Adding short-duration gonadotropin-releasing hormone antagonist and gonadotropin to natural cycle frozen embryo transfer allowed scheduling of transfer day without compromising live birth.

Data from 1,087 natural cycles (scheduled natural cycle n=114; natural cycle n=973) among 795 patients were analyzed. The median age at the time of ovarian stimulation was 35 (IQR 33–38) years and most patients had at least 1 prior failed embryo transfer ( Table 1 ). Those in the scheduled natural cycle group had a small but statistically significantly lower BMI (scheduled natural cycle 23.7 (IQR 21.5–26.8) vs natural cycle 24.7 (IQR 21.8–28.1) kg/m 2 , p=0.001) and were more likely to have unexplained infertility (scheduled natural cycle 35.1% vs natural cycle 25.4%, p=0.029). Otherwise, there were no significant differences in patient characteristics assessed between groups. Regarding cycle characteristics, the groups were comparable in terms of utilization of a single embryo transfer (scheduled natural cycle 96.5% vs natural cycle 92%, p=0.090) and there was no significant difference in the proportion of cycles with PGT-A (scheduled natural cycle 76.3% vs natural cycle 68.9%, p=0.11). The dominant follicle exhibited a small but statistically significant larger mean diameter at the time of trigger in the scheduled natural cycle group (scheduled natural cycle 19.8 (IQR 17.8–21.1) vs natural cycle 18.8 (IQR 17.5–20.4) mm, p=0.030), and peak serum estradiol concentration was also significantly higher in the scheduled natural cycle group (scheduled natural cycle 318 (IQR 208–433) vs natural cycle 249 (IQR 182–326) pg/ml, p=0.0002). There was no significant difference in peak endometrial thickness or serum progesterone concentration on the day of ovulatory trigger between the groups. However, peak serum LH measured on the ovulatory trigger day was significantly lower in the scheduled natural cycle group (scheduled natural cycle 5.7 (IQR 3.1–10.1) vs natural 13.4 (IQR 7.3–28.1) mIU/mL, p<0.0001). The pregnancy outcomes from the primary analysis are shown in Table 2 . The positive hCG (scheduled natural cycle 81.6% vs natural cycle 64.3%, RR 1.26 (95% CI 1.15–1.38)) and clinical pregnancy rates were significantly increased in the scheduled natural cycle group (scheduled natural cycle 68.4% vs natural cycle 57.1%, RR 1.21 (1.06–1.38)). Although there was a higher proportion live births in the scheduled natural cycle group, this difference did not reach statistical significance (scheduled natural cycle 57.0% vs natural cycle 49.4%, RR 1.15 (95% CI 0.97–1.36)). Further, there was no significant difference in spontaneous abortion between groups (scheduled natural cycle 11.8% vs 9.7%, RR 1.27 (95% CI 0.65–2.46)). Similar results were observed in a sub-analysis of the FET cycles utilizing PGT-A. ( Table 2 ). Again, positive hCG was significantly increased in the scheduled natural cycle group (scheduled natural cycle 82.8% vs natural cycle 66.3%, RR 1.24 (95% CI 1.11–1.37)). There was a higher proportion of clinical pregnancies in the scheduled natural cycle group, but this difference did not reach statistical significance (scheduled natural cycle 67.8% vs natural cycle 58.8%, RR 1.16 (0.999–1.36)). Notably, the primary outcome of live birth was similar between the two groups (scheduled natural cycle 55.2% vs natural cycle 52.4%, RR 1.03 (95% CI 0.84–1.26)). There was no significant difference in spontaneous abortion (scheduled natural cycle 12.5% vs natural cycle 7.9%, RR 1.93 (95% CI 0.90–4.13)). All results were similar after accounting for differential cancellation rates using IPSW ( Supplemental Table 1 ).

