such as increased risk of macrosomia, perinatal mortality, and complications including pre-eclampsia, have been reported [33-36]. FET singletons have been shown to have higher mean birth weights than those from fresh transfers or natural conceptions [1]. In a randomized controlled trial of 1,508 patients with polycystic ovary syndrome, the FET group showed a trend toward higher neonatal death compared with the fresh transfer group [37]. Thus, while FET offers advantages, it is not without risks.
The serum estrogen levels in ovarian-stimulated cycles differ markedly from those in artificial FET cycles. If the hyperestrogenic environment is indeed a key factor in the reduced birth weights and shorter gestations seen in stimulated cycles, then evaluating how varying serum E2 levels induced by different E2 regimens affect outcomes in artificial FET cycles is essential. Oral estrogen (Femoston) has low bioavailability due to the molecular structure of 17β-E2, with only a small proportion of the ingested dose reaching systemic circulation. In contrast, vaginally administered 17β-E2 is absorbed through the vaginal epithelium, producing higher systemic bioavailability and exerting direct local effects on the endometrium [17]. Given the increasing use of FET, understanding whether the wide range of serum estrogen levels from different regimens influences receptivity and placentation is clinically important. Few studies have examined the impact of vaginal E2 in FET cycles. One retrospective analysis of 247 artificial FET cycles found no improvement in implantation or pregnancy rates with vaginal estrogen compared to oral administration, but did not assess live birth rate [38]. Another prospective study of 78 artificial FET cycles reported that vaginal E2 increased endometrial thickness but did not evaluate clinical outcomes [39]. Our study expands upon these findings by providing a controlled analysis that assesses both live birth rates and neonatal outcomes. We found that vaginal estrogen did not increase endometrial thickness or FET success rates but was associated with higher circulating E2 levels. Although high E2 concentrations may adversely affect coagulation homeostasis and increase the risk of thromboembolic events [40], the high serum E2 priming in our cohort did not seem to have an impact on birth weight, the preterm delivery rate, the LGA rate, or the SGA rate.
Our study was not designed to identify the mechanisms underlying these results. The absence of reduced birth weight and shorter gestational periods—effects commonly associated with elevated peripheral serum E2—may be explained by the fact that serum E2 levels in our cohort were not high enough to create the supraphysiologic estrogen milieu typical of fresh IVF cycles. In COS, mean peak serum E2 levels range from 10,460 to 15,362 pmol/L, and values above this threshold have been linked to increased risk of preterm labor and SGA [7,11]. In our study, the median serum E2 levels on the day progesterone was started were 6,106 pmol/L in the OVE group and 992 pmol/L in the OE group—both much closer to the levels observed in natural cycles. Furthermore, in the OVE group, vaginal E2 was not used as a complete replacement for OE; rather, oral administration was continued with the addition of vaginal Femoston.
The present study has several strengths. All data were obtained from a single IVF center, ensuring consistency in laboratory equipment, personnel, and procedures. To minimize bias related to newborn sex and gestational age, birth weight Z-scores were calculated. Progesterone regimens were accounted for as potential confounders in the multivariate analysis, given that supraphysiologic progestin exposure in HRT may promote excessively deep placentation, potentially altering obstetric outcomes [41]. However, previous studies have shown that different progesterone regimens generally yield comparable pregnancy rates [42], and including both progesterone and E2 regimens in our model improved its robustness. We further reduced confounding by excluding patients with endometrial thickness <6 mm on the progesterone start day or with possible endometrial lesions. Patients receiving estrogen for more than 28 days were excluded, as prolonged priming has been associated with reduced pregnancy rates in FET [15]. Blastocyst transfers were also excluded to avoid the potential influence of extended in vitro culture on birth weight and gestational age [43,44]. By focusing on a homogeneous cohort of first-cycle FET patients with adequate endometrial thickness and good-quality embryo transfers, and by incorporating long-term follow-up, we were able to evaluate key neonatal outcomes with greater reliability.
This study also has limitations. Its retrospective design introduces the possibility of selection bias, particularly as vaginal E2 was adopted after clinicians in our center became familiar with the method. Not all potential confounding variables could be controlled. Furthermore, we were unable to obtain detailed obstetric data, including information on hypertensive disorders of pregnancy or placental abnormalities, which limited our ability to fully assess risk factors for preterm delivery. Future studies incorporating obstetric outcomes such as pre-eclampsia, placenta accreta, placenta previa, and pregnancy-induced hypertension would strengthen the analysis and help clarify the potential impact of elevated serum estrogen on placental angiogenesis in late pregnancy.
In conclusion, this single-center retrospective study provides evidence on the impact of E2 regimen in HRT-prepared FET cycles. Higher serum E2 levels resulting from combined oral and vaginal administration did not increase live birth rates or the risks of preterm delivery and low birth weight compared with oral administration alone. Vaginal estrogen did not improve endometrial thickness or FET success rates, but did lead to higher serum E2 levels, which may carry potential (albeit unconfirmed) maternal risks. Based on these findings, E2 regimens may be tailored to patient preference without compromising clinical or neonatal outcomes, provided that potential risks are clearly communicated and considered in clinical decision-making. Large, prospective, randomized controlled trials with detailed obstetric follow-up are needed to confirm our results and to better understand the effects of estrogen on placental vascular development and related obstetric complications.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
