Length of estradiol exposure >100 pg/ml in the follicular phase affects pregnancy outcomes in natural frozen embryo transfer cycles

Abstract

STUDY QUESTION

Do the length of follicular phase estradiol exposure and the total length of the follicular phase affect pregnancy and live birth outcomes in natural frozen embryo transfer (FET) cycles?

SUMMARY ANSWER

An estradiol level >100 pg/ml for ≤4 days including the LH surge day is associated with worse pregnancy and live birth outcomes; however, the total length of the follicular phase is not associated with pregnancy and live birth outcomes.

WHAT IS KNOWN ALREADY

An estradiol level that increases above 100 pg/ml and continues to increase is indicative of the selection and development of a dominant follicle. In programmed FET cycles, a limited duration of follicular phase estradiol of <9 days results in worse pregnancy rates, but a prolonged exposure to follicular phase estradiol for up to 4 weeks does not affect pregnancy outcomes. It is unknown how follicular phase characteristics affect pregnancy outcomes in natural FET cycles.

STUDY DESIGN, SIZE, DURATION

This retrospective cohort study included infertile patients in an academic hospital setting who underwent their first natural frozen autologous Day-5 embryo transfer cycle in our IVF clinic between 01 January 2013 and 31 December 2018. Donor oocyte and gestational carrier cycles were excluded.

PARTICIPANTS/MATERIALS, SETTING, METHODS

The primary outcomes of this study were pregnancy and live birth rates. Patients were stratified into two groups based on the cohorts’ median number of days from the estradiol level of >100 pg/ml before the LH surge: Group 1 (≤4 days; n = 1052 patients) and Group 2 (>4 days; n = 839 patients). Additionally, patients were stratified into two groups based on the cohorts’ median cycle day of LH surge: Group 1 (follicular length ≤15 days; n = 1287 patients) and Group 2 (follicular length >15 days; n = 1071 patients). A subgroup analysis of preimplantation genetic testing for aneuploidies (PGT-A) embryo transfer cycles was performed. Logistic regression analysis, adjusted a priori for patient age, number of embryos transferred, and use of PGT-A, was used to estimate the odds ratio (OR) with a 95% CI.

MAIN RESULTS AND THE ROLE OF CHANCE

In the length of elevated estradiol analysis, the pregnancy rate per embryo transfer was statistically significantly lower in patients with an elevated estradiol to surge of ≤4 days (65.6%) compared to patients with an elevated estradiol to surge of >4 days (70.9%; OR 1.30 (95% CI 1.06–1.58)). The live birth rate per embryo transfer was also statistically significantly lower in patients with an elevated estradiol to surge of ≤4 days (46.6%) compared to patients with an elevated estradiol to surge of >4 days (52.0%; OR 1.23 (95% CI 1.02–1.48)). In the follicular phase length analysis, the pregnancy rate per embryo transfer was similar between patients with a follicular length of ≤15 days (65.4%) and patients with a follicular length of >15 days (69.0%; OR 1.12 (95% CI 0.94–1.33)): the live birth rate was also similar between groups (45.5% vs 51.5%, respectively; OR 1.14 (95% CI 0.97–1.35)). In all analyses, once a pregnancy was achieved, the length of the follicular phase or the length of elevated oestradiol >100 pg/ml no longer affected the pregnancy outcomes.

LIMITATIONS, REASONS FOR CAUTION

The retrospective design of this study is subject to possible selection bias in regard to which patients at our clinic were recommended to undergo a natural FET compared to a fresh embryo transfer or programmed FET. To decrease the heterogeneity of our study population, we only included patients who had blastocyst embryo transfers; therefore, it is unknown whether similar results would be observed in patients with cleavage-stage embryo transfers. The retrospective nature of the study design did not allow randomized to a specific ovarian stimulation or ovulation trigger protocol. However, all patients were managed with the standardized protocols at a single center, which strengthens the external validity of our results when compared to a study that only evaluates one specific stimulation protocol.

