Impact of method of endometrial preparation for frozen blastocyst transfer on pregnancy outcome: a retrospective cohort study.

1,182 FETs were performed at Penn Fertility Care during the study period. 102 were excluded for not being autologous cycles, 47 were excluded for being non-blast transfers and 5 were excluded for being modified unstimulated cycle FETs. 1028 total FETs were included, 923 of which were in Group A and 105 of which were in Group B. (See Figure 1 ) FETs were most often performed for excess embryos (51% of Group A and 47% of Group B). Cancellations occurred in 5% of Group A and 14% of Group B cycles. Group B cycles were most often cancelled due to inaccurate determination of timing of ovulation or issues with timing of the transfer, whereas Group A cycles were most often cancelled due to inability to achieve an adequate endometrial thickness or because of an abnormal appearance of the endometrium. The two treatment groups did not differ significantly in terms of age at retrieval or transfer, infertility diagnosis, reason for FET or use of PGS/PGD, with nearly one fourth of both cohorts using PGS/PGD. (See Table 1 ) Women in Group A were more likely to be overweight and obese compared to Group B (Overweight OR 1.9; 95%CI [1.1, 3.4]; Obese OR 4.3; 95%CI [1.7, 10.7]). Women in Group B were more likely to have had at least one prior FET compared to Group A (OR 1.9; 95%CI [1.3, 2.9]). Women in Group B were also more likely to have a single embryo transfer compared to Group A (OR 2.0; 95%CI [1.2, 3.1]). Overall, and when stratified by age for most categories, live birth rates were not higher in Group A compared to Group B in adjusted models (aOR 1.0 95%CI [0.6, 1.5]). Similarly, rates of spontaneous abortion, therapeutic abortion, ectopic, biochemical or stillbirth were not higher in Group A compared to Group B. (See Table 2 ) To further address potential differences in diagnoses of ovulatory dysfunction, a restricted analysis was performed excluding anovulatory patients with PCOS and diminished ovarian reserve (DOR) and similar results were noted (aOR 0.9 95%CI [0.6, 1.4]; aOR 1.0 95%CI [0.6, 1.5] respectively). Live birth rates were not higher in Group A in adjusted analysis irrespective of whether PGS/PGD was (aOR 1.3 95%CI [0.5, 3.3]) or was not performed (aOR 0.9 95%CI [0.5, 1.5]). Similarly, they were not higher in Group A in adjusted analysis when restricting to freeze-only cycles (aOR 1.3 95%CI [0.7, 2.4]). In Group B, there were no patients greater than 40 years of age who conceived with autologous embryos (n=6). Of these six patients, one had preimplantation genetic testing performed. (See Table 3 )

This is a retrospective cohort study conducted at Penn Fertility Care in the University of Pennsylvania from January 2013 to February 2017. Women undergoing autologous blastocyst transfers were included. Those using donor oocytes, gestational carriers or modified unstimulated cycles were excluded. Modified unstimulated cycles were defined as cycles using oral ovulation induction medications, exogenous gonadotropins or trigger shot as part of the endometrial preparation. Patients of all ages and fertility diagnoses were included. The primary exposure was type of endometrial preparation: programmed versus unstimulated FET. Endometrial preparation type was determined at the discretion of the physician. Primary outcome assessed was live birth rates. Secondary outcomes include biochemical pregnancy, spontaneous abortion, therapeutic abortion, stillborn and ectopic pregnancy rates. A programmed FET, defined as Group A, was performed by first administering GnRH agonist suppression in the luteal phase. Ovarian suppression was confirmed after onset of menses with baseline hormonal and transvaginal ultrasound assessment. Oral estradiol was then initiated at a dose of 2mg daily and titrated to 6mg daily over 12 days. Transvaginal ultrasound and bloodwork was performed after 12 days of estradiol and embryo transfer was scheduled if the endometrial thickness was at least 7mm and estradiol levels were at least 200 pg/mL. In cases of inadequate endometrial thickness or morphology or inadequate estradiol level, vaginal estradiol or higher doses of oral estradiol were administered. Intramuscular progesterone was initiated at 50mg when appropriate parameters were met, and blastocyst transfer was scheduled to occur on the 6th day of progesterone supplementation. Beginning in May 2015, surveillance bloodwork was also performed on the day prior to transfer and estradiol and progesterone doses were increased if hormone levels were below the specified threshold. Transfers were cancelled for inadequate lining or inability to achieve appropriate estradiol levels prior to scheduling transfer. An unstimulated FET, defined as group B, involved patients obtaining bloodwork and then monitoring home ovulation predictor kits for LH surge. Patients with positive kits were brought in the following day for bloodwork and ultrasound. Those with predictor kits that were not reliable were brought in starting cycle day 12–14 for bloodwork and ultrasound and monitored for LH surge or collapse of a dominant follicle. Day 0 was defined as the day of follicle collapse. In situations where a discrepancy was noted between home predictor kits and ultrasound monitoring, ultrasound monitoring of follicle collapse was used to determine Day 0. Vaginal progesterone was initiated the evening of day 3 and blastocyst embryo transfer was performed on day 5. Transfers were performed under abdominal ultrasound guidance and embryos were warmed 1–2 hours prior to the scheduled transfer. Serum hCG was performed 10–12 days following embryo transfer. hCGs greater than 1 mIU/mL were considered positive and repeated according to the clinic protocol. Biochemical pregnancy was defined by a positive hCG that spontaneously dropped to <1 mIU/mL and in the absence of an intrauterine gestational sac. A clinical intrauterine pregnancy was defined as the presence of an intrauterine gestational and yolk sacs on transvaginal ultrasound. A spontaneous abortion was defined as loss of a clinical intrauterine pregnancy whereas a therapeutic abortion was an induced loss of a clinical intrauterine pregnancy. Pregnancy losses greater than 20 weeks gestation were defined as stillbirths. In order to have an 80% power to detect a 15% difference in live birth rates, favoring Group A, it was calculated that 412 cycles in Group A and 103 cycles in Group B would be necessary. Multivariable logistic regression was performed with determination of confounders with backward elimination as well as apriori variables that were determined to be of clinical significance. Variables fit in the model include age at retrieval, body mass index (BMI), infertility diagnosis, pre-implantation genetic testing (PGS/PGD), year of transfer and number of embryos transferred per cycle. Overall pregnancy, biochemical, spontaneous abortion, therapeutic abortion, stillborn, ectopic and live birth rates in Group A were compared to Group B. Analysis was performed using STATA version 14. Approval for the study was obtained from the University of Pennsylvania’s Institutional Review Board (Protocol 827237).