This was a retrospective cohort study across a single network of fertility clinics with natural cycle FET performed at multiple embryology labs in the United States. This research was conducted with approval from Advarra IRB (R2009.1). All autologous natural cycle vitrified-warmed blastocyst FET cycles performed between January 2019 and April 2022 were included. Natural cycle was defined as FET preparation where at least one dominant follicle developed spontaneously without any exogenous stimulation and ovulation occurred due to HCG administration and/or documented spontaneous LH surge. Patients with oligo-ovulation (or a diagnosis of polycystic ovary syndrome) were excluded given that these patients typically require ovulation induction with either clomiphene, letrozole, or exogenous gonadotropins prior to the natural development of a dominant follicle. As in many centers, given weekend staffing constraints, FETs are preferentially performed on weekdays in our network, to optimize standardization, quality control, and outcomes ( 21 ). Therefore, patients undergoing natural cycle FET are counseled that if their natural follicular progression threatens to result in a weekend FET, adjunctive medications may be used to delay ovulation and ensure weekday FET. Unless they have a contraindication to exogenous estradiol administration, patients are also offered programmed FET preparation. The referent group (natural cycle FET) consisted of cycles where patients underwent monitoring according to the standard protocol at our clinic ( Figure 1A ). Generally, patients presented for natural cycle monitoring approximately four days before their expected ovulatory trigger administration based on the length of their natural cycle. Serum hormone levels (estradiol, progesterone, and luteinizing hormone) and transvaginal ultrasound of the uterus and ovaries were performed serially until the endometrial thickness was considered adequate and a dominant follicle >17mm was measured at which point an hCG ovulatory trigger was administered that evening. Occasionally, hCG trigger was omitted in the setting of a clear LH surge, typically >20 IU/mL, and also taking into account baseline LH and rise over time. If any uncertainty regarding the endogenous LH surge, hCG trigger was also administered. The FET was performed 7 days after hCG trigger or 6 days after LH surge ( 5 ). The study group (scheduled natural cycle) was initially managed with the same protocol until a follicle was measured between 14–17mm on a Friday (not ready for hCG ovulatory trigger but at risk for an LH surge prior to Monday evening). At that point, these patients received GnRH antagonist (0.25mg, 1 ampule) and gonadotropins (75 IU) add-back (either recombinant FSH or human menopausal gonadotropin at provider discretion) were administered daily for typically 3 nights (median 3, range 1–4). These patients then returned for serum hormone levels and transvaginal ultrasound on Monday and received an hCG ovulatory trigger that evening if they met criteria ( Figure 1B ). Following the ovulatory trigger or LH surge in both groups, supplemental luteal progesterone with micronized vaginal progesterone (200 mg BID) was initiated 3–4 days before the FET. No additional bloodwork or ultrasound were performed between ovulatory trigger and pregnancy test other than ultrasound guidance at the time of the embryo transfer. The primary outcome was live birth, defined as a live born infant at 22 weeks gestation or beyond. Secondary outcomes included positive beta hCG, clinical pregnancy (presence of an intrauterine gestational sac), and spontaneous abortion. Continuous variables were expressed as medians with interquartile ranges and categorical variables were expressed as counts with percentages. P-values comparing distributions of covariates by cycle type were obtained using regression models fitted with generalized estimating equations (GEE) to account for cycle repeats per patient. The outcome of these models was a given covariate, and cycle type was entered as a single predictor. The models were linear, log-linear, logistic, or ordinal multinomial logistic depending on the distribution of the outcome. The P-value was obtained from a Wald test of the model parameter for the cycle type predictor. GEE modeling of the cycle and pregnancy outcomes was performed to control for potential confounding variables, selected a priori based on biologically plausible association with the exposure and primary outcome. Variables in the model included: female age at time of transfer (years), female age at time of embryo creation (years), BMI at transfer (kg/m 2 ), number of prior failed transfers, number of embryos transferred, endometrial thickness (mm), notated diagnosis of ‘uterine factor’ infertility, and PGT-A utilization. Since untriggered cycles were eligible to be included in both the referent and study groups and the choice to trigger or not occurred after the choice to convert a patient’s cycle to a scheduled natural cycle, we chose not to adjust for trigger versus surge in the models. These models were fitted with an exchangeable correlation structure and were weighted by the inverse of the number of cycles contributed by a patient, in order to account for informative cluster sizing in which patients who successfully become pregnant are likely to contribute fewer cycles to the analysis ( 22 ). A subgroup analysis was performed, limited to PGT-A cycles. A P-value of >0.05 was considered statistically significant. It is possible that the rate of cycle cancellation may differ between the study and referent groups, creating possible selection bias. In order to assess this, we compared the probability of cycle cancellation among scheduled natural cycles vs. among natural cycles in which at least one follicle reached a size of 14mm (i.e. the group of natural cycles that would have been eligible to become scheduled natural cycles, if the large follicle was observed on a Friday). These probabilities were 3.6% (N=10) and 8.1% (N=36) among scheduled natural cycles and natural cycles, respectively (P=0.02). We therefore conducted a sensitivity analysis accounting for potential selection bias using inverse probability of selection weighting (IPSW). Stabilized weights were constructed using a logistic regression model with non-cancellation as the outcome and predictors including cycle protocol (scheduled natural vs. natural), patient age, maximum observed follicle size, and primary diagnosis. These weights were applied to the same GEE models as in the main analysis.

