Optimal Endometrial Preparation Protocols for Frozen-thawed Embryo Transfer Cycles by Maternal Age

This retrospective cohort study aimed to explore the optimal endometrial preparation protocols among different maternal age groups. A total of 16,867 frozen-thawed embryo transfer (FET) cycles were categorized into three groups based on endometrial preparation protocols: Natural cycle (NC n = 3893), artificial cycles (AC, n = 11456) and AC with GnRH-a pretreatment (AC+GnRH-a, n = 1518). To account for repeat cycles, a generalized estimating equation (GEE) method was applied to examine the associations between cycle regimens and pregnancy outcomes. Subgroup analyses were conducted to evaluate the best preparation methods for different maternal age groups. Primary outcomes were live birth and early miscarriage rates. After completing GEE, in overall population, the live birth rate [(NC as reference; AC: adjusted odds ratio (aOR) = 0.837, 95% confidential interval (CI) 0.771–0.908; AC+GnRHa: aOR = 0.906, 95%CI 0.795–1.031)] in NC was significantly higher than that in AC, while comparable that in AC+GnRH-a. The early miscarriage rate (AC: aOR = 1.420, 95%CI 1.225–1.646; AC+GnRHa: aOR = 1.545, 95%CI 1.236–1.931) was significantly lower in NC compared to either AC group. Subgroup analysis showed that in younger women, the incidences of live birth (AC: aOR = 0.900, 95%CI 0.804–1.007; AC+GnRHa: aOR = 1.091, 95%CI 0.904–1.317) were equivalent between groups, with a slightly higher in AC+GnRH-a. Early miscarriage rate (AC: aOR = 1.462, 95%CI 1.165–1.835; AC+GnRHa: aOR = 1.137, 95%CI 0.948–1886) was only significantly lower in NC compared to that in AC. In older women, the live birth rate (AC: aOR = 0.815, 95%CI 0.722–0.920; AC+GnRHa: aOR = 0.759, 95%CI 0.627–0.919) was significantly higher, and early miscarriage rate (AC: aOR = 1.353, 95%CI 1.118–1.638; AC+GnRHa: aOR = 1.704, 95%CI 1.273–2.280) was significantly lower in NC compared to either AC group. Our study demonstrated that NC is associated with lower early miscarriage late in overall IVF population. There is a mild favor of AC+GnRH-a in younger women, while the priority of NC is remarkable in older women. Maternal age should be a considerable factor when determining endometrial preparation method for FET.


Introduction
Since the first report on successful pregnancy after frozenthawed embryo transfer (FET) in 1983, the amount of FET has been increasing worldwide, particularly driven by the advanced vitrification technique allowing safe and efficient cryopreservation, storage and warming of embryos [1,2]. Current evidence indicates that FET cycles not only produce non-inferior live birth rate to fresh cycles, but reduce multiple pregnancy rate by selecting single good-quality blastocyst for transfer, and minimize the risk of ovarian hyperstimulation syndrome [3][4][5]. Regardless, the optimal method of endometrial preparation remains a topic of ongoing debate.
The major endometrial preparation protocols can be generally classified into natural cycle (NC) and artificial cycle (AC). In NC, follicle development and ovulation are required to maintain the physiological hormonal milieu for endometrial growth and embryo implantation. Alternatively, in AC, this physiological process is overridden by the administration of exogenous estrogen and progesterone. AC with GnRH-a pretreatment is applied to minimize the risk of premature ovulation and prevent cycle cancellation [6], it was also reported to increase live birth in patients with adenomyosis [7,8].
In clinical practice, AC is increasingly adopted as NC may not be possible in patients with ovulatory disorders. Additionally, AC provides a better control of FET timing and transfer, which is convenient for both patients and physicians. However, question remains unanswered whether AC is equivalent to NC regarding pregnancy outcomes and safety, as several recent publications have reported increased adverse obstetric and perinatal outcomes in the absence of a corpus luteum [9][10][11][12]. In this study, we aimed to compare pregnancy outcomes after different endometrial preparation protocols, and explore the most optimal cycle regimens in different age groups.

