In high responding patients undergoing an initial IVF cycle, elevated estradiol on the day of hCG has no effect on live birth rate

In IVF cycles, high responding patients are characterized as having a large number of follicles that are associated with elevated levels of estradiol and frequently by elevated levels of progesterone on the day of hCG administration. In most situations, this results in an increased number of embryos from which candidates can be chosen for embryo transfer. Due to its potential clinical importance the impact of elevated estradiol on the day of hCG on IVF outcomes has been debated for over 25 years and even after this length of the time the issue of how, or if, the supra-physiological hormonal environment exhibited in high responding patients has an adverse effect on pregnancy outcome is still not settled. Regarding elevated estradiol and pregnancy outcome, some investigators [1‐4] have shown a positive effect and others [5‐8] a negative effect; while most studies [9‐18] have shown no effect. A meta-analysis conducted in 2004 [19] reviewed nine studies and concluded that no high quality evidence existed to support or refute the value of estradiol levels on the day of hCG administration as a determining factor in establishing pregnancy in IVF cycles. Possibly contributing to the disparity in outcomes from these studies are differences in the way trials were conducted, including the use of repeated measures, patient exclusion criteria, type of stimulation, day of transfer, number of embryos transferred and how estradiol levels were categorized (cutoff, percentile or area under the curve). There is also considerable variation among studies in how pregnancy rates were expressed with most using clinical pregnancy rate, with or without implantation rate. Few studies have based their conclusions on live birth outcome which arguably provides the most information on the embryos ability to tolerate the hormone environment to which it has been exposed.

The purpose of this study was to determine the relationship between the levels of estradiol on the day of hCG administration and pregnancy outcomes, with emphasis on live birth rate after controlling for embryo quality. We addressed this by initially categorizing patients according to the level of estradiol on the day of hCG administration, which has been the basis of the majority of previously published studies. We then performed additional analyses controlling for embryo quality and for confounding and interactions occurring between IVF response variables. In a secondary analysis we compared differences in estradiol levels between patients achieving a live birth and those that did not (NLB), since differences in estradiol levels have been reported in previously published reports when stratified by birth outcome [17, 20, 21]. We also tested if the day of embryo transfer made a difference in live birth (LB) rate in patients with an elevated level of estradiol on the day of hCG since some prior studies [17, 20, 22] had suggested that elevated estradiol only had an adverse effect on pregnancy outcome when embryo transfer was conducted on day 3.

Patient demographics for the entire population are shown in Table 1. Pregnancy rates and IVF cycle response parameters for the entire population (n = 489) and the sub-group (n = 428) that had only good or fair quality embryos transferred, are summarized in Tables 2 and 3 respectively. For the entire population (Table 2), the higher estradiol groups were associated with greater biochemical and clinical pregnancy rates. Whereas, in the sub-group (Table 3), with fair to good embryo quality no association with estradiol was observed on BCP, CPR or LBR in contrast to all other IVF response variables. Although live birth rates were greatest (NS, P = 0.052) in higher estradiol groups from the categorical analysis shown in Table 2, following the univariate analysis (Table 4) in which estradiol level was evaluated as a continuous variable, estradiol level had a significant (P = 0.013) effect on LB as did other covariates including age, FSH dosage, number of metaphase II (MII) oocytes, total 2PN embryos, AEQS and embryo transfer day. In the multivariate analysis (Table 5) however, estradiol level was not related to live birth. Only the main effects of AEQS, patient age, and the transfer of two embryos as opposed to 1 or 3, were found to significantly affect LB in the final model. The multivariate regression model also revealed significant two-way interactions between transfer on day 5 and patient age (P < 0.01) and transfer on day 5 and FSH dose administered (P < 0.04). IVF response variables in the total population and in the subgroup containing only good and fair quality embryos on the day of embryo transfer were evaluated by live birth status. In the total population for day 3 transfers (Table 6) there were differences between the NLB and LB groups in patient age, FSH dose, number of 2PN embryos and AEQS, whereas, in the subgroup-only analysis, patient age and AEQS were different. In the total population for day 5 transfers (Table 7), FSH dose, number of MII eggs and AEQS differed. In the subgroup however, only FSH dose differed between NLB and LB. In the entire population the LB rate was greater for day 5 than for day 3 transfers. However, when the subgroup containing only good and fair embryos was evaluated, although the LB rate was slightly higher (65.3% vs 59.1%), the difference was no longer statistically significant.

In the final analysis, a subgroup of patients having embryo transfer on day 3 or day 5 that had an estradiol level of > 3000 pg/ml on the day of hCG were compared (Table 8). When embryo transfer was conducted on day 3 or day 5 with high quality embryos, patients had similar estradiol levels, with the number of oocytes retrieved, mature oocytes and number of 2PN embryos all being higher in the day 5 transfer group. However, there was no significant difference in the CPR or LBR for patients receiving a transfer on day 3 or day 5.

