Introduction
Frozen embryo transfers (FET) are increasingly common (De Geyter et al., Reference De Geyter, Calhaz-Jorge, Kupka, Wyns, Mocanu, Motrenko, Scaravelli, Smeenk, Vidakovic and Goossens2018) due to the modification of clinical protocols favouring freezing in patients at risk of hyperstimulation and from improvements in biological techniques such as cryopreservation and oocyte vitrification. The endometrium is prepared to allow transferred embryo implantation via various endometrial preparation protocols. In a stimulated cycle, endometrial preparation is performed through ovulation induction using follicle stimulating hormone (FSH). Ovulation is triggered with human chorionic gonadotropin (hCG) when there is a main follicle, producing a corpus luteum that secretes progesterone in the luteal phase and can be sustained with exogenous progesterone (Labrosse et al., Reference Labrosse, Lobersztajn, Pietin-Vialle, Villette, Dessapt, Jung, Brussieux, Bry-Gauillard, Pasquier and Massin2020). This method requires regular monitoring. In a natural (spontaneous) cycle, endometrial preparation is not induced; ovulation can be natural or induced by an hCG trigger. This method requires regular monitoring and is complicated when the transfer falls on a weekend. In an artificial cycle (AC), the endometrium is prepared by oral or transdermal estradiol to stimulate the growth of the endometrium, whilst blocking the patient’s gonadotropic pathway, and administering progesterone in the luteal phase to differentiate the endometrium. This method is easier to monitor and manage. The choice of protocol depends on the existence or not of an ovulatory cycle, the presence of endometriosis and adenomyosis, and the patient’s preference.
There is no consensus on which type of preparation is best, and implantation rates per embryo are comparable between the methods (Peeraer et al., Reference Peeraer, Couck, Debrock, De Neubourg, De Loecker, Tomassetti, Laenen, Welkenhuysen, Meeuwis, Pelckmans, Meuleman and D’Hooghe2015). However, results on the rate of pregnancy loss with substituted cycles are contradictory (Ghobara et al., Reference Ghobara, Gelbaya and Ayeleke2017; Mackens et al., 2017; Tomás et al., Reference Tomás, Alsbjerg, Martikainen and Humaidan2012), (Mackens et al., Reference Mackens, Santos-Ribeiro, van de Vijver, Racca, Van Landuyt, Tournaye and Blockeel2017) there is no apparent difference in the rate of ongoing pregnancies or birth rate according to cycle type (Peigné et al., Reference Peigné, Devouche, Ferraretto, Gricourt, Luton, Patrat and Epelboin2019). However, a recent retrospective study on a large multicentre cohort showed a significant increase in early pregnancy loss when using AC for endometrial preparation before FET (Vinsonneau et al., Reference Vinsonneau, Labrosse, Porcu-Buisson, Chevalier, Galey, Ahdad, Ayel, Rongières, Bouet, Mathieu d’Argent, Cédrin-Durnerin, Pessione and Massin2022). Some authors have shown that the AC resulted in fewer clinical pregnancies when the progesterone level on the day of the FET was lower, especially <9.2 ng/ml (Cédrin-Durnerin et al., Reference Cédrin-Durnerin, Isnard, Mahdjoub, Sonigo, Seroka, Comtet, Herbemont, Sifer and Grynberg2019; Labarta et al., Reference Labarta, Mariani, Holtmann, Celada, Remohí and Bosch2017). Another team recently proposed a higher progesterone threshold of 20.6 ng/ml (Boynukalin et al., Reference Boynukalin, Gultomruk, Turgut, Demir, Findikli, Serdarogullari, Coban, Yarkiner and Bahceci2019). Other studies have demonstrated better clinical pregnancy rates if the progesterone level on Day 10 of the transfer was >35 nmol/l (Alsbjerg et al., Reference Alsbjerg, Thomsen, Elbaek, Laursen, Povlsen, Haahr and Humaidan2018), >50 nmol/ml on Day 16 (Basnayake et al., Reference Basnayake, Volovsky, Rombauts, Osianlis, Vollenhoven and Healey2018), or >7.9 ng/ml in the middle of the luteal phase (Arce et al., Reference Arce, Balen, Platteau, Pettersson and Andersen2011). A recent meta-analysis found a cut-off of 10 ng/ml for the luteal serum concentration of progesterone, although there is no consensus on when to measure progesterone or the optimum threshold (Melo et al., Reference Melo, Chung, Pickering, Price, Fishel, Khairy, Kingsland, Lowe, Petsas, Rajkhowa, Sephton, Tozer, Wood, Labarta, Wilcox, Devall, Gallos and Coomarasamy2021).
