Does Estradiol Orally and Vaginally Administered Together Impact Live Birth and Neonatal Outcome in Articial Frozen-thawed Embryo Transfer Cycles: A Retrospective Cohort Study

Background: Previous studies have demonstrated that newborns from fresh embryo transfer are with higher risk of small for gestation (SGA) rate than those from frozen-thawed embryo transfer (FET). It is suggested that supraphysiologic serum estradiol in controlled ovarian stimulation (COS)is one of reasons. Out study aims to investigate whether exogenous estradiol delivered regimens have an impact on live birth rate and singleton birthweight in hormone replacement (HRT)-FET cycles. Methods:This retrospective study involved patients undergoing their rst FET with HRT endometrium preparation followed by two cleavage-staged embryos transfer, comparing orally and vaginal estradiol tablets (OVE) group versus oral estradiol tablets (OE) group from January 2015 to December 2018 at our center. A total of 792 patients fullled the criteria, including 282 live birth singletons. Live birth was the primary outcome. Secondary outcome included clinical pregnancy rate, singleton birthweight, large for gestational age (LGA) rate, SGA rate, preterm delivery rate. Results:Patients in OVE group achieved higher serum estradiol level with more days of estradiol treatment. No difference in live birth (Adjusted OR 1.327; 95%CI 0.982, 1.794, p=0.066) and clinical pregnancy rate (Adjusted OR 1.278; 95%CI 0.937, 1.743, p=0.121) was found between OVE and OE groups. Estradiol route did not affect birth weight (β=-30.962, SE=68.723, p=0.653), the odds of LGA (Adjusted OR 1.165; 95%CI 0.545, 2.490, p=0.694), the odds of SGA (Adjusted OR 0.569; 95%CI 0.096, 3.369, p=0.535) or the preterm delivery rate (Adjusted OR 0.969; 95%CI 0.292, 3.214, p=0.959). Conclusion:Estrogen orally and vaginally together did not have an impact on clinical outcomes and singleton birthweight compared to


Background
Vitri cation technology has played an instrumental role in the implementation of fertility preservation, single embryo transfer and freeze-all cycles and helped avoid ovarian hyperstimulation syndrome (OHSS) in assisted reproductive technology (ART). During frozen-thawed embryo transfer (FET), though no endometrial preparation protocol superior to others is found, hormone replacement treatment (HRT) is preferred by many centers due to the controlled exposure to exogenous steroids and adjustment of the timing of transfer. In HRT endometrial prepared FET cycles, estradiol and progesterone are sequentially administered to synchronize the window of embryo implantation with embryo transfer. Endometrium proliferated and ovulation was hampered due to the continued estrogen administration during the initial estrogen-only phase. Daily progesterone started before the scheduled embryo transfer. Adequate estradiol priming induced endometrial cells proliferation and appropriate progesterone receptors to develop endometrial receptivity.
The growing number of FET has made researchers to examine various aspects of this treatment strategy. Epidemiologic data has indicated that FETs are associated with increased risk of large and post-date babies and adverse perinatal outcomes like pregnancy related hypertension, postpartum hemorrhage, compared to fresh transfer [1][2][3][4][5][6]. The increased risk of preterm labor and small for gestational age (SGA) babies was found in fresh-transfer ART and was fundamentally decreased in FET.
The reasons behind that varied. Though some reports suggested embryo cryopreservation altered epigenetics regulation inducing abnormal placentation and fetal growth, more studies advised the elevated level of estradiol which has been shown to impair trophoblastic invasion led to the smaller birthweight and shorter gestational weeks in fresh IVF cycles [7][8][9][10][11][12]. The negative effect of supraphysiologic E2 milieu on implantation and placentation was believed eliminated by the subsequent transfer of frozen-thawed embryos in the relatively steady endocrine environment. However, the duration and dose of estradiol administration in HRT-FET cycles intervened with the clinical and neonatal outcomes. Sekhon et al. have indicated that duration of estradiol supplementation before progesterone in arti cial cycles did not impact frozen blastocyst transfer outcome, but inversely correlated with gestational age [13]. Constant dose or increasing dose of estrogen in HRT for endometrium preparation though didn't change live birth rate [14], but excessive duration of estradiol priming which was more than 28 days induced slacked pregnancy rate and increased miscarriage rates in FET [15].
In HRT endometrium preparation, different routes estrogen absorbed induced different serum estradiol level. Estrogen administered vaginally was rstly reported in 1988 [16]. The alternative vaginal rings allowed a steady release of estrogen into the bloodstream. Estrogen administered vaginally seemed to cause higher circulating E2 level and even higher in endometrial tissue than estrogen administered orally [17]. The objective of the current retrospective study was to investigate whether estrogen orally and vaginally together administration made a difference to the live birth rate and singletons birthweight compared to estrogen orally administration. This model was chosen to speci cally focus on the effect of vaginal absorbed estrogen induced higher E2 priming on FET clinical and neonatal outcomes.

