Strengthen Luteal Phase Support for the Patients With Low Serum Progesterone on the Day of Embryo Transfer in Articial Endometrial Preparation Cycles: a Large-sample Retrospective Trial

Background: The low serum progesterone (P) on the day of frozen embryo transfer (FET) is associated with diminished pregnancy rates in articial endometrium preparation cycles using vaginal micronized P, but it is no consensus whether the strengthened luteal phase support (LPS) for the patients with low P on the FET day in articial cycles is benecial. A single centric, large-sample retrospective trial was aimed to investigate the contribution of strengthened LPS to the pregnancy outcomes for the groups of low P levels on the FET day in articial endometrium preparation cycles. Methods: Women who had undergone rst articial cycle for endometrium preparation after freeze-all in our clinic during 2016 and 2018 were classied into two groups depending on the serum P levels on the FET day, routine LPS was administered for group B (P ≥ 10.0ng/ml on the FET day, n=1261) and strengthened LPS (routine LPS+ im P 40mg daily) for group A (P <10.0 ng/ml on the FET day, n=1295), the primary endpoint was the live birth rate and secondary endpoints were clinical pregnancy, miscarriage and neonatal outcomes. Results: The results showed the clinical pregnancy rate in group A was lower than group B (48.4% vs 53.2%, aRR 0.81, 95% CI 0.68, 0.96), the miscarriage rate was similar between the two groups (16.0% vs 14.7%, aRR 1.09, 95%CI 0.77, 1.54). The live birth rate was slightly lower than group B (39.5% vs 43.3%, aRR 0.84, 95%CI 0.70, 1.0). The birthweights and other neonatal outcomes were found no difference between the two groups (P>0.05). Conclusions: The strengthened LPS for the section of patients of low P levels on the FET day might help to have a reasonable pregnancy outcome, although the live birth rate was slightly lower than the groups with normal serum P levels on the FET day and usage of routine LPS. Trial registration: no available.


Introduction
Nearly 5 million babies have been delivered following assisted conception (IVF/ICSI) and the demand is increasing. Meticulous ovarian stimulation and well programmed luteal phase support (LPS) are the landmarks of treatment success. Although the importance of LPS in IVF/ICSI cycles is well established, the optimal route, dose and duration of this support is still a matter of debate [1]. Arti cial endometrial preparation for frozen embryo transfer (FET) cycle is different from stimulated IVF cycle in that there is no endogenous progesterone (P) production, therefore instead of luteal phase supplementation, there is a need for luteal phase "creation" or replacement [2]. In the context, arti cial cycles provide us a chance to explore the optimal P amount for supporting embryo implantation without the interferences with endogenous P in stimulation cycles [3]. P administration for LPS has multiple routes including oral, intramuscular, vaginal and rectal routes. Oral route undergoes extensive rst-pass metabolism which limits its e cacy for luteal support. Vaginal administration is preferred over intramuscular injection for its uterine rst-pass effect, its convenience and good tolerability [3]. However, body mass index (BMI) and vaginal environment affect serum levels after vaginal route, the serum P concentrations show a marked inter-individual difference despite the similar dose of P administration by vaginal route [4]. Low serum P levels on the FET day in arti cial cycles using vaginal P have been reported to be associated with poorer reproductive outcomes [5][6][7], however, it is no consensus whether the strengthened LPS is bene cial for the patients with low P levels in arti cial cycles [8,9]. The debate of using one-size-t-all regimen or individualized protocols in arti cial endometrial preparation cycles has still continued.
In our clinic, the serum P levels on the FET day was routinely monitored in arti cial endometrium preparation cycles, and for the cases with low P levels (< 10.0 ng/ml) on the FET day, we added intramuscular P 40 mg daily to salvage the FET cycles. Aimed to investigate the contribution of strengthened LPS for the women of low P levels on the FET day, we performed a single centric, largesample retrospective trial to compare the pregnancy outcomes of arti cial cycles between the two treatments.

