End of stimulation endocrine environment and the impact on clinical outcomes
It is essential to consider all aspects when choosing the most appropriate gonadotrophin therapy for OS as different treatments could affect the end-of-stimulation endocrine environments, consequently impacting clinical outcomes and the overall goal of achieving a live birth. The MERiT trial (a prospective, randomised, controlled, multicentre study), primarily investigated the clinical outcome of 731 IVF patients treated with HP-hMG (Menopur®) versus rFSH alfa (Gonal-F®), while also collecting data on the endocrine profiles achieved in the patients following treatment [
11,
84]. The trial was the subject of a
post-hoc analysis carried out by Smitz et al., who found that, although there was no significant difference on stimulation Day 6 (
P = 0.333), progesterone levels were significantly higher at the end of stimulation in the rFSH alfa group versus HP-hMG treatment (23% higher on the last day of stimulation, 3.4 ± 1.7 nmol/L vs 2.6 ± 1.3 nmol/L,
P < 0.001; and 31% higher at oocyte retrieval, 36.3 ± 25 vs 24.5 ± 15.6,
P < 0.001) [
84]. The significance remained when adjusting for ovarian response (28% higher at the end of stimulation when adjusting for the number of follicles and 29% higher at oocyte retrieval when adjusting for the number of oocytes retrieved). In addition, a higher number of patients developed progesterone levels of > 4 nmol/L at the end of stimulation in the rFSH alfa group compared with the HP-hMG group (23% vs 11%, respectively), which was linked with reduced pregnancy rates [
11,
84]. Oestradiol levels were found to be significantly higher by 20% in the rFSH alfa group on Day 6 of stimulation (1.1 ± 1.0 vs 1.0 ± 0.9,
P = 0.004), 10% higher in the HP-hMG group by end of stimulation (7.2 ± 4.3 vs 6.6 ± 4.0,
P = 0.031) and 16% higher at oocyte retrieval (3.9 ± 2.1 vs 3.4 ± 1.9,
P = 0.001) [
84]. Androstenedione levels were significantly increased in the HP-hMG versus rFSH alfa group at stimulation on Day 6, last stimulation day and after oocyte retrieval (Day 6: 6.0 ± 2.5 vs 5.5 ± 2.4,
P = 0.002; last stimulation day: 11.9 ± 5.2 vs 9.5 ± 3.8,
P < 0.001; oocyte retrieval: 13.6 ± 5.5 vs 10.8 ± 4.2,
P < 0.001). When considering these results in combination with the outcomes of ongoing pregnancy rate (OPR) and live birth rate (LBR) from the primary study (OPR: 27% with HP-hMG group vs 22% with rFSH alfa,
P = 0.204; LBR: 26% HP-hMG vs 22% rFSH alfa,
P = 0.236), the data suggest that the differences in outcome observed between the HP-hMG and rFSH alfa groups may have been influenced by the different endocrine profiles induced by the respective treatments [
11].
This hypothesis is further supported by the results from a retrospective study that analysed progesterone levels and OPR in more than 4000 OS cycles (IVF and intracytoplasmic sperm injection [ICSI] included in the analysis) and found that OPRs are inversely correlated with progesterone levels on the day of hCG administration. In addition, this study indicated that pregnancy rates are significantly higher in women with progesterone levels < 1.5 ng/mL than in those with progesterone levels > 1.5 ng/mL [
85]. Furthermore, a meta-analysis of more than 60,000 fresh, frozen-thawed and donor/recipient IVF cycles, investigating the link between progesterone levels on the day of hCG administration and pregnancy outcomes, found that elevated progesterone levels are associated with a reduced probability of pregnancy in fresh cycles only (
P < 0.05 for all thresholds defined as ≥ 0.8 ng/mL progesterone) [
86]. Interestingly, elevated progesterone levels on the day of hCG administration were not associated with a reduced probability of pregnancy after frozen and donor/recipient IVF cycles, suggesting that a premature rise in progesterone levels during the follicular phase may act directly on the endometrium and impair its normal development. This may result in asynchrony between the embryo development and endometrium receptivity through advanced secretory transformations of the endometrium, thus shifting the implantation window so that it occurs earlier than normal in fresh cycles. A shifted implantation window could have subsequent detrimental effects on implantation and explain the poorer pregnancy outcomes that have been observed in clinical trials [
87‐
93].