Study design and patient population
This is a retrospective study which was undertaken at the assisted reproduction medicine department of Shanghai general hospital a liated to Shanghai Jiao Tong university school of medicine. It involved 1005 women who had undergone their rst FET from January 2015 to January 2018. Data was extracted from the database in our center.Patients inclusion criteria wasmaternal age <48, two Day 2 or Day 3 cleavage-staged embryos transfer following HRT endometrium preparation. The following patients were excluded from the analysis: (1) Patients with cryopreserved oocytes or donor oocytes. (2) Patients who had prior attempts at conception via IVF and FET. (3) Patients who were in HRT treatment with Sidena l or growth hormone to improve endometrium development. (4) Patients with endometrium thickness less than 6mm at endometrium transform day. (5) Patientswith endometrial polyps, submucosal myomas, or endometrium separation. (6) Patients with estrogen administration beyond the doctor's order. (7) Patients with duration of estradiol administration > 28 days. 792 women and 228 live birth singletons were nally included in the criteria (Figure 1). Institutional review board and ethics committee of Shanghai General Hospital approval was obtained (2020KY016).

IVF and Laboratory protocols
Ovarian stimulation, oocyte retrieval and fertilization procedures have been previously presented [18].
As for IVF, oocytes were inseminated with human tubal uid with 10% serum substitute supplement and around 300000 progressively motile spermatozoa. As for ICSI, oocytes were laid in the fertilization medium immediately after microinjection. Fertilization was evaluated 18 hours post fertilization procedures. Embryos were cultured in early cleavage medium before Day3 and in multiblast medium afterwards. All embryos were cultured in incubator at 37°C, under 6% CO 2 and 5% O 2 . Embryo development was evaluated on Day2, Day3, Day5 and Day6. Day2 or Day3 cleavage-stage embryos with at least three or six blastomeres respectively and less than 20% fragmentation were eligible for cryopreservation. The criteria for good-quality embryo were: 4 to 5 cells without fragmentation for Day 2 embryos, 7 to 9 cells with less than 20% fragmentation for Day3 embryos.

Frozen embryo transfer protocols
In a subsequent cycle, patients were administered estrogen and consecutive progesterone for endometrial preparation before FET. All patients started with estrogen administered orally (Estradiol Valerate or Estradiol Femoston) 6mg per day from the second day of menstruation and continued for one week. Subsequently, some patients continued the same regimen until progesterone was applied. "OE group" was used for these patients. Other patients were given extra estradiol supplement vaginally, Estradiol Femoston 2mg per day, combined with the oral usage for another7-21 days. "OVE group" was used for these patients ( Figure 2). The whole estradiol administration duration could vary from 10 to 28 days. Patients came for visit every week. We performed transvaginal ultrasonagraphy every week to assess patients' endometrium. Vaginally administered estrogen was usually added if the endometrium thickness was less than 6mm in patient's rst visit. Additionally, it was also depended on physician's preference. Serum E2, LH and progesterone level were measured at each visit to detect premature ovulation. Once the timing of FET was determined, progesterone intramuscularly or Crinone vaginally was initiated daily. They both combined with dydrogesterone orally 40mg per day and estradiol orally 6mg per day. Intramuscular Progesteron 60mg per day or Vaginal Crinone 90mg per day was chosen according to patient preference. Patients who were undertaken two Day 2 or Day 3 cleavage-staged embryos transferred started the progesterone 2 or 3 days before FET, respectively. The vitri cation and thawing procedures were previously described [19]. Embryo transfer was performed via the same exible catheter (20G)and transabdominal ultrasonography guidance. After embryo transfer, same daily estrogenand progesterone administration continued until a negative pregnancy testat the 14thday after transfer. If pregnancy was achieved, hormone administration continued until 12 weeks' gestation.