Study design and population
A retrospective study was conducted at the Department of Assisted Reproduction of the Ninth People's Hospital of Shanghai Jiao Tong University School of Medicine. Women who had undergone rst arti cial cycle for endometrium preparation in a freeze-all policy during the period from January 2016 to December 2018 were enrolled. Exclusion criteria were as follows: patients older than 42 years of age, a history of recurrent miscarriages or recurrent implantation failure (e.g. unsuccessful transfer of ≥ 3 times), with systemic diseases, uterine diseases (e.g. broids and congenital uterine malformation) or hydrosalpinx. The cycles without measuring serum P levels on the FET day were excluded. Only the rst arti cial cycle in our clinic was included for each patient, although some patients had previous FET failures elsewhere.
This study protocol was approved by the ethical committee of the hospital and was carried out in accordance with the Helsinki Declaration. Due to the retrospective nature, informed consent was not required and patients' data were used anonymously.
Frozen embryo transfer and luteal phase support For endometrium preparation in FET cycles, oral 17β-estradiol (Fematon red tablets 4 mg, twice daily; Abbott Healthcare Products B.V.) or ethiny oestradiol 25ug three times daily was commenced on the third day of a natural or progesterone withdrawal menstrual cycle. After 12-14 days, vaginal ultrasound examination was performed, when the endometrial thickness reached ≥7 mm, ultrasound detected quiescent ovaries and serum P level was <1.0 ng/ml, P transformation was initiated using fematon yellow tablets (containing estradiol 4 mg and dydrogesterone 20mg twice daily, Abbott Healthcare Products B.V) and vaginal micronized progesterone capsule 200mg twice daily (Utrogestan, Laboratoires Besins International, France). Embryo transfer was performed 3 days after P administration for cleavagestage embryos or 5 days later for blastocyst transfer.
All patients completed the serum hormone examination of P and estradiol at 8:00-9:00 of transfer day, and the last dose before FET was administered on the morning of the FET day. The vitri ed-warmed embryo transfers were arranged at 13:30-15:30. Based on the serum P levels of the FET day, patients were classi ed into the two groups. For the cases with the serum P levels ≥ 10.0ng/ml, the mentioned luteal phase support of dydrogesterone and vaginal P continued to use (Group B: normal P+ routine LPS). For the cases with the serum P levels < 10.0ng/ml, the intramuscular P 40mg daily was added to strengthen luteal phase support after FET (Group A: low P+ strengthened LPS). LPS was stopped if the serum hCG test was negative on the 14 th day after FET. If the serum hCG test was positive, the LPS was continued to 10 weeks of gestation.

Embryo quality assessment and vitri cation
The ovarian stimulation regimen included the GnRH antagonist protocol, GnRH agonist long protocol and progestin-primed ovarian stimulation. The embryo morphology assessment was evaluated on day 3, 5 and 6 after oocyte retrieval. Cleavage-stage embryos with at least 7 blastomeres and fragmentation <20% were regarded as high-quality embryos. Blastocysts were scored according to the Gardner and Schoolcraft grading system [10] and recorded as high quality if they reached at least an expansion stage 3 with A or B for inner cell mass and trophectoderm (3BB). In all FET cycles, up to 2 embryos can be transferred. The vitri cation and thawing procedure were previously described by Kuwayama et al [11]. Brie y, embryo vitri cation was carried out via a Cyrotop carrier system, in combination with DMSO-EG-S as cryoprotectants. For thawing, embryos were transferred into dilution solution in a sequential manner.