WIDER IMPLICATIONS OF THE FINDINGS

Our observations provide cycle-level characteristics that can be applied during a natural FET cycle to help optimize embryo transfer success rates. Physicians should consider the parameter of number of days that oestradiol is >100 pg/ml prior to the LH surge when determining whether to proceed with embryo transfer in a natural cycle. This cycle-specific characteristic may also help to provide an explanation for some failed transfer cycles. Importantly, our findings should not be used to determine whether to recommend a natural or a programmed FET cycle for a patient, but rather, to identify natural FET cycles that are not optimal to proceed with embryo transfer

STUDY FUNDING/COMPETING INTEREST(S)

No financial support, funding, or services were obtained for this study. The authors do not report any potential conflicts of interest.

TRIAL REGISTRATION NUMBER

N/A.

Introduction

In the USA, embryo transfers are now most commonly performed in frozen embryo transfer (FET) cycles. In 2016, there were 95 024 FETs reported in the CDC National Summary Report (SART, 2017). A cryopreserved embryo is most commonly transferred in either a programmed cycle, when exogenous hormones are used to prepare the endometrium and luteal phase support is required, or in a natural cycle, when endogenous hormones prepare the endometrium for embryo transfer and a corpus luteum supports the embryo development with or without additional progesterone support (Sun et al., 2017; Devine et al., 2018). Overall, programmed and natural FET cycles result in comparable live birth rates, but there may be patient- or clinic-specific factors that could alter the success rates of each cycle type (Groenewoud et al., 2013; Yarali et al., 2016; Ghobara et al., 2017; Alur-Gupta et al., 2018; Kang, 2018).

Programmed FET cycles involve a scheduled treatment protocol that mimics the hormone exposure that occurs in a natural menstrual cycle. This type of cycle is generally preferable for patients with menstrual cycle irregularities or for patients for whom a planned embryo transfer date is required for medical or logistical reasons. However, the self-administration of daily i.m. progesterone injections for 8–10 weeks for luteal support and recent evidence of higher rates of hypertensive disorders in pregnancy and postpartum hemorrhage when compared to natural FET cycles are disadvantages of programmed FET cycles (Devine et al., 2018; Ginström Ernstad et al., 2019; von Versen-Höynck et al., 2019). For these reasons, natural FET cycles are preferred for patients with predictable ovulatory cycles.

In programmed FET cycles, Navot et al. (1989) reported that follicular phase oestradiol exposure ranging from 6 to 35 days resulted in a histologically normal secretory phase endometrium after the addition of progesterone. Further, prolonged follicular phase oestradiol exposure up to 28 days in programmed FET cycles does not appear to negatively impact implantation and live birth outcomes (Bourdon et al., 2018; Sekhon et al., 2019). In natural FET cycles, it is currently unknown whether a short or prolonged follicular phase, resulting in altered endometrial exposure to endogenous hormones such as oestradiol and inhibin B, could affect endometrial receptivity and embryo transfer outcomes. This information would assist in determining whether natural FET cycles are appropriate for all ovulatory women or if there are cycle factors that influence the success rates of implantation and the developing fetus. The objective of this study is to determine whether the length of follicular phase oestradiol exposure or the length of the follicular phase in the natural FET cycle influences the pregnancy outcomes of that cycle.

Materials and methods

Study population and design

This was a retrospective cohort study of patients who underwent their first natural cycle blastocyst FET at our IVF center between 01 January 2013 and 31 December 2018. Donor oocyte and gestational carrier cycles were excluded. Data were collected from our prospectively maintained IVF electronic medical record system. Missing data and key variables were verified within our electronic medical record. This study was approved by the institutional review board at Weill Cornell Medical Center with a waiver for informed consent.

Definition of study groups

A serum oestradiol level that increases above 100 pg/ml and continues to increase is indicative of the selection and development of a dominant follicle. A level of 100 pg/ml has been demonstrated to correlate with a leading follicle that measures greater than 12 mm in diameter (Segawa et al., 2015). For each patient in this study cohort, we calculated the number of days from the oestradiol level increasing above 100 pg/ml to the LH surge day. The median length of time for this to occur was 4 days (Day 1 was counted as the first day with an oestradiol level above 100 pg/ml with all subsequent measurements continuing to increase up to the LH surge day). Based on the median number of days from the oestradiol level increasing above 100 pg/ml to LH surge, patients were stratified into two groups: Group 1 (≤4 days) and Group 2 (>4 days). Patients who presented late in their follicular phase with an oestradiol level already >200 pg/ml were excluded from this analysis.