Outcome measures and de nitions
In order to evaluate the impact of estradiol delivered orally and vaginally together on clinical outcome, the primary outcome analyzed was live birth rate. Clinical pregnancy rate, newborn birthweight, LGA, SGA, preterm delivery rate were secondary outcomes. Live birth was de ned as a delivery of a viable infant after 28th gestational weeks. Clinical pregnancy was a pregnancy con rmed by the gestational sac or heart beat under sonography. Gestational age was calculated from 14 days before the embryo transfer.
Preterm birth was de ned as delivery between 32 to 37 gestational weeks. Small for gestational age (SGA) and large for gestational age (LGA) were de ned as birthweight <10th and >90th percentile, respectively. Z score was administered to calculate birthweight adjusted for gestational age and newborn gender with the formula: Z score=(c-m)/s, where c is the newborn birthweight, m is the mean birthweight for the same sex and same gestational age in the reference group and s is the standard deviation of the reference group. The reference is the Chinese singletons newborns [20].

Statistical analysis
Patients and singletons live birth demographic baseline, cycle characteristics, clinical and neonatal outcomes were compared with Student's t-test, Wilcoxon rank sum tests, chi-square and Fisher's exact tests, as appropriate. Whether binary live birth and clinical pregnancy were affected by the regimen of estradiol was assessed by multivariable logistic regression adjusting for maternal age, BMI, whether duration of estradiol treatment >21 days, whether at least one good quality embryo was transfered, the route of progesterone administered, endometrium thickness at endometrium transform day. Multivariable logistic regression was applied to evaluate estradiol regimen impact on preterm delivery, LGA and SGA rate, adjusting for the major covariates mentioned above plus newborn gender. Multiple linear regression was applied to investigate the independent effect of estrogen regimen on newborn birthweight. Adjusted odds ratios (OR) and 95% con dence intervals (95%CI) were reported. All analyses were conducted with spss statistics. P value<0.05 was considered statistically signi cant.

Clinical outcome
This analysis included 792 women and 228 live birth singletons with the following outcomes: 45.1% clinical pregnancy rate, 36.4% live birth rate. There were 324 patients who were estrogen administered orally and vaginally (OVE group), and 468 patients who were estrogen administered orally only (OE group). Baseline demographics and characteristics were compared between patients with different estrogen regimen (Table 1). Among the 792 women, it did not reveal any signi cant difference for maternal age, BMI, whether there was at least one good quality embryo transferred, progesterone route, E2 level at 14th day after transfer and P level at endometrium transform day between two groups. There were more patients with estradiol treatment days over 21 days in OVE group than in OE group. Serum E2 level at endometrium transform day from OVE group was signi cantly higher than those from OE group. Endometrium thickness at transform day was smaller in OVE group than OE group, as the same result for the endometrium thickness at HCG day in controlled ovarian stimulation (COS). No signi cant difference of live birth rate (Crude OR 1.229, 95CI% 0.917, 1.649) and clinical pregnancy rate (Crude OR 1.260, 95%CI 0.948, 1.675) was found between OVE and OE group (  Figure 1&2). Maternal age and at least one good quality embryo transferred were the independent factors that increased the live birth rate and clinical pregnancy rate (Supplementary Figure 1&2).