Outcome parameters and statistical methods
The primary outcome of the study was the live birth rate. Secondary endpoints included rates of implantation, biochemical pregnancy, clinical pregnancy, miscarriage and perinatal outcomes. Live birth was de ned as a live neonate born after 24 weeks of gestation. A clinical pregnancy was con rmed by the observation of a gestational sac on ultrasound scanning 4-5 weeks after embryo transfer. Miscarriage rate was de ned as a loss of clinical pregnancy before the 24th gestational week. The implantation rate was calculated as the number of gestational sacs visualized on ultrasound examination divided by the number of embryos transferred. All calculations were made on a per transfer cycle.
Low birth weight and macrosomia were de ned as birth weights < 2500 g and ≥ 4000 g, respectively. Preterm birth was de ned as a delivery before complete 37 gestational weeks. Pregnancy-related complications included gestational diabetes, intrahepatic cholestasis of pregnancy, pregnancy-induced hypertension and pre-eclampsia. All neonatal and delivery information were obtained from medical records or telephone interview by a trained nurse.
The baseline characteristics and associated clinical outcomes were compared via t test or chi-square test where appropriate. A multivariate logistic regression analysis was performed to determine the independent effect of serum P on reproductive outcomes after adjustment for possible confounding factors, including maternal age, BMI (underweight, normal weight, overweight and obesity), infertility duration, gravidity, parity, infertility cause (tubal, PCOS, male, endometriosis and other), previous IVF failures, number and stage of embryos transferred. We analyzed the pregnancy outcomes for the groups to evaluate the effects of strengthened LPS for the group with the low serum P on the FET day.
All statistical analyses were performed with the use of the Statistical Package for Social Sciences (SPSS) version 21.0. A P value of < 0.05 was considered to be statistically signi cant difference.

Results
A total of 7674 arti cial endometrium preparation cycles were screened and 2556 cycles met the inclusion criteria and underwent rst arti cial cycle for FET during the study period. Mean female age was 31.3 ± 3.5 years, BMI was 21.76 ± 3.13 kg/m 2 . The median estradiol levels on the FET day in arti cial cycles were 210.0 pg/ml (interquartile range 116.0 pg/ml). The median P levels on the FET day in arti cial cycles were 9.9 ng/ml (interquartile range 4.2 ng/ml). A total of 1261 women (49.3%) with serum P ≥ 10.0 ng/ml on the FET day received routine LPS (group B). A total of 1295 women (50.7%) were recorded with serum P < 10.0 ng/ml on the FET day, then added intramuscular P 40 mg daily from the FET day onwards, and they were classi ed into group A.
A ow diagram of the patient selection process is shown in Fig. 1. Baseline characteristics between the two groups are presented in Table 1. Women with group A were similar with group B in terms of age, infertility duration, gravity, parity, endometrium thickness, the number and stage of embryos transferred.
In group A, the BMI and AFCs were signi cantly higher (P < 0.05), and the basal FSH levels were slightly lower than group B (P < 0.05). Compared with the control, women in group A had more frequently experienced obesity (7.2% vs 4.6%), less proportion of underweights (9.4% vs 13.6%) and slightly higher proportion of PCOS (17.2% vs 13.6%, P < 0.05).  The comparison of neonatal birthweight was similar between the two groups, both in singleton birth (3366.4 ± 536.0 g vs 3317.0 ± 516.8 g, P > 0.05) and twin birth (2461.0 ± 460.1 g vs 2472.6 ± 472.2 g, P > 0.05) ( Table 3). The proportion of pregnancy-related complications, the other neonatal outcomes including the prevalence of preterm birth and low birthweight, were similar between the two treatment groups (P > 0.05). Low serum P on the FET day was a risk factor for the chance of pregnancy The risk factors associated with pregnancy outcomes were explored by logistic regression analysis (Table 4). When the clinical pregnancy was chosen as the dependent factor, and age, BMI (categorical variable), number of embryos transferred, embryo stage, infertility causes and serum P levels (categorical variable) were chosen as independent factors, the variants of age, the number and stage of embryos, infertility causes and P levels on the FET day were found to be signi cant independent prognosticators (P < 0.05). The age and number of transferred embryos had negatively in uential for the pregnancy outcome, the infertility causes of PCOS and male factor had higher chance of pregnancy than the tubal factor in this study (P < 0.05). The low P on the FET day (P < 10.0 ng/ml), in the context of strengthened LPS, still decreased the chance of clinical pregnancy by 19% after adjusting confounding factors. The multinomial logistic analysis, done to predict live birth rate, showed that only age, number and stage of embryos transferred were signi cantly related with live birth rate (P < 0.05). Low serum P values on the FET day slightly reduced the live birth rate but did not reach the signi cance (aRR 0.84, 95%CI 0.70-1.0).