As another part of this analysis we examined if a high level of estradiol (>4200 pg/ml) on the day of hCG had a deleterious effect on pregnancy rate when compared to a group having moderate estradiol levels of >3000 pg ml and <4200 pg/ml, that had embryo transfer on day 3. As shown in Table 9, the estradiol level and number of oocytes retrieved were higher in the >4200 pg/ml group, but there was no difference in either the CPR or LBR between groups.

This study examined the relationship between estradiol level and other IVF cycle response parameters in relation to pregnancy, with a focus on live births after controlling for embryo quality. In high responders there is usually an increased number of eggs retrieved resulting in more embryos and therefore, better embryo selection, leading to more patients undergoing transfer on day 5. In the case of poor responders the opposite is true as was demonstrated in the low estradiol group in Table 2. Here, there were fewer oocytes in spite of a higher dosage of FSH, which also resulted in fewer 2PN embryos and ultimately poorer embryo selection at the time of transfer.

In our experience, patients having embryo transfer on day 3 can attain high pregnancy rates, but only if their embryo quality is not compromised. Therefore, we evaluated a population of patients undergoing their initial IVF cycle to determine the relationship between serum estradiol level and IVF outcomes. We next evaluated a subgroup that included patients having only good or fair quality embryos transferred to study the influence of embryo quality on pregnancy rate and subsequent live birth outcome. Although our initial categorical analysis showed that in an unselected population of first time IVF patients, estradiol level was significantly related to better pregnancy rates (Table 2), in patients having only good or fair quality embryos transferred, (to control for embryo quality influence), estradiol level was no longer related to pregnancy outcome (Table 3). Patients in the lowest estradiol group required the greatest amount of FSH, suggesting a significant component of this group was made up of poor responders. Since our primary interest was in understanding the relationship between live birth and elevated estradiol, we conducted a univariate analysis on the entire population to evaluate the association of IVF response variables with LB outcome. This analysis showed that patient age, estradiol level, FSH dose, number of MII oocytes, total number of 2PN embryos, AEQS and day of transfer (day 3 or day 5) were all related to LB rate. After multiple regression to adjust for confounders and to test for interactions, the only main effects that significantly affected LBR were, AEQS (P < 0.001), the transfer of two embryos (P < 0.01) and the patient’s age (P < 0.03). The level of estradiol had no effect on LB rate (P = 0.308). We also found significant two-way interactions between transfer on day 5 and patient age (P < 0.01) as well as transfer on day 5 and FSH dose (P < 0.04). We interpret these interactions to mean that patients achieving a LB after transfer on day 5 were younger, and also required less FSH.

Some studies [20, 21] have reported that estradiol was elevated in non-pregnant patients compared to pregnant patients suggesting that an elevated estradiol level may be associated with not achieving pregnancy, although in one study [20] this was seen in patients having a day 3 transfer but not on day 5. To examine this question, we partitioned patients on the basis of LB status by day of transfer. Results from our study (Tables 6 and 7) and another study [17] do not support this finding since estradiol was either higher or the same in LB patients compared to NLB patients.

Others [17, 20, 22] have proposed that elevated estradiol may have a negative influence on pregnancy but only when embryo transfer is performed on day 3, arguing that a supra-physiological estradiol environment may cause embryo-endometrial asynchrony which is alleviated if transfer is delayed until day 5. To address this question, we evaluated a subset of patients that had an estradiol level of > 3000 pg/ml on the day of hCG taken from the entire population (described in Table 2), which compares IVF cycle response variables for patients that had embryo transfer on day 3 or day 5. We chose to use the same estradiol cutoff range (>3000 pg/ml) used in a recent publication [20]. Our results (Table 8) show that in contrast to this report, we found no statistical difference in CPR or LBR in patients undergoing embryo transfer on day 3 or day 5 although CPR and LBR were slightly higher for patients having transfer on day 5. The study of Elgindy et al. [20] was a randomized controlled trial that had strict exclusion criteria (patient age <35 with regular cycles and a minimum of 4 good quality embryos on day 3 of culture). By design, the estradiol level, number of oocytes retrieved, number of 2PN embryos and number of good quality embryos were all similar for the day 3 and day 5 transfer groups. Although ours was a retrospective cohort study we believe it is more reflective of a typical population of patients undergoing an initial IVF cycle. In contrast to the referenced study [20] the day of embryo transfer in our study was decided based on the number and quality of embryos, an approach similar to that used in an earlier study [22]. Therefore some bias exists which placed more patients having numerous high quality embryos in the group having embryo transfer on day 5. In our study estradiol levels were not different between patients having transfer on day 3 or day 5 but there were a significantly greater number of oocytes retrieved and 2PN embryos with fewer embryos being transferred on day 5. Of the patients used in our study, 17% were diagnosed as having PCOS using the Rotterdam criteria [24] and therefore probably constituted a significant number of high responding patients. These differences may at least partially explain the divergent conclusions drawn from the two studies. To further examine if patients having an elevated level of estradiol on day 3 had a different pregnancy rate compared to patients having a moderate level of estradiol undergoing embryo transfer on day 3, we performed an additional subgroup analysis. Our high estradiol group had a level of >4200 pg/ml and the moderate estradiol group had a level >3000 and < 4200 pg/ml, which are the same ranges used by Elgindy et al. [20]. Our results, shown in Table 9, demonstrate that the average estradiol level and number of oocytes collected were higher in the high estradiol group but there was no significant difference in CPR or LBR when compared to the moderate estradiol group, which is a conclusion not supported in the previous referenced study [20]. Understanding and managing the high responding patient is a challenge to physicians in the clinical setting of IVF. There is an increased chance of patients developing ovarian hyper-stimulation syndrome (OHSS), which in our study was low, with only four patients out of 489 (0.08%) exhibiting moderate/severe OHSS that required paracentesis, all of which were in the moderate estradiol group (2000-4000 pg/ml) described in Table 2 and none of which had any long term adverse effects. Aside from the possibility of OHSS, much has been written about supra-physiological levels of estradiol and progesterone potentially causing endometrial-embryo asynchrony which could negatively impact pregnancy rates.