Additionally, the median cycle day of LH surge in the cohort was cycle Day 15. Therefore, patients were separately stratified into two groups: Group 1 (follicular length ≤15 days) and Group 2 (follicular length >15 days). The LH surge day in the natural FET cycle was determined by the treating physician based on serial serum LH, oestradiol and progesterone levels.

A subgroup of preimplantation genetic testing for aneuploidies (PGT-A) embryo transfer cycles was additionally included. This subgroup was stratified as described above by the number of days from the oestradiol level increasing above 100 pg/ml to the LH surge and by the follicular phase length. This subgroup was included to perform a subgroup analysis of the study groups that controls for embryo aneuploidy.

Demographics and outcomes

Key demographic variables were collected (Table I). Pregnancy was defined as at least an hCG level of >5 mIU/ml. Spontaneous abortion was defined as a failed pregnancy after the observation of at least one intrauterine gestational sac on ultrasound. Live birth was defined as delivery at ≥24 weeks of gestational age. Embryo morphology was categorized into four groups: excellent (≥3AA), good (3-6AB, 3-6BA, 1-2AA), average (3-6BB, 3-6AC, 3-6CA, 1-2AB, 1-2BA) and poor (1-6BC, 1-6CB, 1-6CC, 1-2BB) (Irani et al., 2017a).

Clinical protocols

Controlled ovarian stimulation, trigger timing and dose, oocyte retrieval, and embryo culture were performed per the standard protocols at our institution (Huang and Rosenwaks, 2014). The protocol for ovarian stimulation utilized GnRH antagonists or short GnRH agonist protocols (Surrey et al., 1998; Cheung et al., 2005). Clomiphene citrate or letrozole was added to the gonadotrophins in some GnRH antagonist protocols at the discretion of the treating physician (Tummon et al., 1992; Yarali et al., 2009). Patients with diminished ovarian reserve received priming in the mid-luteal phase of the cycle preceding the ovarian stimulation cycle with either 0.1-mg oestradiol patches (Climara 0.1 mg, Bayer HealthCare Pharmaceuticals Inc., Montville, NJ, USA) or oral contraceptive pills (Ortho-Novum, Janssen Pharmaceuticals Inc., Titusville, NJ, USA) (Dragisic et al., 2005). Once two leading follicles reached 17 mm in size (20 mm for clomiphene- or letrozole-based protocols), oocyte maturation was triggered with either hCG (Novarel [Ferring Pharmaceuticals Inc., Parsippany, NJ, USA] or Pregnyl [Merck, Whitehouse Station, NJ, USA]) according to a sliding-scale protocol for hCG dose (10 000 IU for oestradiol ≤1500 pg/ml; 5000 IU for oestradiol 1501–2500 pg/ml; 4000 IU for oestradiol 2501–3000 pg/ml; and 3300 IU for oestradiol >3000 pg/ml) or a dual trigger with hCG (1500 IU) and GnRH agonist (4 mg; leuprolide acetate injection [Sandoz Inc., Princeton, NJ, USA]). Approximately 35–37 h after administration of the ovulatory trigger, an oocyte retrieval was performed transvaginally under ultrasound guidance.

Oocyte insemination was performed by standard insemination or by ICSI when indicated. When ICSI was performed, the oocytes were mechanically denuded to assess nuclear maturity and mature oocytes were then injected with a single sperm. Embryos were cultured in sequential media using the Embryoscope (Vitrolife, Göteburg, Sweden) time-lapse system. Embryos were evaluated for fertilization on Day 1 and were morphologically graded on Days 3 and 5 (Veeck and Zaninovic, 2003). Embryos were then cryopreserved at the blastocyst stage. Blastocysts that were selected for PGT-A underwent trophectoderm biopsy and were then cryopreserved. The embryo with the best morphologic grade was then selected to be transferred in a subsequent FET cycle (Irani et al., 2017a).