Discussion
Clinicians frequently preferred the arti cial endometrial preparation for its advantage of easier programming of embryo transfer [3]. The increasing number of arti cial cycles raises the question of the serum P levels required to optimize the pregnancy outcome. In this large-sample retrospective study, we investigated the contribution of strengthened LPS for the pregnancy outcome in patients with low serum P on the FET day in arti cial cycles. The strengthened LPS for the section of patients produced 39.5% live birth in the groups of low serum P, a slightly lower (3.8%) than the groups with normal serum P levels and usage of routine LPS.
The principles for choosing LPS include of the minimum effective dose, good safety and tolerability [12].
This is an open question with many alternative answers, and the heterogeneous applications of the dose and routes make it di cult to compare LPS in different studies [3,13]. The pregnancy outcomes, especially live birth rate, become the endpoint of evaluating the effects of LPS in arti cial FET cycles.
Vaginal administration was the rst choice of doctors, used alone or in combination with oral or intramuscular injection in a recent large survey [14]. In our center, the most popular regimen of LPS was using oral dydrogesterone 20 mg twice daily and vaginal micronized P 200 mg twice daily in arti cial cycles. In this context, serum P level was a surrogate marker re ecting the systematic absorption extent of vaginal P administration. It might be interfered by individual variability, such as BMI, mucus surface area, amount of cervical mucus and the difference of vaginal microbiome [4]. Thus, a marked interindividual difference of serum P concentrations on the luteal phase was present despite the same dose of vaginal P administration, the possibility of LPS insu ciency by vaginal route can not be excluded for the section of patients with low serum P concentration in arti cial cycles. The previous report compared three arms in arti cial cycles using a randomized control trial (vaginal 200 mg twice daily, im P 50 mg daily only, vaginal 200 mg twice daily + im P 50 mg every third day), vaginal P only group was found a signi cantly lower ongoing pregnancy rate compared the other two groups (31% vs 50% vs 47%), and the trial did not continue after inter analysis for the higher proportion of biochemical pregnancy loss and miscarriage [15]. Unfortunately, no data on serum progesterone levels in the three groups were available from that study. The low serum P levels on the FET day in arti cial cycles using vaginal route were previously reported to be associated with poorer reproductive outcomes, the cutoff value was varied in previous reports (5.0-12.0 ng/ml) [5][6][7][8]. In this study, the cutoff value was set as 10.0 ng/ml.
In this study, the strengthened P replacement was a protective approach, with attempting to mitigate the effects of low serum P4 that fell below this threshold. For the topic, the ideal control was using routine LPS for the patients with low serum P on the FET day, but our previous tendency of doctors was to add P to avoid the possible harm and maximize the patients' bene ts in a conservative view. Limited by the realworld data in our clinic, we converted to use the population of normal serum P and routine LPS as controls, the current data showed slight lower pregnant outcomes of strengthened LPS in the lower P group. Although our study can not distinguish the difference origin from the harm caused by lower serum P or the bene ts of additional P, the current comparison data still provided some meaningful information. Firstly, the patients of low serum P and strengthened LPS showed a slightly lower clinical pregnancy and live birth rate compared with the normal control, it decreased the chance of clinical pregnancy by 19% after adjusting confounding factors. Similar results were reported in Alsbjerg et al [16], serum P levels below 11.0 ng/ml (35 nmol/l) decreased the chance of ongoing pregnancy with a risk reduction of 14% in hormone replacement therapy FET cycles. Using the logistic regression model in this study, the low serum P level, as categorical variable, was a risk factor for the chance of pregnancy in arti cial endometrial preparation cycles, which indicated the suboptimal condition in this section of patients with low serum P.
Secondly, the pregnancy outcome of the groups of low P and strengthened LPS was an acceptable result, clinical pregnancy rate (48.4%) and live birth rate (39.5%) appeared reasonable, the similar proportion of biochemical pregnancy loss and miscarriage also approved the e cacy of strengthened LPS. As such, even though serum P has been inadequate on the FET day, this was potentially remedied by additional P administration, reinforcing intervention might still be possible beyond the day of transfer. The results were in coincide with the recent report of Volovsky et al, P replacement enhanced the pregnancy outcome if the P on the FET day was lower than 8.0 ng/ml [8]. In a recent study of Polats et al., the intramuscular P supplementation every third day did not increase the ongoing pregnancy rate compared with vaginal P only in vitri ed blastocyst transfer cycles (48.3% vs 51.8%). However, the patients with serum P less than 8.75 ng/ml in vaginal P only group had a numerically lower ongoing pregnancy rate (28.6% vs 46.6%), but did not reach the statistically signi cant difference [9].
The current study also had several strengths. The vaginal P administration achieved higher endometrial tissue concentrations and lower systemic exposures than observed after intramuscular injection, and the dose-effect relationship was not obvious in the same route of P administration [4], so the combination of two or three routes, rather than increasing dose of single route, is a reasonable choice to meet the requirement of endometrial transfection in arti cial cycles. Multiple routes of P administration provided su cient luteal support in arti cial FET cycles. The secondary strength was the large cohort size, it had su cient power to answer the question whether it was bene cial of strengthened LPS to overcome the possible negative in uence in the cycles with low serum P on the FET day. All cycle data were derived from a single institution, where practice consistency can be assured.
Our study was limited by its retrospective design. In this context, we screened the database with strict inclusion criteria, analysis was restricted to rst arti cial cycles and normal ovarian reserve, and a number of potential confounders were well controlled in the current study. Additionally, the vast majority of patients in the present study were young and normal ovarian reserve, so the extrapolation to an unselected population needs to be validated.
In the previous report of Yovich et al., maternal age, embryo quality and mid-luteal serum progesterone levels were listed as three important factors governing the implantation rates for arti cial FET cycles [17]. In this study, serum P levels on the FET day, as a categorical variable, were found to be a signi cant independent prognosticator of clinical pregnancy, while it did not reach the statistically signi cance for live birth rate. The explanation of the results should be cautious. Strengthen LPS might be useful for the patients with low serum P levels on the FET day in arti cial cycles, further study is need to perform randomized controlled trials to evaluate the individualization of P dosage.

