Increased Risk of High Birthweight in Singleton Newborns after Frozen-Thawed Embryo Transfer According to Endometrial Preparation, Ovulatory or Artificial Cycle

Background: It is unknown whether prolonged artificial hormonal environment during early fetal development affects the birthweight of singletons born after frozen-thawed embryo transfer (FET). Methods: A retrospective observational study included singleton births>22 weeks of gestation obtained after FET between 2013-2019 after endometrial preparation with ovulatory cycle (OC) or artificial cycle (AC). Our primary objective was to compare birthweight of singletons after FET between endometrial preparation by OC or AC. Secondary objectives included prolonged pregnancies, high birthweight, low birthweight, SGA and LGA rates. Multivariate analyses were performed considering potential confounding factors. Results: Among 198 singleton live births after FET, 112 were obtained with OC and 86 with AC. Prolonged pregnancies rate was higher in AC (25.6% vs. 7.1%, respectively, p=0.001). Mean birthweight was higher (+219g) in AC (3386g vs. 3167g, p=0.01; adjusted-p=0.052), as well as the rate of babies exceeding 4000g (16.3% vs. 2.7%, p=0.03, adjusted-p=0.015). The rate of babies <2500g was lower in AC (3.5% vs. 11.6%, respectively, p=0.050, adjusted-p=0.049). Conclusions: Since OC does not strain the chances of pregnancy and in the incomplete knowledge of the consequences of neonatal overweight on the future health of children, OC preparation could be advocated as first-line endometrial preparation in FET.

Besides embryo survival and quality, successful FET relies on endometrial receptivity at the time of transfer. The most common protocols for endometrial preparation are: (1) the artificial cycle (AC), and (2) the ovulatory cycle (OC), either natural cycle (NC) or mild ovarian stimulated (OS) cycle. In OC, a luteal phase support using progesterone can be prescribed but a corpus luteum (CL) naturally secretes progesterone, enabling endometrial receptivity for implantation. In AC, estrogen and progesterone supplementation are mandatory since no CL exists, and is continued up to 12 weeks in case of pregnancy. The choice of the best protocol is still debated. While studies reported equally successful pregnancy rates (PR), ongoing pregnancy rates (OPR) and live birth rates (LBR) between protocols in women with regular cycles [7][8][9], a recent systematic review and meta-analysis [10] suggested mild OS as a promising option since PR and LBR were higher compared to AC. We recently published data showing that OC was associated to higher LBR compared to AC in multivariate analysis [11].
Hence, it seems that the choice of the best protocol should consider obstetric and neonatal outcomes [12]. Numerous studies have indicated that newborns conceived after IVF and immediate transfer (so-called "fresh") had lower average weights compared to natural conception, even for singletons. Subsequently, birthweight, macrosomia (birthweight >4000g), and Large for Gestational Age syndrome (LGA) (birthweight >90th percentile for gestational age) were reported to be higher in children born after FET compared to fresh transfer, regardless of techniques used [13][14][15][16][17][18][19][20][21][22]. However, the underlying mechanisms are still poorly understood [15,[23][24][25][26][27]. Characteristics of the cellular process of fetal growth and epigenetic regulation during pre-implantation are still in question [28]. Intrauterine growth potential may be affected by epigenetic changes in the early embryonic stages during freezing and thawing [29]. Biological conditions (IVF technique, culture medium, embryonic stage) may play a role. A study led by Pinborg et al. [21] reported higher LGA and macrosomia rates in FET vs. fresh transfer, even after adjustment for birth order, suggesting that results could not only be explained by being the second born or by intrinsic maternal factors, but may also be related to freezing/thawing procedures per se. Altogether, whether the periconceptional hormonal environment induced by endometrial preparation protocols for FET has an impact on birthweight remains to be elucidated. A recent study suggested that the absence of CL in AC may play a role in the increased risk of hypertensive disorders during pregnancy [30].
Another study concluded that placental volume and other 1st trimester parameters were modified by IVF with fresh embryo transfer or FET compared to spontaneous conceptions, but with opposite trends, and that hormonal treatment per se may have a major effect on pregnancy outcomes through the modification of placental invasiveness [31].
The main objective of this monocentric retrospective study was to compare the birthweight of singletons conceived after FET between endometrial preparation by OC or AC in daily clinical practice. The secondary objectives were to compare prolonged pregnancies high birthweight, low birthweight, SGA and LGA rates between these two groups and to evaluate whether endometrial preparation was predictive of these neonatal outcomes.