Several reports have shown that in high responding patients there is a concomitant increase in estradiol, progesterone and the number of retrieved oocytes [25‐27] which may have a physiological linkage. In the case of high responders, the relationship between elevated levels of estradiol and progesterone was recently examined. Wu et al. [18] found that elevated levels of estradiol were not detrimental to live birth rate when progesterone levels were < 1.05 ng/ml whereas; when progesterone levels exceeded this threshold LB rates were compromised. Another study [27] examined this same relationship except the focus was on evaluating different progesterone thresholds. Their conclusion was that high progesterone correlated with high estradiol levels and that in high responders’ progesterone levels did not show any clinical impact, although a slight compromise was observed when a threshold of 1.8 ng/ml was exceeded. This study confirmed earlier work [28] which had shown that progesterone levels >1.5 ng/ml on the day of hCG reduced the chance of pregnancy in low and normal responders but not in high responders. On the basis of our results and the results of others, high responding patients are different than poor or normally responding patients regarding pregnancy outcomes and therefore this category of patients should be evaluated aside from the general population of patients. For this reason more large scale studies need to be performed that will elucidate if high responding patients are subject to hormone thresholds that if exceeded, could impair live birth rates. A limitation of our study was that progesterone levels on the day of hCG were not included, but unfortunately these values were not available. It is also important to point out that it is generally accepted that the supra-physiological hormone environment found in high responding patients does not have a negative effect on embryo quality or developmental potential. This has been borne out by studies in which embryos exposed to this environment and transferred to embryo recipients [11] or cryopreserved and transferred in a subsequent cycle [6] had no adverse effect on pregnancy rates. Results from our study indicate that, at least in the case of elevated levels of estradiol, no adverse effect on live birth rate is seen, regardless if embryo transfer was conducted on day 3 or day 5. Rather, our results show that the quality of transferred embryos and patient age are the primary determinants of having a live birth.

We believe that our study has demonstrated that after controlling for embryo quality, an elevated level of estradiol on the day of hCG is neither directly detrimental nor advantageous to having a live birth. Our study shows that average embryo quality at transfer is the primary determinant in establishing a LB while patient age and the transfer of two embryos are also important. Significant interactions showed that there is interplay occurring between embryo transfer on day 5 and a lower dose requirement of FSH and younger patient age, both of which appear to contribute to an increased LB rate. An elevated level of estradiol should be viewed in the context of being part of the profile of a “high responder” requiring less FSH, more oocytes retrieved, more mature oocytes and more 2PN embryos. Yet, it is interesting that all of these response variables were found to not impact LB rate. On the basis of comparable live birth rates, our data also suggest that embryos of high quality appear to tolerate an arguably sub-optimal uterine environment that has been exposed to supra-physiological levels of estradiol.

While our study has demonstrated that elevated estradiol has no effect on live birth rate, in the light of some recent studies [29‐31] showing that elevated estradiol may interfere with some aspects of placentation which in turn can contribute to an increased incidence of preeclampsia and small for gestational age (SGA) offspring, it is imperative that larger prospective studies be conducted to determine if this hormone environment is responsible for these effects. If the initial findings are substantiated perhaps it is time to seek a more comprehensive definition of IVF success that encompasses fetal and neonatal well-being and to look beyond live birth as the singular metric by which IVF success is defined.