Conclusions
In summary, our study con rmed serum P levels on the FET day might be one of risk factors for the chances of pregnancy in arti cial cycles, and careful monitoring of serum P concentration was warranted. The strengthened LPS might help to have a reasonable pregnancy outcome in the section of low P on the FET day in arti cial cycles, although it was a slightly lower than the groups with normal serum P levels on the FET day and usage of routine LPS. Our study added new information to the elds of luteal phase support and further studies should explore P administration to optimize concentrations for individual women to improve pregnancy outcomes.

Declarations
Ethical Approval and consent to participate Human Ethics Committee approval for the retrospective trial of frozen embryo transfer cycles came from the Institutional Review Board (IRB) of the Shanghai Ninth People's Hospital on August 20, 2020 (Number: SH9H-2020-T268-1).

Consent for publication
All patients consent the data used for research and publication Availability of supporting data The data is not publicly shared and please contact author for data requests.

Competing interests
The authors have nothing to declare.

Funding
The study was funded by The National Natural Science Foundation of China (81671520) the Natural Science Foundation of Shanghai (1841196400 and 19411961100).
Authors' contributions QJC, QQH and LHS supervised the entire study, including procedures, conception, design, and completion.
HYG, JY and HJY contributed to the data analysis and manuscript drafting. All authors participated in the ultimate interpretation of the study data and in revisions to the article. The authors have nothing to declare.