Patients were recommended to undergo a natural FET cycle if they had regular or predictable ovulatory cycles. Patients that were not candidates for a natural FET cycle would undergo a programmed FET cycle (these patients did not meet inclusion criteria for this study). All uterine cavities were evaluated by hysterosalpingogram or saline infusion sonogram in a menstrual cycle prior to the FET to rule out structural abnormalities. In the natural FET cycle, patients were first evaluated with a mid-follicular transvaginal ultrasound (cycle Days 8–10 for a typical 28–30 day cycle) to evaluate for a leading follicle and measure the endometrial thickness. Endometrial thickness was measured in the midsagittal plane as the largest distance from the proximal endometrial-myometrial interface to the distal endometria–myometrial interface. Serum LH, oestradiol, and progesterone levels were measured in our endocrinology laboratory using the Siemens Immulite assay for each hormone (Siemens Medical Solutions USA Inc., Malvern, PA, USA). The intra-assay coefficient of variation for the LH, oestradiol and progesterone assays were all less than 10%. The inter-assay correlation coefficients for the LH, oestradiol and progesterone assays were all greater than 0.97. Hormone level and ultrasound evaluation were continued serially until a dominant follicle was measured at >14mm with an adequate endometrial thickness, at which point hormone levels only were measured daily until the LH surge was confirmed. The LH surge was defined as the first day with an LH level ≥17 mIU/ml (Irani et al., 2017b). The embryo transfer was performed 5 days after the LH surge day. A vaginal progesterone suppository (endometrin 100 mg twice daily [Ferring Pharmaceuticals Inc., Parsippany, NJ, USA]) was administered nightly, starting the day after the embryo transfer, for luteal phase support. Serum progesterone, oestradiol and hCG were measured 10 days after the embryo transfer. Patients with a normally rising hCG underwent a viability ultrasound between 16 and 30 days after embryo transfer (5–7 weeks of gestation). Follow-up ultrasounds were performed to monitor fetal growth and development as indicated. Patients initiated care with an obstetrician between 8 and 10 weeks of gestation.

During the study period, there were no significant changes to protocols used or embryo transfer outcomes. On 01 January 2017, our laboratory transitioned from array comparative genomic hybridization for PGT detection of chromosomal copy number imbalance to next-generation sequencing technology.

Statistical analysis

The primary outcomes of this study were pregnancy and live birth rates. Logistic regression analyses of the entire study cohort (the number of days with oestradiol above 100 pg/ml to LH surge groups and the follicular phase length study groups; Tables II and III, respectively) were adjusted a priori for patient age at the time of retrieval, number of embryos transferred, and use of PGT-A to estimate the odds ratio (OR) with a 95% CI among the study groups. A subgroup analysis of PGT-A embryo transfer cycles was performed to compare the number of days with oestradiol above 100 pg/ml to LH surge groups and the follicular phase length study groups (Tables IV and V, respectively). Logistic regression analysis of the PGT-A subgroups was adjusted a priori for patient age at the time of retrieval and number of embryos transferred to estimate the OR with a 95% CI. A separate subgroup analysis of single embryo transfer cycles was performed to compare the number of days with oestradiol above 100 pg/ml to LH surge groups and the follicular phase length study groups (Tables VI and VII, respectively). Logistic regression analysis of the single embryo transfer subgroups was adjusted a priori for patient age at the time of retrieval, morphologic grade of the embryo and use of PGT-A to estimate the OR with a 95% CI. Statistical analyses were performed using StataSE 16 (StataCorp LLC, College Station, TX, USA). A value of P < 0.05 was considered to indicate statistical significance.

Results

There were 1891 patients included in the days of elevated oestradiol cohort (≤4 days: 1052 patients and >4 days: 839 patients) and 2358 patients included in the follicular phase length analysis (follicular length ≤15 days: 1287 patients and follicular length >15 days: 1071 patients) (Fig. 1). Demographic characteristics of the patients stratified by days of elevated oestradiol are displayed in Table I. The mean ± SD age was 35.6 ± 4.3 years in Group 1 and 36.1 ± 4.1 years in Group 2. On the day of LH surge, Group 1 had a mean oestradiol level of 253.4 ± 96.8 pg/ml, a mean LH level of 42.2 ± 21.3 mIU/ml, and a mean progesterone level of 0.70 ± 0.31 ng/ml. Group 2 had a mean oestradiol level of 326.5 ± 121.4 pg/ml, a mean LH level of 41.8 ± 20.6 mIU/ml, and a mean progesterone level of 0.88 ± 0.34 ng/ml. The number of embryo transfers that were PGT-A screened was 319 (30.3%) in Group 1 and 316 (37.7%) in Group 2.