Our aim was to compare progesterone levels on the day of transfer according to endometrial preparation protocol (spontaneous, stimulated or AC). The secondary objectives were to characterize normal progesterone levels depending on the day of embryo transfer, to observe the rate of pregnancies, live birth and pregnancy loss depending on the preparation protocol, and to determine the progesterone level threshold for clinical pregnancies.
Materials and methods
Study design and participants
This was a retrospective observational study on data from patients who had undergone an FET cycle at Nîmes University Hospital from May 2019 to December 2019. Medical and sociodemographic data were collected using medical JFIV® software. The study was approved by the local Institutional Review Board (IRB no. 20.02.01). Patients had previously agreed to the use of their data in future research. All patients who did not respond to the non-opposition letter were studied. Patients without a progesterone level test on the day of the transfer were excluded from the study.
Embryo freezing, thawing and transfer
Patients were grouped according to endometrial preparation protocol: spontaneous cycle (SC), ovarian stimulation (OS) or artificial cycle (AC). For endometrial preparation protocols, see Methods S1.
Frozen embryos at the cleavage or blastocyst stage were transferred between Days 2–5. The embryos were frozen by vitrification (Vit Kit-Freeze, Irvine Scientific®, Paris, France) according to the manufacturer’s procedure (Wirleitner et al., Reference Wirleitner, Vanderzwalmen, Bach, Baramsai, Neyer, Schwerda, Schuff, Spitzer, Stecher, Zintz and Zech2013). After thawing, embryos were eligible for FET if >50% of the cells were intact on Day 2 or 3. Blastocyst-stage embryos were eligible for transfer if <25% lysis was observed. The development stage and the number of embryos transferred were determined on a case-by-case basis, up to a maximum of three embryos.
Outcomes
The primary outcome was progesterone level (ng/ml) on the day of embryo transfer. Samples were taken at 10 a.m. The demographic variables analyzed were body mass index (BMI), the type (primary/secondary) and duration of infertility, cause of infertility (tubal, ovulatory, male factor, endometriosis), smoking status and hormone levels on Day 3 [FSH, luteinizing hormone (LH), estradiol and anti-Müllerian hormone (AMH)] and day of transfer (cleavage stage or blastocyst).
The embryonic development stage was assessed at the time of transfer (early transfer on Days 2–3, cleavage or prolonged culture on Days 4–6), the number of embryos transferred, the maternal age at freezing and thawing, the thickness of the endometrium before progesterone treatment, and the levels of LH and estradiol on the day of transfer were also recorded.
Positive pregnancy was considered for a positive hCG blood test (>100 UI/l). Progressive pregnancy with heartbeat activity detected at 7 weeks using ultrasound was classed as ‘clinical pregnancy with fetal heartbeat’. A pregnancy documented by positive hCG without fetal heartbeat at 7 weeks of gestational age was classed as ‘pregnancy loss’. A ‘live birth’ was considered as the birth of at least one living baby after 22 weeks of gestation.
We compared, according to the endometrial preparation method, the serum progesterone level at the end of the FET cycle in terms of pregnancy rate per cycle, the rate of pregnancy loss per pregnancy and the rate of clinical pregnancies with fetal heartbeat per cycle (Zegers-Hochschild et al., Reference Zegers-Hochschild, Adamson, Dyer, Racowsky, de Mouzon, Sokol, Rienzi, Sunde, Schmidt, Cooke, Simpson and van der Poel2017).
Data sources/measurements
Progesterone, LH and estradiol levels were analyzed via electrochemiluminescence using the Elecsys Progesterone III, Elecsys LH and Elecsys Estroadiol III assays, respectively, performed on a Cobas e801 (Roche Diagnostics) analyzer. The Day 3 hormone tests (FSH, LH, estradiol and AMH) were either performed in our laboratory or in a community laboratory. Endometrial thickness was measured using an S10 ultrasound machine (Voluson, GE Healthcare) and a transvaginal probe.