Neonatal outcome
To further explore the estrogen regimen impact on birthweight and gestational age, a cohort of 228 live birth singletons from 792 patients was further investigated. The singletons were divided in two groups which were OVE group, and OE group. Baseline demographic and cycle characteristics were presented in Table 3. Comparison between the two groups did not reveal any signi cant difference for maternal age, BMI, whether there was at least one good quality embryo transferred and progesterone route. OVE group from 228 singletons cohort had more percentage of patients with longer days of estradiol treatment, thinner endometrium before FET and higher serum E2 level at endometrium transform day than OE group, as the same result with the 792 cohort (Table 3).
Neonatal outcomes strati ed by the regimen of estradiol administered were also presented in Table 3.
Preterm delivery rate, mean birthweight and Z-scores were not different across two groups ( Table 3). Given that gestational age at delivery is an important determinant of newborn birthweight, a subgroup analysis was conducted strati ed by term delivery. The analysis of gestational age between 32 to 36 weeks was not performed because the number in this category was too small. As shown in Table 3, Newborn gender, mean birthweight, LGA rate and SGA rate for term singletons were not different across the two groups.
In multivariate analyses (Table 4), the risk of preterm delivery (Adjusted OR 0.969, 95%CI 0.292, 3.214) and the risk of LGA (Adjusted OR 1.165, 95%CI 0.545, 2.490), SGA in term delivery (Adjusted OR 0.569, 95%CI 0.096, 3.369) were not signi cantly different between two groups after adjusting for maternal age, BMI, transfer with at least one good quality embryo, whether estrogen administration more than 21 days, endometrium thickness at endometrium transform day, progesterone route and newborn gender. The multivariate analysis of LGA and SGA for gestational age between 32 to 36 weeks was not performed because the number in this category was too small. After correction for a number of potential confounders, the route of estradiol administered was not correlated with neonatal birthweight (β=-30.962, SE=68.723, p=0.653) ( Table 5).