Study design and population
This retrospective cohort study included 16867 FET cycles with autologous oocytes from January 2015 to June 2019 in our fertility center. Excluding criteria were cycles with no viable embryos available for transfer or underwent preimplantation genetic diagnosis, cycles in which transferred embryos came from different ovarian stimulation cycles or sequential embryo transfer mixed with cleavage and blastocyst stage embryo, and cycles lost to followup. The eligible cycles were classified into three groups according to the endometrial preparation protocols: natural cycle (NC), artificial cycle (AC) without GnRH-a, and AC with GnRH-a pretreatment (AC+GnRH-a). This study was approved by the Institutional Review Board of the Shenzhen Zhongshan Urology Hospital (approval SZZSECHU-F-2020043). The requirement of informed consent was waived due to the retrospective nature of the study.

Cryopreservation and thawing
Standard regimens for controlled ovarian stimulation (COH) were applied by clinicians in our center based on the individual ovarian reserve and response, including long agonist protocols, antagonist protocols, and clomiphene-based mild stimulation protocols. Ovulation was triggered with either human chorionic gonadotropin (hCG, Lizhu; Zhuhai, China) 6 500-10,000 IU or a single subcutaneous bolus of triptorelin (Diphereline; Ipsen, France) 0.2 mg, oocyte retrieval was performed 34-36 hours after. Insemination of mature oocytes was performed by conventional IVF or ICSI according to the sperm parameters. Details on embryo culture, vitrification, thawing and transfer procedures have been described in our previous studies [13,14]. Embryos on Day 3 were graded according to the morphology criteria [15]. Good and fair embryos were cryopreserved or underwent blastocyst culture. The Gardner grading system was used to evaluate the blastocyst quality [16]. Only blastocysts better than grade 3CC were selected for vitrification. The laboratory procedures were performed by well-trained embryologists, each with over 5 years of laboratory experience. There were no substantial changes of laboratory practices over the course of study.

Endometrial preparation protocols
The patients were allocated to different endometrial preparation protocols based on the experience of the clinician and patients' characteristic. NC was chosen if the patient had regular menses or refused to take medication, AC with or without GnRH-a was selected in patient with irregular menses, or who lived at considerable distance and did not wish to be frequently monitored.
NC: Surveillance of the cycle was done from day 8-10 of the cycle with vaginal ultrasound until the leading follicle was ≥18 mm or the urine LH surge was observed. Ovulation may occur spontaneously or triggered by 10,000 IU hCG (Lizhu; Zhuhai, China). Oral administration of 20 mg progesterone twice daily was prescribed for 3/5 days before FET of cleavage/blastocyst stage embryo.
AC: Endometrial preparation was started from day 2-3 of menstrual cycle, 4 mg of oral estradiol valerate (Progynova; Bayer, Germany) per day was started, and the dose was increased by 2 mg every 5 days. Estrogen was given for 15 days. Intramuscular administration of 60 mg progesterone daily (ZheJiang XianJu Pharmaceuticals, China) was prescribed for 4/6 days before FET of cleavage/blastocyst stage embryo.
AC+GnRH-a: The injection of leuproreline acetate (Diphereline; Ipsen, France) 3.75 mg i.m., was administered during the mid-luteal phase of the menstrual cycle, twenty-nine days after the hormone replacing protocol as in AC was started.
In all three protocols, serum estradiol (E 2 ), progesterone (P) levels, and endometrium thickness were measured on hCG day in NC or on progesterone day in AC. One-two blastocysts or 1-3 cleavage stage embryos were transferred under ultrasound guidance. On the day of embryo transfer, luteal support was prescribed with 2 0 m g p r o g e s t i n t a b l e t s (D up h a s t o n; A bb o t t , Netherlands) orally twice daily and 90 mg progestin gel (Crinone; Merck, Germany) vaginally daily until a serum beta hCG assay was performed 11 (if blastocyst was transferred) or 13 days (if cleavage embryo was transferred) after FET. Luteal support was continued to 12 weeks of gestation if pregnancy was resulted.

