Increased risk of maternal and neonatal complications in hormone replacement therapy cycles in frozen embryo transfer

In 1984, the success of first live birth after thawing the frozen human embryos was reported by the team of Zeilmaker [1]. Since then, the proportion of frozen embryo transfer (FET) has been increasing in in-vitro fertilization (IVF) along with the development of laboratory techniques for cryopreservation and the reduced number of embryo transfer in fresh cycle [2]. Recently, the so called “freeze-all strategy”, i.e. selectively freezing all embryos and performing FET later has become an optimal choice in cycles with high risk of ovarian hyperstimulation syndrome (OHSS), preimplantation genetic testing (PGT) and double ovarian stimulation (DuoStim) protocols [3]. Several studies reported that FET achieved higher pregnancy rates and lower complications rates compared with fresh embryo transfer [4‐6]. Therefore, the “freeze-all strategy” has been widely used to improve live birth rates and decrease potential complications.

Multiple protocols for endometrium preparation have been explored during FET. The common protocols include natural cycles (NC), artificial cycle with hormone replacement therapy (HRT), and cycle with ovulation induction (OI) [7, 8]. Groenewoud et al. conducted a randomized controlled trial (RCT) concluding that HRT cycle was not inferior to modified natural cycle for FET with regard to live birth rates (LBRs), clinical and ongoing pregnancy rates [9]. The latest Cochrane review suggested there were no significant differences in pregnancy rates, miscarriage rates, or live birth rates among different endometrial preparation protocols for FET [7]. However, the effect of different protocols on maternal and neonatal outcomes is still uncertain. Here, we performed a retrospective study to compare the maternal and neonatal outcomes in women underwent NC, HRT, or OI protocol for endometrial preparation.

The number of women receiving NC, HRT and OI protocol for FET were 4727, 1642 and 517, respectively. The baseline characteristics were listed in Table 1. BMI (23.2 ± 3.4 vs. 22.8 ± 3.3 vs. 22.5 ± 3.2 kg/m², p < 0.001), AFC (15.6 ± 6.6 vs. 15.3 ± 6.0 vs. 14.9 ± 5.9, p < 0.001), and levels of testosterone (26.0 ± 11.9 vs. 25.6 ± 12.5 vs. 24.4 ± 11.7 ng/dL, p < 0.001) and AMH (5.3 ± 3.6 vs. 5.0 ± 3.5 vs. 4.7 ± 3.7 ng/mL, p < 0.001) were higher in HRT group compared to OI and NC group. As expected, HRT group and OI group had higher rates of irregular menstruation compared with NC group (21.9% vs. 21.7% vs. 5.6%, p < 0.001). Systolic pressure (120.8 ± 11.6 vs. 120.7 ± 11.9 vs. 119.5 ± 11.8, p < 0.001) and diastolic pressure (72.4 ± 8.6 vs. 73.3 ± 8.8 vs. 71.7 ± 8.7, p < 0.001) were higher in HRT and OI groups than NC group. And difference was also found in the FET cycle number among NC, HRT and OI groups (1.2 ± 0.5, 1.3 ± 0.6, 1.6 ± 0.8, p < 0.001). However, no difference was observed in the rate of donor sperm using (9.3% vs. 8.0% vs. 7.4%, p = 0.115) and vanishing twin gestation (3.0% vs. 3.2% vs. 2.5%, p = 0.710) among the three groups.

Table 2 listed the maternal and neonatal outcomes. For the maternal outcomes, women using HRT protocol had higher rates of HDP (7.9% in HRT vs. 4.6% in OI vs. 3.5% in NC, p < 0.001) than women in OI and NC group. Significant difference was observed in the rates of GDM among the three groups (6.5% in HRT vs. 7.5% in OI vs. 5.2% in NC, p = 0.030). The subgroup analyses showed that the GDM rate in OI group was higher than that in NC group, however, no difference existed between HRT group and NC group. The rates of placenta previa (1.5% in HRT vs. 1.2% in OI vs. 1.0% in NC, respectively, p = 0.299) and oligohydramnios (1.0% in HRT vs. 1.7% in OI vs. 1.3% in NC, respectively, p = 0.437) were similar among the three groups. For the neonatal outcomes (Table 2), HRT and OI groups had higher risk of PTB compared to NC group (7.9% in HRT vs. 4.6% in NC, p < 0.001 and 7.7% in OI vs. 4.6% in NC, p = 0.001), but no significant difference was found between HRT and OI groups (7.9% vs. 7.7%, p = 0.773). The HRT group had increased risk of LBW compared with the NC group (4.5% vs. 2.8%, p = 0.001); however, no significant difference was observed between the OI group and the NC group (3.7% vs. 2.8%, p = 0.289). In addition, different LGA risk was found among three groups (26.2% in HRT vs. 22.1% in OI vs. 23.4% in NC, p = 0.040), although there was no significant difference between HRT and NC groups (p = 0.112). Additionally, the three groups were comparable in terms of SGA risk (2.9% in HRT vs. 5.0% in OI vs. 3.5% in NC, p = 0.073) and gender of neonates (p = 0.890).

After adjusting for the effect of age, BMI, irregular menstruation, AFC, endometrial thickness, the levels of testosterone, AMH, PFG, systolic and diastolic pressure et al. (Table 3, Fig. 2), HRT group still showed higher risk of HDP (aOR 2.00, 95% CI 1.54–2.60) and LBW (aOR 1.49, 95%CI 1.09–2.06) compared with the NC group. Risk of PTB in the HRT and OI groups were elevated compared with NC group (aOR 1.78, 95%CI 1.39–2.28 and aOR 1.51, 95%CI 1.02–2.23, respectively).