Our results indicate that overall, patients with programmed FETs did not have higher pregnancy or live birth rates compared to patients with unstimulated FETs. Although the rates appear to be similar and strongly suggest no difference, larger sample sizes would be needed in order to confirm non-inferiority or equivalence of the two types of endometrial preparation. Our findings contrast with several other retrospective studies showing superiority of either the artificial or unstimulated endometrial preparation methods. The studies by Xiao et. al and Morozov et. al found higher pregnancy rates in patients using an unstimulated cycle compared to those with programmed cycles.( 16 , 17 ) However, both used non-blastocyst embryos and neither evaluated live birth rates, which are more clinically meaningful as compared to pregnancy rates alone. By contrast, Zheng et. al and Hill et. al noted higher pregnancy rates in women using programmed cycles.( 14 , 15 ) While Zheng et. al evaluated live birth rates as an outcome, the study was limited in its generalizability to current practices by use of nonblastocyst embryos and relying on high thresholds of progesterone instead of follicle collapse or LH surge in the unstimulated cycle regimen to time ovulation. Hill et. al also evaluated live birth rates as an outcome with a similar retrospective cohort design to ours. However, they did not account for the use of preimplantation genetic testing, which has become more widespread in current practice. Two prospective studies have found similar implantation, pregnancy and live birth rates between unstimulated cycles and artificial embryo transfers however restricted inclusion criteria to women under age 40.( 11 , 12 ) A prospective non-inferiority randomized trial of modified unstimulated cycle FETs versus artificial cycles found non-significant higher live birth rates in the modified unstimulated cycle group but included both cleavage stage and blastocyst embryos.( 13 ) A systematic review and meta-analysis of 33 studies found that while unstimulated cycles had increased clinical pregnancy rates compared to non-suppressed artificial cycles, there was no significant difference in live birth rates.( 9 ) In our study, there was appropriate representation of patients from the full spectrum of infertility diagnoses in both cohorts, increasing its generalizability. In addition, nearly one fourth of patients from both cohorts used preimplantation genetic testing. Although women using unstimulated FETs had a lower mean BMI, this is likely correlated to overweight women having anovulatory cycles( 23 ) thus predisposing them to be offered programmed FET cycles that do not rely on regular menses for timing of transfer. Interestingly, women undergoing unstimulated FETs were more likely to have a single embryo transfer compared to women undergoing programmed FETs. This finding may be reflective of a shift in practice patterns towards more single embryo transfers in parallel with the updated American Society of Reproductive Medicine guidelines for number of embryos to transfer.( 7 ) Regardless, number of embryos transferred was incorporated into the multivariate regression model to account for this difference. Strengths of our study include its relatively large sample size, limited exclusion criteria and incorporation of women of all age ranges and infertility diagnoses. Preimplantation genetic testing utilization was also evaluated and there was consideration of clinically meaningful outcomes including live birth rates. By following patients to live birth, it is also possible to evaluate whether the effects of differences in endometrial preparation type persist beyond the first trimester, as increasing evidence suggests that the peri-implantation environment may impact perinatal outcomes.( 24 ) Results remained robust in restricted analyses of nonanovulatory patients and after accounting for preimplantation testing and freeze-only cycles. To our knowledge, this is the first such study to incorporate many of the modern practice patterns of reproductive medicine including vitrification, blastocyst transfer and use of preimplantation genetic testing. Findings can be limited however by its retrospective nature and by the higher proportion of programmed cycles performed at our institution. The higher proportion of programmed cycles is likely due to physician preference for logistical reasons. Anticipated difference in outcomes is not thought to play a role in the decision by physicians. Thus, any bias would be non-differential in relation to outcome. This study provides evidence that, in most women, exogenous hormone preparation does not result in improved outcomes compared to unstimulated endometrial preparation with luteal support in a contemporary population presenting to an academic fertility center. Given that the difference in live birth detected was smaller than what our study was powered for, larger studies would need to be conducted in order to confirm if the two methods are equivalent. In addition, there are many other components to consider when deciding upon the optimal endometrial preparation route for an individual. Logistical aspects such as number of appointments, costs of medications and clinic visits and a patient’s desired timeline are all important factors to incorporate into the shared decision-making process with the patient. Clinicians should also discuss with patients the delay that could be imparted by cancelled cycles. Consideration for what would be the best option should therefore incorporate all these elements in order to help increase accessibility to treatments for patients. However, women over the age of 40 may benefit from programmed FETs due to challenges associated with increased cycle variability that is expected in this age group. Further prospective studies with larger sample sizes are warranted in this population.