Definition of clinical outcomes
Primary outcomes were live birth and early miscarriage rate per cycle. Secondary outcomes were clinical pregnancy and ongoing pregnancy rate. Live birth was defined as delivery of a live fetus after 24 completed weeks of gestational age. Early miscarriage was defined as loss of clinical pregnancy before 12 weeks of gestation. Clinical pregnancy was defined as the observation of at least one gestational sac on vaginal ultrasound at 6-7 weeks of gestation. Ongoing pregnancy referred to a clinical pregnancy proceeding beyond 12 weeks of gestation.

Statistical analysis
The demographic characteristics and clinical outcomes were described as mean ± SD for continuous variables and as frequency with proportion for categorical variables. The differences between groups were tested using the ANOVA test for continuous variables and the Pearson's chi-square test for categorical variables. As repeat cycles were included in the cohort, a generalized estimating equations (GEE) method was applied to examine the associations between replacement regimens and pregnancy outcomes after controlling for potential confounders, which included maternal age, BMI, parity, duration of infertility, indication of treatment, fertilization method, endometrium preparation protocol, embryo stage, number of embryos transferred, number of top embryos and endometrium thickness. Endometrial preparation protocol was included as a categorical variable, and NC was selected as the reference. The results were reported as adjusted odds ratios (aORs) with 95% confidential intervals (CIs). Subgroup analyses s t r a t i f i e d b y m a t e r n a l a g e ( < 3 5 y e a r o r ≥ 3 5 year)were performed to investigate the optimal endometrial preparation protocols for different age groups. Pvalues (two-tailed) less than 0.05 were considered statistically significant. SPSS version 19.0 (IBM SPSS, Chicago, IL) was applied for the statistical analysis.

Results
Totally there were 16,867 FET cycles (from 11,183 patients) enrolled in the study: 3893 NC cycles, 11,456 AC cycles without GnRH-a and 1518 AC cycles with GnRH-a pretreatment. The demographic data are summarized in Table 1. Maternal age at transfer in AC without GnRH-a was about two years younger than the other groups. When stratified by maternal age, there were 9339 cycles with younger age (<35 years), and 7528 cycles with advanced age (≥35 years).
The pregnancy outcomes stratified by maternal age are shown in Fig. 2. Differences regarding pregnancy outcomes between groups seemed apparent in older women, but obscure in younger women. GEE models suggested that in younger women (