We describe a novel natural cycle FET protocol that allows for more flexibility in scheduling the day of transfer. In this protocol, ovulation was delayed with short duration GnRH antagonist along with low dose gonadotropin add-back, resulting in a corresponding shift in day of transfer. This protocol maintained comparable pregnancy outcomes when compared to patients undergoing the traditional natural cycle FET. While positive hCG and clinical pregnancy rates were statistically higher in those that used the scheduled protocol, these data should be interpreted with caution given limitations of sample size in this cohort and ultimately did not result in difference in live birth, particularly among PGT-A cycles. The study findings may enable more clinics and their patients to undertake FET in the setting of a corpus luteum, which may be preferred by patients and has been hypothesized to result in improved maternal-fetal outcomes. These data may become particularly valuable if forthcoming RCT data confirm decreased hypertensive disorders of pregnancy following natural cycle FET compared to programmed FET cycles ( 18 ).

A growing body of evidence suggests endometrial preparation via the natural cycle approach, which allows for FET in the presence of a corpus luteum, results in similar or improved live birth outcomes while possibly reducing obstetrical risk when compared to programmed FET cycles ( 16 , 20 ). Although demand for natural cycle FET is expected to rise, a major barrier to its uptake is that scheduling of transfer is dictated by the ovulatory cycle and thus has potential to fall on a date precluded by patient or clinic availability ( 21 ). In this study, we described a novel natural cycle FET protocol that allows more flexibility in scheduling the day of embryo transfer without compromising clinical outcomes. A recent survey of fertility clinics in the United States, found that 17% of clinics do not offer natural cycle FET, and 81% of those reported not doing so due to “lack of timing predictability for transfer” ( 21 ). Notably, these clinics tended to be smaller practices located in less population dense areas. These data highlight the barrier that scheduling challenges pose to implementation of natural cycle FET. Scheduling challenges are multifactorial including expertise level, chain of custody minimums and economics. Therefore, strategies that allow for more control over the timing of FET may serve to reduce geographic disparities and improve accessibility to natural cycle FET. Our protocol used GnRH antagonist along with low dose gonadotropins for a short duration to allow for an adjustable delay in ovulation and corresponding shift in timing of transfer. There was no significant difference in live birth among cycles utilizing this protocol compared to the referent group of natural FET cycles. The positive hCG and clinical pregnancy rates were increased in the scheduled natural cycle group, while live birth rate was not statistically different between the two groups. These results were consistent across univariate analyses and multivariate analyses. Furthermore, similar results were observed in a subgroup analysis limited to PGT-A cycles. These data suggest that delaying ovulation in natural cycle FET did not negatively impact FET outcome. Notably, the scheduled natural cycle protocol was associated with significant changes to the late follicular phase as evidenced by larger follicle size, higher serum estradiol, and lower serum LH. One potential advantage of the scheduled natural cycle protocol is the suppression of LH until ovulatory trigger is administered. This may have allowed for greater precision for the time interval between ovulation and embryo transfer as compared to the natural cycle protocol in which some patients may experience an initial LH increase prior to the ovulatory trigger administration. However, ultimately the live birth was comparable between the two groups suggesting that patients managed with a scheduled natural cycle FET will have comparable outcomes to patients managed with a traditional natural cycle FET. It is interesting that the scheduled natural cycle FET group had a significant increase in positive hCG and clinical pregnancy and it is possible that the greater precision from time to ovulatory trigger to FET in the scheduled natural cycle protocol resulted in improved accuracy with regards to FET timing with the window of implantation. Further, the larger follicular diameters and higher serum estradiol levels may correspond to a generally more developed follicle, which may have resulted in more robust endogenous luteinization in the scheduled natural cycle FET group. However, it should also be noted that this secondary outcome may have occurred due to either random chance or confounding by prognosis if patients with a better prognosis were more likely to have undergone a scheduled natural cycle FET. Ultimately the increased clinical pregnancy did not result in a statistically significant difference in live birth between the two groups, particularly in PGT-A cycles. This study was strengthened by use of live birth as the primary outcome, which was high in both cohorts. Additionally, the data was derived across a multi-center IVF network with multiple embryology labs enhancing generalizability of the results. Further, the high rate of PGT-A utilization and single embryo transfers, reflect contemporary U.S. practice. This study also has several limitations to acknowledge. First, this study was limited by minor heterogeneity in natural cycle FET protocols inherent in a retrospective study design including small variations in the timing of FET relative to LH surge/trigger. We did not perform separate analysis of true natural cycle since the omission of ovulation trigger was relatively rare, and therefore this may limit the generalizability of these results to natural cycle FET cycles that do not receive an ovulatory trigger. However, randomized controlled trials have concluded that pregnancy outcomes are comparable between natural cycle with surge versus natural cycle with ovulatory trigger protocols, which is reassuring ( 23 , 24 ). Further, the potential for selection bias in terms of which patients underwent a scheduled natural cycle compared to a traditional natural cycle exists given the lack of randomization and standardized criteria for selecting candidates for scheduled natural cycle FET. Overall, the findings of the current study justify further prospective studies comparing scheduled natural cycle FET to traditional natural cycle and/or programmed FET protocols to confirm the findings of this study.