Endometrial preparation protocols
The choice of endometrial preparation protocol depended on the physician's decision. The 2 endometrial preparation protocols were: weeks of gestation (WG) if pregnancy, since the major source of progesterone derived from the CL. FET was performed 5 days after r-hCG injection for cleaved embryos and 7 days after for blastocysts.
Ultrasound and hormonal testing were performed Days 12-14 and repeated if needed. When endometrial thickness reached 7 mm and serum progesterone level <1ng/ml, VMPg (600 mg daily) was administered (3 and 5 days before FET for cleaved embryos or blastocysts respectively) until the 12th WG if pregnancy.

Embryos
Embryos were obtained from conventional IVF or ICSI cycles. Egg donation cycles were not included. Embryos were mainly frozen at cleavage stage until April 2016, then at blastocyst stage after evolution of laboratory policy. Cleaved embryos were considered alive if at least 50% blastomeres were intact after thawing.

Births
All live births > 22 WG were included. Medical terminations of pregnancies and antenatal deaths were excluded. No perinatal deaths were reported. Premature delivery was defined as < 37 WG. Prolonged pregnancies were considered for terms > 41WG [35].
Macrosomia was defined for any child born at term > 4000g. SGA and LGA were defined as birthweight < 10th or > 90th percentile for gestational age, respectively.

Statistics
Descriptive and regression analyses were performed using Stata for Windows (version 14; StataCorp). Paternal and maternal ages at freezing, maternal smoking status, body mass index, PCOS, parity, technique (IVF/ICSI), embryo stage and freezing technique were included in the multiple regression model as potential confounders.

Ethics
The Ethical committee gave its unrestricted approval for the study and all patients had previously given their consent to use their data (CEERB Paris Nord, IRB 00006467: ID 2018-013). All data were collected from Medifirst® software, meeting recognized medical and ethical specifications according to the French information protection commission (ID 2068638).

Results
Among the 198 singletons born after FET during the 5-year study period (2013-2018), 112 were conceived with OC and 86 with AC, respectively. No difference was observed concerning parental characteristics, except for PCOS (more frequent in AC, p = 0.002).
These discrepancies are highlighted in the rugplot and shift in birthweight density curves according to OC or AC preparation ( Figure 1). When representing birthweight distribution in relation to gestational age at birth, the increased incidence of > 4000g is objectified in extended terms > 41WG, standing out in the AC group (Figure 2). SGA and LGA rates were not different.