Discussion
With the continuous increase in the number of FET cycles, determining the optimal endometrial preparation protocol has become paramount to maximize IVF success. Our study demonstrated that NC is superior to AC regarding pregnancy outcomes in general IVF population. To the best of our knowledge, our study is the first to evaluate the optimal endometrial preparation protocols between different age groups and  concludes that there is a mild favor of AC+GnRH-a in younger women, but a remarkable priority of NC in older women. The availability of a good quality, euploid embryo, a receptive endometrium and synchrony of these factors remains central for a successful pregnancy. In FET, endometrium receptivity is achieved by dedicated endometrial preparation protocols, which can be largely divided into natural and artificial cycles. In NC, usually solely menstrual cycle monitoring is performed without any pharmacological intervention prior to ovulation, which may be spontaneous (true NC) or triggered by HCG (modified NC). However, the timing of ovulation in NC may pose scheduling difficulties, and premature ovulation may occur and increase cancellation rates. In AC, exogenous hormone is administered to prepare the endometrium for embryo implantation, with estrogen to prime the endometrium, while progesterone to complete endometrial maturation. AC with GnRH-a pretreatment offers the most control over the timing and minimize the risk of premature ovulation, but the cycle is much more prolonged and expensive.
Several studies have compared NC and AC cycles in endometrial preparation, unfortunately the results have been conflicting. Two large scale meta-analyses comparing different cycle regimens of FET revealed no significant advantage of one approach over the others in terms of reproductive outcomes, however, the studies included were often of a low or very low quality of evidence as most were retrospective character with limited sample size, and some failed to report important clinical outcomes, or had a poor reporting of study methods [6,17]. A retrospective cohort study of 1265 cycles showed that the implantation rate was significantly higher in NC, while there were no significant differences between groups in the clinical pregnancy, ongoing pregnancy, live birth, and miscarriages rates [18]. El-Toukhy T and coworkers [19] conducted a prospective randomized trial of 234 patients and found that using GnRH-a prior to exogenous steroid  supplementation for endometrial preparation achieved significantly higher clinical pregnancy (24% vs 11.3%, OR 2.5, 95%CI 1.2-5.5) and live birth rates (20% vs 8.5%, OR 2.9, 95%CI 1.2-8). Our study suggested that in overall population, the chance of the live birth was higher in NC compared to AC without GnRH-a, but comparable to AC with GnRH-a. While the early miscarriage rate in NC was significantly lower than the two AC groups, either with or without GnRH-a pretreatment. The decreased live birth and increased early miscarriage rate in the two ACs might due to the absence of a corpus luteum (CL). Indeed, it has been reported that pregnancies achieved in the absence of a CL are at higher risk of adverse obstetric and perinatal outcomes, such as hypertensive disorder, preeclampsia, cesarean section delivery, low birth weight (LBW) and macrosomia [9,11,12,20,21]. Relaxin is~6kDa peptide hormone secreted by CL and plays a key role in the transformation of the maternal circulation during early pregnancy especially before the "corpus luteal-placental shift" [22]. Absence of CL results in undetectable level of relaxin, as well as decreased levels of certain angiogenic and immunoregulatory factors, leading to insufficient cardiovascular  adaptation and adverse pregnancy outcomes [23,24]. The mechanism of GnRH-a pretreatment in improving pregnancy outcomes in AC is unclear. It is proposed that GnRH-a pretreatment may suppress untimely rises of progesterone levels during hormonal supplementation, which may advance the endometrium and hamper pregnancy outcomes [25]. Besides, animal study suggested that GnRH-a up-regulated the uterine expression levels of key receptivity markers including Hoxa10, Hoxa11, Lif and integrin b3 mRNA and protein, as well as increased the abundance of pinopodes in adenomyosis, therefore restoring endometrial receptivity [26].
Another intriguing finding of our study was that same endometrial protocol using in different age groups resulted in different reproductive outcomes. In younger women, there seemed to be a mild favor of AC+GnRH-a protocol as the early miscarriage rate in NC was only significantly lower than that in AC, but comparable to that in AC+GnRH-a. While the incidences of clinical pregnancy, ongoing pregnancy and live birth were slightly higher in AC+GnRH-a, although no reached statistical significance. Comparative proteomic analysis indicated that GnRH-a could affect 87 proteins expression, of which downregulated proteins were associated with energy metabolism and upregulated proteins were linked to cytoskeleton maintenance [27]. Younger patients present more vigorous metabolism than the older. It is possible that the downregulation of energy-metabolism proteins under GnRH-a treatment exerts a positive effect on the endometrium receptivity of younger women, but a negative effect on that of the older. In older women, the priority of NC to both AC groups was remarkable, as the ongoing pregnancy and live birth rates of NC were much higher, and early miscarriage rate was much lower in NC compared to AC, either with or without GnRH-a pretreatment. A recent study by Liu J and coworkers [28] suggested that in women aged 38 years or over, the endometrial preparation protocols did not affect FET outcomes. However, the study was limited by the small sample size as only 457 cycles with advanced age were included. Advanced maternal age is an independent risk factor of thrombotic events [29], exogenous hormone is associated with increased risk of vascular thrombosis [30]. Moreover, premature estradiol elevation may lead to apoptosis of trophoblast and is associated with uteroplacental insufficiency, hence further worsen the pregnancy outcomes in older women using AC protocols [31].

Strengths and limitations
This study has several strengths. First, the large sample size enhances the statistical power. Second, the clinical and laboratory practices did not substantially change over the course of study, which should minimize the possible confounders associated with pregnancy outcome. Third, we adjusted for more potential confounders that might otherwise have biased the findings, and used proper statistical methods to ensure structural stability. In addition, this study has good representativeness as we avoid strict inclusion and exclusion criteria. This study is mainly limited by its retrospective nature, and we could not control all the confounders. Therefore, prospective randomized trials are warranted to verify these findings.

Conclusions
In conclusion, our study reveals that NC protocol is associated with lower early miscarriage rate in overall population. There is a mild favor of GnRH-a cycles in younger women, while the priority of NC protocol is remarkable in the older. Maternal age should be a considerable factor when determining endometrial preparation method for FET.