Over the past decade, there has been a significant increase in the number of frozen embryo transfers (FET) ( 1 ). In 2021, there were >250,000 FETs performed in the United States alone ( 2 ). This trend has been driven by advances in extended culture and cryopreservation techniques. Further, by allowing embryo transfer to take place outside the supra-physiological hormonal milieu of an egg retrieval cycle, FET reduces the risk for ovarian hyperstimulation syndrome and allows embryo selection with preimplantation genetic testing (PGT) ( 3 , 4 ). Generally, there are two approaches to preparing the endometrium for FET — programmed and natural cycles ( 5 , 6 ). Programmed cycles, a.k.a. hormone replacement cycles, utilize exogenous estradiol and progesterone to prepare the endometrium. The administration of estradiol in the follicular phase suppresses ovulation, and the corpus luteum remains notably absent ( 4 , 5 ). In contrast, natural cycle FET ta kes advantage of follicular development, ovulation and luteinization as a “natural” means of endometrial preparation. The FET is scheduled based on LH surge or hCG trigger following follicular and endometrial development ( 5 , 6 ). Thus, one notable advantage of a programmed cycle is that it facilitates scheduling of the FET compared to natural cycle where timing of the FET is dictated by ovulation. As compared to fresh embryo transfers, pregnancies achieved via FET have been associated with an increased risk for certain adverse obstetrical outcomes, including hypertensive disorders of pregnancy ( 7 , 8 ). However, more recent evidence suggests that this association may be limited to programmed FET, where the corpus luteum is notably absent ( 4 , 9 ). Observational studies have reported decreased risks for preeclampsia, postpartum hemorrhage, and macrosomia in natural cycle compared to programmed FET ( 10 – 15 ). With regards to preeclampsia, a recent meta-analysis found the risk in natural cycle to be half that of programmed FET ( 16 ). Overall, these observational studies suggest the potential importance of the corpus luteum in early placentation and vascular remodeling ( 4 , 17 ). However, given that the data thus far is observational, data from experimental studies are still needed before a cause and effect relationship can be established and a randomized trial (NatPro) is currently underway ( 18 ). Aside for secreting steroid hormones, the corpus luteum produces a variety of vasoactive substances, including relaxin and vascular endothelial growth factor (VEGF), that are thought to play a role, not only in placentation, but also in cardiovascular and renal adaptions of pregnancy ( 4 , 9 , 17 , 19 ). Despite observational studies suggesting similar live birth and decreased obstetrical risk, many clinics either do not offer natural cycle FET or significantly restrict its use ( 16 , 20 , 21 ) due to staffing and scheduling challenges relative to programmed FET cycles, for which transfers can be easily scheduled. A recent survey study of fertility clinics found that “lack of timing predictability for transfer” (81%) and “increased burden on staff / laboratory personnel on holidays and weekends” (54%) comprise the main reasons that the majority of FETs are still performed in programmed cycles ( 21 ). Given that timing of transfer is dictated by ovulation in natural cycle FET, novel strategies that delay the initiation of ovulation would allow for increased flexibility of transfer timing, which may result in increased utilization. In this study, we describe a novel protocol that uses a short duration of GnRH antagonist along with low dose gonadotropins to delay the development of a dominant follicle in order to provide more flexibility in timing of FET. Our primary objective was to determine whether the use of a scheduled natural cycle protocol compared to a traditional natural cycle protocol was associated with live birth outcome. We hypothesized that live birth would be similar between cycles that employed the scheduled natural cycle protocol when compared to the traditional natural cycle protocol that is primarily used at our center.