Discussion
This retrospective analysis of neonatal outcomes of 198 singletons born after FET showed that endometrial preparation with AC was associated to higher birthweights and rates of birthweight > 4000g compared to OC. These results remained significant after multivariate analysis. Among the maternal factors known to influence neonatal birth weight, BMI and parity were identical in both groups. PCOS, more frequent in the AC group, was included as an adjustment variable. Gestational age was not significantly different, nor were preterm or very pre-term birth rates. However, prolonged pregnancies > 41 WG were significantly higher in AC, which may impact the increased birthweight observed (+ 219g).
Our study focused on the difference in neonatal growth indicators according to treatments NC [36]. Nevertheless, the study comprised a very selected population, as it excluded women over 39, with severe endometriosis, or PCOS. Recently, Jing et al. [37] showed that AC was associated to higher birthweights compared to NC in multiple pregnancies (2550g vs. 2600g, respectively, p = 0.023). Consistently to our results, Ishii et al. [38] found significantly greater birthweights (+ 137g, p < 0.01) after AC (n = 403) compared to OC (n = 117). In blastocyst transfers, average birthweight was significantly higher in AC (p < 0.01).
For cleaved stage embryos, although not significant, average BW was higher in AC.
Average BW from the AC-blastocyst transfers was, as described in our study, higher compared to OC-cleaved stage embryos (p < 0.01). Putative factors affecting BW such as stimulation protocols, stage and quality of embryos could not explain the difference observed.
To date, multiple findings indicate that FET leads to heavier babies compared to fresh transfers [12-19 ,21,23,26,37]. However, the underlying mechanisms remain unclear. The first hypotheses evoked were maternal factors and biological conditions. Vidal et al.'s [40] analysis including 14262 singletons births > 24 WG suggested that maternal factors may play a role, as babies born from FET (n = 1158) had significantly higher birthweights (+ 190g), lower LBW and SGA rates (p < 0.001) compared to fresh transfer (n = 5560) in autologous cycles, while no difference was observed for egg donation transfers.
Conversely, the role of maternal factors was questioned by Pinborg et al. [21]  , respectively), and the increased risk was confirmed after adjustment on birth order. Therefore, results could not only be explained by being the second born, or by intrinsic maternal factors, but could also be related to freezing/thawing procedures per se.
Consistently, Anav et al. [41] suggested that cryopreservation in itself could be responsible of birthweight variations, independently of parity. Furthermore, animal studies such as the description of Big Calf Syndrome [42] and murine models focusing on the role of freezing techniques on genomic imprint showed that culture media constituents could affect birthweight, although no consensus exists concerning human data. Indeed, while Dumoulin et al. [43] and Nelissen et al. [28] showed a significant impact of embryo culture medium on early embryonic and fetal development and birthweight, no significant association was reported by Eaton et al. [44] and Vergouw et al. [45], even when adjusted for gestational age, gender and parity.
Concerning a possible influence of embryo stage on birthweight, Ishihara et al. [46] compared pregnancy outcomes between 4 groups: fresh Day 2 (n=10928); fresh Day 5 (n = 5981); vitrified-FET Day 2 (n = 3841); vitrified-FET Day 5 (n = 27408). FET was associated to lower LBW < 2500g and SGA, and to higher LGA rates (AOR 1.48). Lower rates of SGA and higher rates of LGA were observed for Day 5 vs. Day 2 embryos. Conversely, Belva et al. [47] found no influence of embryo stage on neonatal outcomes in live born singletons after fresh or FET cycles, although lower rates of SGA (p = 0.005), higher rates of birthweight standard deviation score (SDS) (p = 0.008), length at birth SDS (p = 0.001), and head circumference SDS (p = 0.005) in FET groups were reported. The impact of endometrial preparation was not considered. Our work comprised more D5 and vitrified FET in OC than in AC group, but no difference in birthweight and LGA according to embryo stage was observed.
Although no transnational report exists on the proportion of protocols used for endometrial preparation, it seems that FET is mostly conducted using AC worldwide. However, using AC implies the continuous administration of E2 and progesterone during the first trimester of pregnancy. Specific attention should be paid to potential foetal effects of extended hormonal treatments, especially because epigenetic modifications can induce specific pathologies revealed decades after exposure. For instance, one must consider the dramatic consequences of diethylstilbestrol (DES) exposure during early pregnancy and its persistent effect in subsequent generations [48][49][50][51]. Unlike oocyte donation in which AC is mandatory to prepare embryo implantation, the choice of protocol remains open for FET.
In case of endometrial preparation with OC, a corpus luteum is active and progressively secretes progesterone until stabilisation of serum levels, leading to constant progesterone levels [52]. Moreover, ovulation triggering with hCG sustains the luteotropic effect in early luteal phase [53]. Finally, other factors are also secreted by the CL, leading to possibly more natural endometrial protein secretion profiles compared to AC [52], and recent findings suggest increased obstetrics risks in the absence of CL [54]. Besides maternal complications, adverse fetal outcomes such as stillbirth, shoulder dystocia, hypoglycemia, respiratory distress and perinatal mortality are increased in macrosomic babies, and further concerns during child-and adulthood include metabolic, cardiovascular and endocrine complications [55][56][57][58].

Conclusions
In conclusion, this study demonstrates that endometrial preparation by AC and its prolonged artificial hormonal environment during early foetal development in FET was significantly associated to higher mean birthweight, macrosomia and prolonged pregnancy rates compared to OC. The precautionary principle implies that the prevention of short or long term potential consequences of these neonatal characteristics, since possible, seems coherent. Since results obtained by OC do not strain the chances of pregnancy, and in the incomplete knowledge of the consequences of neonatal overweight on the future health of children, OC preparation could be advocated as first-line endometrial preparation in FET.
Randomized controlled trials should be undertaken to assess these preliminary results. Informed Consent Statement: According to French regulations, written informed consent from the patients was not required for this study. Patient consent was waived due to French regulations. All data were rendered anonymous before access and analysis.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.