The comparisons between IVF and ICSI cycles were listed in Additional file 2: Table S2-S13.

In this study, we compared the maternal and neonatal outcomes of singletons born after FET with different endometrial preparation protocols in a large cohort. A higher risk of HDP and LBW was observed in the HRT group, as well as increased risks of PTB in both HRT and OI group compared with NC group. The differences in the maternal and neonatal outcomes could be explained by the excessive estrogen exposal, absence of corpus luteum (CL) and distinct clinical characteristics of population in the HRT group.

During early pregnancy, the uterine spiral arteries transform from high-resistance, low-capacity to low-resistance, high-capacity vessels by trophoblasts immigrating, invading and replacing the endothelial and smooth muscle wall of the spiral arteries [14, 15]. The remodeling of the uterine spiral arteries is crucial for sufficient nutrient and oxygen supply from the placenta to the fetus through an optimal uteroplacental blood flow. Excessive estrogen levels have been reported to impair the invasion of trophoblastic vessels during pregnancy. In pregnant baboon, it has been shown that estrogen plays a major role in regulating morphological and functional differentiation of the villous trophoblasts and signals between the placenta and fetus. Moreover, an elevated estrogen level markedly suppressed the vascular invasion [16‐18]. Attenuation of trophoblast vascular invasion and spiral artery restructure will result in placental defect, which subsequently interferes the pregnancy process and end up with complications such as hypertensive and growth disorders [19‐23].

It has been evidenced that high levels of estrogen could increase the risks of maternal and neonatal outcome during IVF-ET treatment. Imudia et al. discovered that an elevated serum E₂ concentration during controlled ovarian hyperstimulation (COH) was associated with higher risk of maternal preeclampsia and SGA newborns for fresh embryo transfer [24]. Pereira N et al. conducted a retrospective cohort study of 4071 patients undergoing fresh IVF-ET cycles, indicating that serum E₂ levels exceeding 2500 pg/ml during COH seemed to be an independent predictor for LBW in full-term singletons [25]. Additionally, compared to FET, fresh embryo transferred showed worse obstetrical outcome, including LBW and PTB, which was also attributed to a hyperestrogenic milieu generated during ovarian stimulation [26‐29].

During the preparation with HRT protocol, the patients were prescribed estrogen normally from 2 weeks before embryo transfer till 8 ^th -10th week of pregnancy. It is assumed that the intake of exogenous estrogens during the period of trophoblastic vessels invasion may result in increased risk of maternal and neonatal complications, such as HDP and LBW. Study from Tatsumi et al. compared the pregnancy and neonatal outcomes among OI with letrozole cycle, natural cycle and HRT cycle for FET and discovered differences in terms of gestational weeks at delivery, birth weight and SGA/LGA among three groups [30].

Another explanation for high risk of HDP in HRT protocol could be ascribed to the absence of corpus luteum (CL) in the first trimester when CL contributed most to hormone secretion. According to a recent study by von Versen-Höynck F, CL defect was associated with elevated rate of preeclampsia, which was probably caused by lack of circulating relaxin, a potent vasodilator secreted by CL yet not supplemented in the luteal phase support [31]. Another study also stressed that the vascular health was impaired when no CL was present during early pregnancy, which suggested an insufficient vascular adaption involved in the development of preeclampsia [32]. Ginström Ernstad E et al. conducted a population-based retrospective study in Sweden, in which FET cycles were grouped according to presence or absence of a CL [33]. His results demonstrated an increased risk of hypertensive disorders in HRT FET cycles, which was in accordance with our findings.

Thirdly, the increased risk of obstetric and neonatal complications in the HRT group was also caused by the distinct characteristics of the population. In our study, the preferred choice for endometrium preparation was NC protocol; while for patients with irregular menstruation or history of oligo-ovulation or anovulation, HRT or OI will be suggested. Although patients with PCOS had been excluded in the study, it was till reasonable to expect more women with endocrine disturbance in HRT and OI group than NC group. Not unexpectedly, compared to the NC group, women in HRT group had more AFC, higher BMI, and serum testosterone levels. Increased BMI and high levels of testosterone were involved in obstetrical complications through altered trophoblast invasion and placentation [34]. Obese patients were more likely to have dyslipidemia, which was associated with pregnancy complications and adverse pregnancy outcomes owing to vascular damage and endothelial dysfunction caused by oxidative stress from free radicals and lipid peroxides [34‐36]. Hyperandrogenism and insulin resistance were also reported to alter endovascular trophoblast invasion and placentation [37‐39]. Excess maternal androgens reduced placental weight and affected fetal growth in rats [40]. Therefore, the endocrine-metabolic disturbance in women using HRT protocol may also contributed to the adverse pregnancy outcomes.

Our sample size was large enough to detect the differences of maternal and neonatal complications in different groups as well as to adjust for the maternal characteristics. One of the limitations was the estradiol concentration during early pregnancy was unavailable thus the comparisons of estradiol level during pregnancy among different FET groups were lacking. Another problem was that since OI protocol was costlier and more time intensive for patient visiting compared to HRT protocol, patients in the OI group might tend to have a history of thin endometrium or cancelled cycles which could lead to different maternal characteristics in this group.

Hormone replacement therapy protocol for frozen embryo transfer of blastocysts may be associated with adverse maternal and neonatal outcomes, such as high risks of HDP, LBW and PTB. Our results provide reference for endometrial preparation during FET treatment. More attention should be paid to the potential harmful effects of excessive estrogen and corpus luteum defect on maternal and neonatal complications during pregnancy.