Bias
To reduce the risk of inclusion bias, we included all patients with known progesterone levels during the inclusion period. We performed a logistic regression analysis to limit the effect of possible differences in the population studied.
Sample size
Based on a previous study in which women received 200 mg vaginally administered progesterone twice a day (versus three times per day here), we estimated a Cmax progesterone level with a substituted cycle of 13 ± 4 ng/ml 18. The expected average progesterone level at the middle of the luteal phase with an SC was 19.4 ± 6.4 ng/ml. Therefore, the number of subjects necessary to observe a difference of 6 ng/ml with an alpha risk of 5% and a power of 90% in a unilateral test was 19 subjects in each group.
Statistical methods
Quantitative variables are described by mean ± standard deviation or median and interquartile range, whilst qualitative variables are described by percentages. Means were compared by parametric analysis of variance (ANOVA) test or Kruskal–Wallis test, and percentages were compared using the chi-squared test or non-parametric Fisher’s exact tests, as appropriate. Patients with missing variables for the primary outcome were not included in the study. If a variable was missing for a secondary outcome measure, the patient was excluded from the analysis.
When the association of the quantitative and qualitative variables (three modalities) was significant, multiple comparisons (2 × 2) were made with alpha risk adjustment using the Benjamini–Hochberg method. A multinomial logistic regression model was used to evaluate the primary endpoint and we performed adjustments for age, dysovulation, BMI and AMH, because these factors could affect the choice of preparation protocol (Yu et al., Reference Yu, Ma, Wu, Li, Tang, Li and Deng2015). A second model was made to account for the number of follicles to make sure that the difference in progesterone level was not linked to a difference in the number of follicles, only in the spontaneous versus stimulated groups.
A P-value <0.05 was considered to be statistically significant. Calculations were performed using SAS (SAS Institute, Cary, NC, USA) version 9 and R version 3.6.1 (R Development Core Team, 2019) and R® statistical software (version 3.6.1; Foundation for Statistical Computing, Vienna, Austria).
Results
Participants
During the inclusion period, 226 patients underwent FET, of whom 49 underwent a natural cycle (SC group), 48 a stimulated cycle (OS group), and 129 an artificial cycle preparation (AC group; Figure S1). Twenty-nine patients in the SC group, 24 patients in the OS group and 64 patients in the AC group were excluded, largely due to missing progesterone concentration levels measured on the day of the transfer due to staffing levels. Therefore, 20 patients were eligible for inclusion in the SC group, 27 patients in the OS group and 65 patients in the AC group.
No difference was seen between the three groups concerning the number of embryos transferred early or at the blastocyst stage, and there was no significant difference in patient age at embryo freezing or transfer (Table 1). BMI also did not differ significantly between groups.
Table 1. Baseline patient characteristics
More patients in the AC group had ovulatory infertility (35% vs 11 % in the SC group and 20% in the OS group; P = 0.04; Table 2). In contrast, the type of infertility, duration of infertility and the baseline hormone levels, the thickness of the endometrium and the number of embryos transferred were not different between groups.
Table 2. Demographic data of the study population
AMH: anti-Müllerian hormone; FSH: follicle stimulating hormone; LH: luteinizing hormone; MD: missing data.
Primary outcome: progesterone level on the day of transfer according to endometrial preparation protocol
Serum progesterone levels on the day of transfer were significantly higher in the OS group, whilst estradiol was significantly higher in the AC group and LH was highest in the SC group (Figure 1, Table 3). The median progesterone level in the AC group was 13.2 ng/ml (50th percentile), the 30th percentile in our population was for patients with a serum progesterone level of 9.82 ng/ml, and the 10th percentile for patients with a serum progesterone level of 6.38 ng/ml.
Figure 1. Box plots showing serum progesterone levels (ng/ml) by endometrial preparation protocol in which diamonds denote the mean, horizontal lines denote median, and boxes show the interquartile range (IQR).
Table 3. Hormone levels on day of transfer and pregnancy outcomes
* Missing data for nine patients in the SC group, 19 patients in the OS group and 38 patients in the AC group.