Discussion
From our study, estrogen vaginally and orally regimen did not improve clinical pregnancy rate or live birth rate compared to estrogen orally regimen in HRT endometrial preparation FET cycles. Adding estrogen vaginally supplement did not change the newborn weight, SGA, LGA or the preterm delivery rate of the live birth singletons either.
Nowadays, a growing number of reports advised better IVF outcomes after freeze-all policy and elective embryo cryopreservation [21][22][23]. Some research has suggested COS has made a detrimental effect on endometrium receptivity [24,25]. They reported lower implantation rates in fresh transfer cycles compared to FET cycles due to the supraphysiologic estrogen level on embryo-endometrium asynchrony and impaired endometrial receptivity after ovarian stimulation [26]. Adverse perinatal outcomes such as low birth weight newborn in term singletons have been associated with fresh IVF-ET because of the hyperestrogenic milieu generated during COS in IVF [7][8][9][10][11]. Excessive E2 priming induced from development and recruitment of multiple follicles can affect gene expression pattern change like Grb10 and GATA3, and generate epigenetic alterations in developing embryo and fetus through alterations in DNA methylation, histone alteration and others [27][28][29][30]. It is also reported that immune environment changed due to the increase of natural killer cells in oocytes during stimulated cycles compared with natural cycles [31]. Ovarian stimulation would probably not bene t implantation, placentation and subsequent fetal growth. Since freeze-all policy and FET have been welcomed as global trend in recent years due to the endometrial synchrony and less perinatal morbidity, we must recognize the adverse perinatal outcomes of FET, like higher risk of macrosomia, perinatal mortality and pregnancy complications like preeclampsia [32][33][34][35]. There is evidence that FET singletons have higher mean birthweight than singletons born after fresh embryo transfer and natural pregnancy [1]. A trend toward higher neonatal death was found in FET group compared to fresh ET group in a RCT of 1508 PCOS patients [36]. FET is not perfect without risks.
Dramatically large-spanned levels of serum estrogen were found between ovarian stimulated cycles and arti cial FET cycles. If the supraphysilogic E2 priming was one important reason for the low birth weight and shorter gestational weeks in stimulated cycles singletons, we wanted to explore the impact of different serum E2 levels induced by different routes of estrogen on arti cial FET clinical and neonatal outcome. Estradiol valerate orally has the structure of 17 β-estradiol, which possessed low bioavailability. Only very small part of ingested dose reaches the circulation intact. While the vaginally delivered 17 β-estradiol exerts the local effect on endometrium after vaginal epithelium absorption, bypassing the liver metabolism and achieving higher bioavailability [17]. The regimen of estrogen administered usually varies depending on clinical variables like endometrium thickness, previous clinical records and physician preference. Considering the rise in FET cycles applied in recent years, it is essential to investigate whether the peripheral serum estrogen levels induced by estradiol absorbed regimens can impact endometrial receptivity and placentation. There is scarce publication available regarding the effect of estrogen administered vaginally route in FET cycles. To our knowledge, one retrospective study of 247 arti cial FET cycles has suggested that vaginal estrogen regimen did not promote implantation and pregnancy rate compared to oral regimen. But they didn't investigate the live birth rate [37]. The other prospective study of 78arti cial FET cycles showed vaginal estradiol administration improved endometrial proliferation by eliciting thicker endometrium, but it did not dig and analyze clinical outcomes [38]. Our present study has expanded this concept by providing the comprehensive wellcontrolled analysis, which evaluated estrogen administration route impact on both live birth rate and perinatal outcome.
Sekhon et al. suggested duration of estradiol supplementation before progesterone in arti cial cycles did not impact frozen blastocyst transfer outcome [13]. This effect may extend beyond the route estradiol absorbed as we demonstrated that estrogen vaginally or orally administered did not impact clinical pregnancy rate and live birth rate in arti cial FET cycles. Estrogen administered vaginally and orally together induced higher serum estrogen with longer duration of estrogen supplement, but ended up with thinner endometrium compared to orally only group. It is noteworthy that the patients in OVE group probably had inferior endometrium proliferation capability, considering their thinner endometrium both in COS and the rst visit in arti cial FET cycles. From our result, estrogen vaginally administered didn't improve endometrium thickness and neither increased the successful rate in FET, but was accompanied with the higher circulating estradiol level. The extremely high E2 level might adversely affect the clotting system homeostasis with higher incidence of thromboembolic events [39]. Nevertheless, the high serum E2 priming caused by estrogen vaginally and orally delivered seemed to not have an impact on singletons preterm delivery rate, LGA, and SGA rate.
Our study was not designed to identify the mechanisms underlying the results. The reason why we didn't see the lower birthweight and shorter gestational weeks induced by the higher peripheral serum estradiol in our cohort may be speculated that the serum estradiol level in our analysis is still not high enough to create the supraphysiologic estradiol milieu as in fresh IVF cycles. The mean serum peak estradiol value in COS varied from 10460-15362 pmol/L. Estradiol higher than that cutoff value proclaimed the high risk of preterm labor and SGA [7,11]. The median value of serum estrogen level at endometrium transform day in our study was 6106 and 992 pmol/L in two groups respectively, which were both much closer to serum E2 level in natural cycle. Additionally, we didn't use estrogen vaginally gel insertion replaced of estrogen orally taken. In OVE group, we still continued the orally-taken estrogen valerate and subsequently added estrogen gel vaginally in case of vaginal malabsorption in some patients.
The present study has following strengths. The data was collected from one IVF center whichguaranteed the same lab equipment and associated operating people and sonographers. This ensured the practice consistency. Male neonates have been predicted to be higher-weighted than female neonates. To control for infant gender and gestational age bias, z score was calculated across different groups. We also accounted for progesterone regimen as potential confounder in multivariate analysis.
Progestin has been suggested to directly advance vascular proliferation during placentation [40]. A supraphysiologic progestin exposure in HRT could initiate excessively deep placentation which would make a difference in infant birthweight and obstetrical consequences [41]. The progesterone supplementation route in this analysis included two regimens. Though different progesterone regimens seemed to induce comparable pregnancy rate [42], taking into consideration of progesterone and estrogen regimen simultaneously greatly bene tted in our model. We reviewed the database and restricted the analysis to strict criteria. We tried to minimize any effect associated with uterine factor infertility by excluding patients with endometrium thickness less than 6mm at endometrium transform day and patients who have presented with endometrial polyps, submucosal myomas, or endometrium separation. We didn't include patients with estrogen administered more than 28 days in our study to minimize the effect of extended duration of estrogen, as E2 priming longer than 28 days showed a slacked pregnancy rate in FET [15]. We didn't include blastocyst embryo transfer into the whole cohort to mitigate the possibleimpact of prolonged in vitro embryo culture on subsequent newborn birthweight and gestational weeks [43,44]. Only homogenous group of patients who achieved adequate endometrial thickness in their rst FET cycles were included to guarantee the relative good-quality embryo transfer and exclude the recurrent FET failure cases, as poor-quality embryos were put forward to be associated with lower birthweight [45]. The long-term follow up of the cohort allowed to provide important neonatal outcomes.
Our study is limited by its retrospective design. Whether and when the estrogen vaginally was administered depended on both patients' endometrium thickness and physician's preference. Patients added with estrogen vaginally administered are likely to possess inferior capability of endometrium growth and inherent uterine defects that may predispose them to thinner endometrium. The selection bias was possible. We couldn't account for all possible cofounding variables. We couldn't follow up the patient obstetrics details including hypertensive disorders of pregnancy or placental abnormalities, which hampered us to assess the pregnancy risk factors associated to preterm delivery. An evaluation of other perinataloutcomes such as pre-eclampsia, placenta accrete, previa, abruption hypertensive would allow us a better understanding of high serum estrogen level's impact on placental angiogenesis in late trimester.