The pregnancy outcomes for patients stratified by the number of days measuring an oestradiol level >100 pg/ml to the LH surge are shown in Table II. The pregnancy rate per embryo transfer was statistically significantly lower in patients with an elevated oestradiol to surge of ≤4 days (65.6%) compared to patients with an elevated oestradiol to surge of >4 days (70.9%; OR 1.30 (95% CI 1.07–1.59)). The live birth rate per embryo transfer was also statistically significantly lower in patients with an elevated oestradiol to surge of ≤4 days (46.6%) compared to patients with an elevated oestradiol to surge of >4 days (52.0%; OR 1.24 (95% CI 1.03–1.50)). Among patients who achieved a pregnancy, there were no statistically significant differences between any of the pregnancy outcomes, including live birth rate per pregnancy (elevated oestradiol to surge of ≤4 days (71.0%) versus elevated oestradiol to surge of >4 days (73.3%); OR 1.08 (95% CI 0.84–1.40)).

The pregnancy outcomes for patients stratified by length of the follicular phase are displayed in Table III. The pregnancy rate per embryo transfer was similar between patients with a follicular length ≤15 days (65.4%) and patients with a follicular length >15 days (69.0%; OR 1.12 (95% CI 0.94–1.33)). The live birth rate was also similar between groups (45.5% vs 51.5%, respectively; OR 1.14 (95% CI 0.96–1.35)). Among patients who achieved a pregnancy, there were no statistically significant differences between any of the pregnancy outcomes (biochemical, spontaneous abortion, therapeutic abortion, ectopic, stillbirth or live birth).

The subgroup analysis results among PGT-A embryo transfer cycles are displayed in Tables IV (elevated oestradiol to LH surge groups) and V (follicular phase length groups). The pregnancy rate per embryo transfer was not statistically significantly different in patients with an elevated oestradiol to surge of ≤4 days (68.3%) compared to patients with an elevated oestradiol to surge of >4 days (75.0%; OR 1.42 (95% CI 1.00–2.01)). The live birth rate per embryo transfer was statistically significantly lower in patients with an elevated oestradiol to surge of ≤4 days (55.5%) compared to patients with an elevated oestradiol to surge of >4 days (63.9%; OR 1.45 (95% CI 1.05–2.00)). Among patients who achieved a pregnancy, there were no statistically significant differences between any of the pregnancy outcomes, including live birth rate per pregnancy (elevated oestradiol to surge of ≤4 days (81.2%) vs elevated oestradiol to LH surge of >4 days (85.2%); OR 1.35 (95% CI 0.82–2.21)).

Among PGT-A embryo transfer cycles stratified by the follicular phase length, the pregnancy rate was similar between patients with a follicular length of ≤15 days (70.0%) and patients with a follicular length of >15 days (73.8%; OR 1.21 (95% CI 0.88–1.65)). The live birth rate was also similar between groups (58.4% vs 61.2%, respectively; OR 1.10 (95% CI 0.82–1.47)). Among patients who achieved a pregnancy, there were no statistically significant differences between any of the pregnancy outcomes (biochemical, spontaneous abortion, therapeutic abortion, ectopic, stillbirth or live birth).

The subgroup analysis results among single embryo transfer cycles are displayed in Tables VI (elevated oestradiol to LH surge groups) and VII (follicular phase length groups). The pregnancy rate per embryo transfer was statistically significantly lower in patients with an elevated oestradiol to surge of ≤4 days (62.2%) compared to patients with an elevated oestradiol to surge of >4 days (69.5%; OR 1.32 (95% CI 1.05–1.66)). The live birth rate per embryo transfer was not statistically significantly different in patients with an elevated oestradiol to surge of ≤4 days (43.8%) compared to patients with an elevated oestradiol to surge of >4 days (50.8%; OR 1.24 (95% CI 1.00–1.55)). Among patients who achieved a pregnancy, there were no statistically significant differences between any of the pregnancy outcomes, including live birth rate per pregnancy (elevated oestradiol to surge of ≤4 days (70.4%) vs elevated oestradiol to LH surge of >4 days (73.0%); OR 1.07 (95% CI 0.79–1.43)).