Univariate analysis
The difference between the groups was significant in the univariate model (P < 0.0001). Increasing the progesterone level by 1 ng in a univariate analysis showed an OR of 1.15 [95% CI (1.08; 1.22), P < 0.0001] of the patient being in the stimulated group and 1.09 [95% CI (1.02; 1.15), P = 0.0073] of being in the spontaneous group compared with the artificial group.
Progesterone level did not differ according to cleavage stage versus morula–blastocyst stage, regardless of the protocol (Figure 3 and Table 4).
Table 4. Progesterone levels according to transfer day and type of endometrial preparation
Multivariate analysis
The results did not change significantly after adjustment for AMH and dysovulation and BMI (Table 5). We found an adjusted OR of 1.15 [95% CI (1.08; 1.23), P < 0.0001] for the patients in the stimulated group and 1.08 [95% CI (1.01; 1.15), P = 0.02] in the spontaneous group compared with the artificial group (Table 5).
Table 5. Multivariate analysis
Secondary outcomes
We observed no between-group difference in rates of pregnancy (40% SC, 22.2% OS versus 29.8% AC; P = 0.42), pregnancy loss (12.5% SC, 30% OS versus 43% AC; P = 0.9) or clinical pregnancy with fetal heartbeat (35% SC, 14.8% OS versus 16.4% AC, P = 0.21; Table 3). In all patients, no significant difference was found in progesterone levels on the day of transfer between patients with clinical pregnancy with fetal heartbeat versus those experiencing pregnancy loss or no pregnancy (17.49 ng/ml vs 20.83 ng/ml respectively, P = 0.7; Figure 2).
Figure 2. Box plots showing serum progesterone level (ng/ml) by pregnancy outcome: no pregnancy (no clinical pregnancy/pregnancy loss) versus live birth (single/multiple). Diamonds denote the mean, horizontal lines denote median, and boxes show the interquartile range (IQR).
Figure 3. Box plots showing serum progesterone level (ng/ml) according to transfer day and type of endometrial preparation. Circles and crosses denote the mean, horizontal lines denote median, and boxes show the interquartile range (IQR).
Effect of number of follicles on progesterone level and endometrial preparation method
To ensure that the difference in progesterone levels observed between spontaneous and stimulated cycles was not due to a higher number of follicles in the stimulated cycle (and therefore the corpora lutea), we performed a multivariate analysis of progesterone levels according to the number of follicles on the day of induction (for the SC and OS groups). The adjustment did not modify the effect of the group of membership on the progesterone level.
Discussion
Our study shows a significant difference in serum progesterone levels depending on the type of endometrial preparation. Endometrial preparations during a natural or stimulated cycle, with a corpus luteum secreting progesterone, resulted in higher serum progesterone levels. This luteal insufficiency with endometrial preparations with an artificial (substituted) cycle, without a corpus luteum, marked by lower progesterone levels on the day of transfer, might partly explain the higher rate of pregnancy loss observed in these patients in certain studies (Tomás et al., Reference Tomás, Alsbjerg, Martikainen and Humaidan2012). However, we observed no significant difference in the rate of pregnancy loss or live birth rate.
Following the work of Labarta et al. (Labarta et al., Reference Labarta, Mariani, Holtmann, Celada, Remohí and Bosch2017; Labarta, Reference Labarta2019), many teams measure progesterone in the luteal phase of ACs in clinical practice (Alsbjerg et al., Reference Alsbjerg, Thomsen, Elbaek, Laursen, Povlsen, Haahr and Humaidan2018; Arce et al., Reference Arce, Balen, Platteau, Pettersson and Andersen2011; Basnayake et al., Reference Basnayake, Volovsky, Rombauts, Osianlis, Vollenhoven and Healey2018). This is controversial. The receiver operating characteristic (ROC) curve of progesterone predictive of ongoing pregnancy in patients with an AC showed a threshold value of 11 ng/ml (Labarta et al., Reference Labarta, Mariani, Holtmann, Celada, Remohí and Bosch2017). However, the area under the curve (AUC) was 0.59, a very low score, questioning the clinical significance of this marker and showing a weak discriminating property of the progesterone threshold. The same team recently found a threshold of 8.8 ng/ml, in which 31% of the total population had a progesterone level below the target (Labarta et al., Reference Labarta, Mariani, Paolelli, Rodriguez-Varela, Vidal, Giles, Bellver, Cruz, Marzal, Celada, Olmo, Alamá, Remohí and Bosch2021). Therefore, Labarta et al. found that 30% of patients had insufficient luteal phase impregnation and would benefit from a modification of progesterone support. In our study, the 30th percentile was 9.82 ng/ml, which is consistent with the work of several teams. Having such a large proportion of the population below the cut-off value is concerning; rather than measuring the progesterone level in clinical practice, another type of endometrial preparation might be more appropriate. Here, we showed that a spontaneous or stimulated cycle produces higher progesterone levels. Alternatively, we could modify our AC protocol to obtain higher serum progesterone levels for the whole population.