Conclusion
This single-center analysis provided some evidence of the estradiol regimen for women undergoing HRT endometrium preparation FET cycles. Relatively higher serum E2 level induced by estrogen orally and vaginally together did not increase successful IVF outcome or patients' risk of delivery premature and low birth weight infants compared to estrogen orally only administration. Estrogen vaginally delivered didn't improve endometrium thickness much and neither successful rate in FET, but induced higher serum E2 value with potential maternal undesirable risks. Based on our ndings, upon safe and effectiveness consideration, estrogen route could be adjusted to patient preference without compromising FET outcome. Patients should be adequately informed the potential risks and patient's opinion should be taken into consideration in medical practice. Large randomized controlled prospective clinic trials with detailed obstetric follow-up are needed to verify our result and allow a better understanding of the estrogen impact on placental-vessel associated obstetric complications.  Note: Analyses were adjusted for maternal age, BMI, whether days of estrogen treatment> 21, whether at least one good quality embryo was transferred, endometrium thickness at endometrium transform day, progesterone regimen.OE group stands for estradiol orally administered group. COS stands for controlled ovarian stimulation. CI=con dence interval; OR=odds ratio. Table 3 Baseline demographics, cycle characteristics and neonatal outcomes of singletons live birth according to different estradiol route quartile, third quartile) for continuous variables in informal distribution. P values were assessed with the use of t tests or Wilcoxon rank sum tests or chi-square (Fisher's exact tests as appropriate). COS stands for controlled ovarian stimulation. SGA stands for small for gestational age.
LGA stands for large for gestational age. Note: Analyses were adjusted for maternal age, BMI, transfer with at least one good quality embryo, endometrium thickness, whether estrogen administration >21 days, progesterone route, newborn gender.
CI=con dence interval; OR=odds ratio.SGA stands for small for gestational age.
LGA stands for large for gestational age. Figure 1 Flow diagram of patient inclusion.