Among single embryo transfer cycles stratified by the follicular phase length, the pregnancy rate was similar between patients with a follicular length of ≤15 days (62.4%) and patients with a follicular length of >15 days (67.2%; OR 1.10 (95% CI 0.89–1.35)). The live birth rate was also similar between groups (42.7% vs 50.4%, respectively; OR 1.16 (95% CI 0.95–1.42)). Among patients who achieved a pregnancy, there were no statistically significant differences between any of the pregnancy outcomes (biochemical, spontaneous abortion, therapeutic abortion, ectopic, stillbirth or live birth).

Discussion

The primary aim of this study was to identify whether cycle characteristics during a natural FET cycle affect the pregnancy and live birth outcomes per embryo transfer cycle. We identified that the length of time of oestradiol exposure of >100 pg/ml in the follicular phase does influence pregnancy and live birth outcomes. Patients with ≤4 days of elevated oestradiol to LH surge day had significantly lower pregnancy and live birth rates. However, the total length of the follicular phase in the natural FET cycle was not associated with pregnancy or live birth outcomes. In all analyses, once a pregnancy was achieved, the length of the follicular phase or the length of elevated oestradiol to LH surge no longer affected the pregnancy outcomes.

The effects of hormone exposure on FET pregnancy outcomes have been well studied in programmed FET cycles (Navot et al., 1989). Follicular phase oestradiol given for up to 4 weeks results in similar pregnancy outcomes compared to a more traditional regimen of 2 weeks of oestrogen (Bourdon et al., 2018; Sekhon et al., 2019). On the other hand, there is evidence that the endometrium has a minimum length of time that follicular phase oestrogen exposure is needed in order to optimize receptivity. Even when endometrial thickness is optimal, follicular phase oestrogen exposure of <9 days results in lower pregnancy rates in programmed FET cycles (Dougherty et al., 2017). Our findings in natural FET cycles support the literature from programmed FET cycles. We observed that women with a long follicular phase had comparable pregnancy outcomes to women with a follicular phase of ≤15 days. This finding is supported by previous research that observed similar clinical pregnancy rates in natural FET cycles in women with a total follicular phase length <21 days to women with a follicular phase length ≥21 days (Ying et al., 2020). However, our cohort of patients who had a short exposure to elevated oestrogen in the follicular phase prior to the natural LH surge, independent of the total follicular phase length, had worse pregnancy and live birth outcomes after embryo transfer.

In the subgroup analyses of PGT-A cycles and single embryo transfer cycles in the elevated oestradiol to LH surge groups, some associations observed in the primary analysis were mitigated. In the PGT-A cycles analysis, the increased live birth rate persisted in patients with >4 days of elevated oestradiol exposure; however, the pregnancy rate was no longer significantly different between groups. One explanation for this is that the subgroup analysis of PGT-A cycles eliminates unknown confounding variables related to embryo ploidy that could not be accounted for in the primary analysis. An alternative, and more likely, explanation is that the study cohort in the subgroup analysis is underpowered to observe the association of an increased pregnancy rate in patients with >4 days of elevated oestradiol exposure that was observed in the primary analysis. This rationale is supported by this subgroup analysis having less patients than the primary analysis, a similar OR, and a CI lower limit of 1.00. Similarly, in the subgroup analysis of single embryo transfer cycles, the increased pregnancy rate persisted in patients with >4 days of elevated oestradiol exposure, but the live birth rate was no longer significantly different between groups. Again, this is most likely due to the subgroup being underpowered as there were less patients than the primary analysis, a similar OR, and a CI lower limit of 1.00.