Increasing the doses of vaginal progesterone did not noticeably increase the serum progesterone level. Indeed, it is the method of application that affects serum progesterone level, with intramuscular application of subcutaneous progesterone providing a greater effect (Paulson et al., Reference Paulson, Collins and Yankov2014). Preliminary data suggest that when the progesterone level is below 9 ng/ml on the day of transfer, subcutaneous progesterone can achieve a clinical pregnancy rate comparable with patients with a normal level of serum progesterone 20. Indeed, according to Sator and colleagues (Sator et al., Reference Sator, Radicioni, Cometti, Loprete, Leuratti, Schmidl and Garhöfer2013), the maximum concentration of progesterone, obtained at 0.92 h after the injection of 25 mg of subcutaneous progesterone, was 57.84 ng/ml. The half life was 13.06 h, which guarantees sufficient progesterone levels with an injection every 24 h.
Dydrogesterone could also be added orally in the luteal phase. Oral dydrogesterone is non-inferior compared with vaginally applied micronized progesterone in in vitro fertilization using fresh embryos (Tournaye et al., Reference Tournaye, Sukhikh, Kahler and Griesinger2017). Few studies have investigated pregnancy rates with luteal phase support with dydrogesterone, and the results do not appear to differ from luteal phase support, in a modified natural cycle, with vaginal micronized progesterone (Ozer et al., Reference Ozer, Yuksel, Yucel Cicek and Kahraman2021). Furthermore, the application of dydrogesterone does not increase serum progesterone levels, as it is not easy to measure in routine care. It is therefore difficult to monitor and confirm its value in the luteal phase.
Whilst most studies show no difference in birth rate irrespective of the endometrial preparation protocol, they are mainly retrospective studies or studies with small numbers pooling ovulating and anovulating patients. One recent study showed an improved birth rate with a SC versus AC in patients without ovulatory infertility (Singh et al., Reference Singh, Reschke, Segars and Baker2020). Finally, the absence of corpus luteum in protocols with substituted cycles could lead to a risk of obstetric complications, especially pre-eclampsia (von Versen-Höynck et al., Reference von Versen-Höynck, Schaub, Chi, Chiu, Liu, Lingis, Stan Williams, Rhoton-Vlasak, Nichols, Fleischmann, Zhang, Winn, Segal, Conrad and Baker2019), postpartum haemorrhage, and caesarean section (Ginström Ernstad et al., Reference Ginström Ernstad, Wennerholm, Khatibi, Petzold and Bergh2019). The limitation of our study was that the sample size was too small to detect a difference in the rate of clinical pregnancy, pregnancy loss or live birth rate depending on the protocol.
In conclusion, the serum progesterone level on the day of transfer was significantly lower for an AC preparation compared with a spontaneous or stimulated cycle preparation in FET. Further studies are required to evaluate the effect of a lower progesterone level in the luteal phase at the time of FET and to evaluate whether modifying the protocol with an AC could correct the progesterone level in the luteal phase.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0967199423000163
Acknowledgements
We wish to thank Genevieve Pascal, the midwife in our unit, Teresa Sawyers for translating the manuscript and Thierry Chevalier for methodological advice.
Financial support
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Conflict of interest
The authors declare none.
Ethical standards
The authors assert that all procedures contributing to this work complied with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