During assisted reproduction, ovaries are artificially stimulated in order to induce the development of multiple follicles for oocyte retrieval. In line with the ‘two-cell, two-gonadotrophin theory’, administration of supraphysiological levels of FSH stimulates the development of multiple follicles, consequently increasing the production of progesterone [
94]. Treatment supplementation with LH or hCG during the stimulation is thought to instead drive the conversion of pregnenolone to androgens (via the Delta 5 pathway), which are further converted by FSH into oestrogens, thereby limiting the conversion of pregnenolone to progesterone [
95,
96]. As a result, less progesterone is available to enter the bloodstream [
88]. Data from a recent clinical trial that studied the follicular steroidogenesis pathways and progesterone levels in oocyte donors treated with either rFSH alfa or HP-hMG for OS support the importance of attaining a balance of FSH and LH bioactivity when stimulating using gonadotrophins [
97]. Indeed, there was a significant increase in serum progesterone following rFSH alfa stimulation compared with HP-hMG stimulation both on the day of trigger (0.68 ± 0.50 ng/mL vs 0.46 ± 0.27 ng/mL, respectively;
P = 0.010), and on stimulation Day 8. Consistently, treatment with HP-hMG was associated with a significant increase in androstenedione compared with rFSH alfa (3.0 ± 1.4 vs 2.4 ± 1.1, respectively;
P = 0.015) on both stimulation days, and a significantly higher pregnenolone:progesterone ratio on the day of trigger (
P = 0.019). Although the ovarian response remained comparable between the two groups (17.5 ± 7.9 oocytes retrieved in the rFSH alfa group vs 16.5 ± 7.5 in the HP-hMG group,
P = 0.49), these results indicate that different follicular steroidogenesis pathways are at play with rFSH alfa versus HP-hMG OS.
Several reviews have recently discussed the controversy surrounding the origin of premature progesterone elevation in OS based on evidence to date [
90,
93,
98]. Initially, it was hypothesised that the LH bioactivity of hMG preparations may cause elevations in circulating progesterone in the context of premature LH surges. Contrary to expectations, studies comparing circulating progesterone levels following the administration of FSH or hMG revealed that FSH was associated with a similar or an increased elevation in progesterone compared with hMG (progesterone levels increased from 0.37 ± 0.12 ng/mL before FSH administration to 0.86 ± 0.12 ng/mL approximately 15 h post-administration,
P < 0.01 [
93,
99]). These data suggest that FSH stimulation alone could be associated with elevated circulatory progesterone levels at the end of stimulation, but did not distinguish between the individual influences of hCG and LH. A study by Sebag-Peyrelevade et al. (2015) analysed whether supplementation of rFSH alfa treatment with rLH could lead to comparable progesterone levels on the day of hCG triggering compared with HP-hMG (Menopur®) treatment alone. The study, conducted on pituitary-desensitised IVF patients, revealed that rFSH alfa supplementation with rLH was not sufficient to diminish the circulating progesterone levels to the same level as those receiving HP-hMG (median of 0.63 ng/mL for HP-hMG vs 0.91 ng/mL for rLH/rFSH alfa median,
P < 0.0001). This relationship remained significant even when adjusted for the number of growing follicles to control for the extent of ovarian response (0.055 ng/mL/growing follicle vs 0.077 ng/mL/growing follicle) [
100]. A lack of effect was also reported following administration of a higher rLH dose in the rFSH alfa/rLH group: it was, therefore, suggested that the lower progesterone levels induced with HP-hMG treatment may be attributable to the hCG content (not LH) [
100].
Interestingly, embryo quality may be equally as important in determining pregnancy outcomes as the endometrial receptivity induced by elevated progesterone levels. It has been suggested that patients with a good ovarian response produce oocytes/embryos of increased quality, and so may be better equipped to overcome the challenges presented by an impaired endometrial receptivity brought about by high progesterone levels. In contrast, patients with a poor ovarian response are more likely to produce oocytes of lower quality, which are less likely to be able to compensate for the impaired endometrial receptivity, thus resulting in a lower OPR [
93,
98]. Data from multiple clinical trials have indicated that a progesterone level > 0.9 ng/mL on the day of hCG administration is associated with lower pregnancy rates in patients with a weak ovarian response. Therefore, it has been speculated that previous trials not reporting a link between premature elevated progesterone levels and reduced pregnancy outcomes may have unknowingly included in their analyses a majority of normal or high ovarian responders, with resulting good-quality embryos being capable of compensating for impaired endometrial receptivity [
93]. Taken together, these findings indicate that OS could potentially be adapted to prevent progesterone elevation by individualisation according to a patient’s predicted ovarian response and embryo quality [
101].