At the level of the endometrium, oestradiol is the principle hormone responsible for endometrial proliferation during the follicular phase, with receptors located in epithelial, stromal and myometrial cells (Press et al., 1986). Oestrogen receptor binding at this location controls both epithelial proliferation and expression of progesterone receptors, two necessary steps that prepare the endometrium for the luteal phase and potential implantation (Levy et al., 1980). It has been estimated that a threshold level of 50–100 pg/ml of oestradiol is necessary to initiate these downstream effects of oestradiol action (Key and Pike, 1988). It is also reasonable to hypothesize that this oestradiol threshold must be maintained for a certain duration of time in order for the endometrium to adequately proliferate and upregulate enough progesterone receptors to develop into a receptive endometrium once exposed to adequate levels of progesterone in the luteal phase. The association observed in our study that patients with a longer duration of oestradiol exposure >100 pg/ml prior to LH surge have increased pregnancy and live birth rates supports the hypothesis that an increased duration of elevated oestradiol exposure better prepares the endometrium for the luteal phase. Importantly, patients in our study cohort with an oestradiol level >100 pg/ml for ≤4 days had decreased, although still clinically acceptable, rates of pregnancy and live birth. The threshold of >4 days established here is not as simple as an adequate or not adequate duration of exposure. Patient characteristics, including genetic and epigenetic factors, may also contribute to the endometrial response to oestradiol exposure in the follicular phase.

This cycle-specific characteristic in natural cycles is important as it may provide both an explanation for some failed transfer cycles and may be helpful to use prospectively to identify natural FET cycles that are not optimal to proceed with embryo transfer. The subgroup analysis of PGT-A cycles yields helpful information with which to counsel patients after a failed euploid embryo transfer. It can be difficult to identify the cause of a failed euploid embryo transfer in which the endometrial lining and luteal progesterone were normal (Kliman and Frankfurter, 2019; Pirtea et al., 2021). This can make patient counseling challenging in this situation when few risk factors for a failed euploid embryo transfer have been identified (Ben Rafael, 2020). Our finding that ≤4 days of oestradiol of >100 pg/ml to the LH surge is associated with lower live birth outcomes in PGT-A cycles provides another explanation for why euploid embryos may fail to implant. Despite similar endometrial thickness measurements in our study cohort, the length of exposure to elevated oestradiol is an important factor in preparing the endometrium for implantation.

Importantly, our findings should not be used to determine whether to recommend a natural or a programmed FET cycle for a patient. Attempts to use a patient’s predicted length of menstrual cycle or luteal phase length have not been able to predict success in natural FET cycles (Reljič and Knez, 2018). Rather, these results should be used to guide decision-making during the natural FET cycle regarding whether to proceed with an embryo transfer or whether to cancel the cycle and begin monitoring again for potential FET in another natural menstrual cycle.

This study has several limitations. The retrospective design of the study is subject to possible selection bias in regard to which patients at our center were recommended to undergo a natural FET compared to a fresh embryo transfer or programmed FET. To decrease the heterogeneity of our study population, we only included patients who had blastocyst embryo transfers; therefore, it is unknown whether similar results would be observed in patients with cleavage-stage embryo transfers. Further, due to the retrospective nature of the study design, patients could not be randomized to a specific ovarian stimulation or ovulation trigger protocol. However, all patients were managed with the standardized protocols at a single center, which strengthens the external validity of our results when compared to a study that only evaluates one specific stimulation protocol.

Conclusion

In summary, the length of the follicular phase in natural FET cycles does not appear to impact pregnancy or live birth rates. However, when the dominant follicle develops with an accelerated trajectory, reflected by an oestradiol level >100 pg/ml to the LH surge in ≤4 days, pregnancy and live birth rates were decreased. This should be discussed with patients as a parameter that is associated with slightly lower success rates. Once a patient achieved a pregnancy, these variables did not affect the outcome of that pregnancy. These associations were confirmed in the subgroup analysis of PGT-A embryo transfers that was performed to account for potential unrecognized differences in aneuploidy between groups. Our observations provide cycle-level characteristics that can be applied during a natural FET cycle to help optimize success rates following embryo transfer.

Data availability

The data underlying this article cannot be shared publicly to protect the privacy of individuals included in the study.

Acknowledgements

The authors would like to thank Alexandra MacWade for her help in proofreading the manuscript.

Authors’ roles

Phillip A. Romanski: participated in study design, execution, analysis planning, manuscript drafting and editing. Pietro Bortoletto: participated in study design, analysis planning, manuscript drafting and editing. Yung-Liang Liu: participated in study design and analysis planning. Pak Chung: participated in study design, analysis planning and manuscript editing. Zev Rosenwaks: participated in study design, analysis planning and manuscript editing.

Funding

No financial support, funding or services were obtained for this study.

Conflict of interest

The authors do not have any conflict of interest disclosures.

References