Number of oocytes retrieved and proportion of top-quality embryos achieved with different gonadotrophin preparations
Although live birth is the most clinically meaningful goal of ART, the number of oocytes retrieved following OS is frequently used as a surrogate measure of clinical success [
102]. Data from several studies indicate that rFSH alfa and beta trigger an increased ovarian response, in terms of number of oocytes retrieved, compared with HP-hMG. In a study by Bosch et al. (2008), in which 280 patients undergoing IVF/ICSI were randomised to receive HP-hMG (Menopur®) or rFSH alfa (Gonal-F®) in GnRH antagonist protocols, the numbers of cumulus–oocyte complexes (COCs) retrieved and metaphase II (MII) oocytes obtained (ICSI cycles) were significantly higher with rFSH alfa than HP-hMG (14.4 ± 8.1 vs 11.3 ± 6.0,
P = 0.001 and 9.7 ± 6.0 vs 7.8 ± 4.0,
P = 0.004, respectively) [
78]. The MERiT [
11] and MEGASET [
12] studies also reported higher numbers of oocytes retrieved in the rFSH alfa and beta groups (respectively) than in the HP-hMG group (11.8 ± 5.7 vs 10.0 ± 5.4,
P < 0.001 and 10.7 ± 5.8 vs 9.1 ± 5.2,
P < 0.001, respectively). These results were further replicated in the recently concluded MEGASET-high responder (MEGASET-HR) trial, in which 620 women predicted to be high responders were randomised to receive either HP-hMG (Menopur®) or rFSH alfa (Gonal-F(R)) in a GnRH antagonist cycle [
13,
14,
79]. As in previous studies, the number of oocytes retrieved was higher in the rFSH alfa arm than in the HP-hMG arm (22 ± 11.54 vs 15.1 ± 10.12) [
14,
79]. Interestingly, the PERSIST trial comparing a 2:1 formulation of rFSH alfa plus rLH (Pergoveris®) administered from Day 1 versus rFSH alfa (Gonal-F®) administered during Days 1–5 and supplemented with rLH from Day 6 of the stimulation cycle did not report a significant difference in terms of oocytes retrieved (9.7 vs 10.9, 95% confidence interval [CI]: − 3.15 to 0.59), therefore failing to meet its primary endpoint and yielding inconclusive evidence on the influence of rLH on oocyte yield [
82].
To understand whether differences in the source of LH activity affect ovarian stimulation characteristics (number of oocytes retrieved and percentage of mature oocytes) and IVF outcome, a critical appraisal of studies comparing hMG and rFSH alfa/beta + rLH was performed [
103]. Of the 11 studies included, most were observational, with only two randomised controlled trials (RCTs) evaluated. Moreover, only one of the RCTs included compared hMG and rLH from day 1 of ovarian stimulation [
104]. This RCT, performed on 111 patients, found that hMG was associated with a longer length of treatment but, surprisingly, a lower total amount of FSH administered compared with the rFSH alfa + rLH group. Notably, this study was performed in Italy at a time when only 3 oocytes were permitted to be inseminated by law, an additional reason to exert caution when interpreting these results. The author found that there was insufficient evidence to form any firm conclusions on whether the source of LH activity affects ovarian stimulation characteristics and concluded that further RCTs on this subject are needed [
103].
Although studies consistently show that rFSH alfa and beta stimulation is associated with the retrieval of a higher number of oocytes than HP-hMG, the data also indicate that rFSH alfa/beta and HP-hMG stimulation may influence the quality of the oocytes retrieved, and this, in turn, could impact clinical outcomes. In the MERiT study, a higher proportion of retrieved oocytes developed into high-quality embryos in the HP-hMG group than in the rFSH alfa group (11.3 ± 16.1% versus 9.0 ± 13.0%,
P = 0.044) [
11,
105]. The MEGASET trial then reported how, despite the higher number of oocytes retrieved in the rFSH beta group, the number and quality of blastocysts on Day 5 was comparable between the two treatment arms [
12]. Again, in the MEGASET-HR study, the difference in the number of oocytes retrieved, as calculated by the Hodges-Lehmann estimate, was seven fewer with HP-hMG compared with rFSH alfa, which narrowed to zero for excellent quality blastocysts (numerical difference of 0.9). MEGASET-HR also reported cumulative higher rates of early pregnancy loss in the rFSH alfa treatment arm, suggesting poor oocyte quality in comparison with the HP-hMG treatment arm [
79]. Taken together, these data suggest that treatment with rFSH alfa/beta is likely to result in a higher oocyte yield compared with HP-hMG, but that the quality of oocytes produced with HP-hMG is proportionally higher than that produced with rFSH alfa/beta. Oocyte quality-related issues may explain why the incremental oocyte yield with rFSH alfa/beta does not translate into a similarly augmented abundance in numbers of good-quality blastocysts between the two groups. Furthermore, these studies also highlight why oocyte yield is not a clinically meaningful endpoint to evaluate ART efficiency.