全文

Abstract

The human chorionic gonadotrophin (HCG) trigger used for final follicular maturation in connection with assisted reproduction treatment combines ovulation induction and early luteal-phase stimulation of the corpora lutea. The use of a gonadotrophin-releasing hormone agonist (GnRHa) for final follicular maturation has, however, for the first time allowed a separation of the ovulatory signal from the early luteal-phase support. This has generated new information that may improve the currently employed luteal-phase support. Thus, combined results from a number of randomized controlled trials using the GnRHa trigger suggest an association between the reproductive outcome after IVF treatment and the mid-luteal-phase serum progesterone concentration. It appears that a minimum mid-luteal progesterone threshold of approximately 80–100 nmol/l exists, which, when surpassed, results in reduced early pregnancy loss and an increased live birth rate. Further, the trade off between the HCG bolus and the subsequent risk of ovarian hyperstimulation syndrome has resulted in a trend to reduce the HCG bolus from 10,000 IU to 6500–5000 IU, which augments the HCG/LH deficiency during the early/mid-luteal phase. The mid-luteal HCG/LH shortage results in an altered progesterone profile, showing the highest concentration during the early luteal phase, contrasting with the mid-luteal peak seen in the natural menstrual cycle.

Treatment of infertility involves administration of different drugs that affect the reproductive organs and collectively increases chances of becoming pregnant. Often treatment is divided into different phases such as: (i) pituitary down-regulation; (ii) stimulation of follicular growth; (iii) final maturation of preovulatory follicles; (iv) in-vitro culture of oocytes and embryos; and (v) luteal-phase support. In particular, luteal-phase support has not received the same interest as the other phases of treatment, and administration of exogenous progesterone is considered the gold standard for luteal-phase support. However, recent randomized controlled trials involving the use of a gonadotrophin-releasing hormone agonist for final maturation of follicles have revealed that the luteal phase may be optimized and result in an augmented reproductive outcome. There appears to be a minimum threshold limit of mid-luteal-phase progesterone that needs to be surpassed to get the best chances of implantation. Current luteal-phase support does not, in all circumstances, reach this threshold or reaches it too early during the luteal phase. Further, the shape of the progesterone profile during the luteal phase with the current treatment may not be optimal. Based on these observations new ways to optimize the luteal-phase support are discussed.

Keywords: corpus luteum function; early pregnancy loss; GnRH agonist trigger; low dose hCG; luteal phase; progesterone
 

Introduction

Ovarian stimulation protocols for patients who receive fertility treatment have undergone numerous refinements during the last two to three decades. Efforts to optimize ovarian stimulation protocols during the follicular phase have been the focus of many clinical trials and studies in which the whole armamentarium of drugs affecting follicular development has been tested in different combinations. In parallel, a huge effort has been made to optimize treatment outcomes by introducing the highest standards of embryo culture and laboratory techniques. The two remaining essential parts for a successful assisted reproduction treatment – the type of stimulation used for final follicular maturation and the luteal-phase support – have received less interest. Thus, final follicular maturation and rescue of the disrupted luteal phase seen after ovarian stimulation has almost exclusively been performed by the use of a large bolus of human chorionic gonadotrophin (HCG; i.e. 5000–10,000 IU), which, apart from the induction of the ovulatory processes, also directly stimulates the endogenous progesterone production by the multiple corpora lutea during the early luteal phase. The two combined effects of the HCG bolus trigger merge the period of events leading to final maturation of oocytes and appropriate luteal-phase stimulation prior to implantation.

However, the use of an HCG bolus for final follicular maturation as the golden standard has recently been challenged by the use of a GnRH agonist (GnRHa) trigger in patients cotreated with a GnRH antagonist. The GnRHa trigger concept utilizes a different mechanism to induce final follicular maturation than that of HCG, resulting in the release of an endogenous surge of both LH and FSH from the pituitary. The GnRHa trigger was originally developed in the early 1990s (Andersen et al., 1993, Gonen et al., 1990 and Itskovitz et al., 1991), but cannot be used in connection with the use of GnRHa for pituitary down-regulation, as in the long agonist protocol, and has only received renewed interest in parallel to the introduction of the antagonist protocol in the early part of the 21st century (Felberbaum et al., 2000).

Although the GnRHa-elicited surge is sufficient to induce final maturation of oocytes with a similar efficacy as HCG (Acevedo et al., 2006, Bodri et al., 2009 and Griesinger et al., 2007), it is short lived and does not provide the continuous prolonged stimulation of the early corpus luteum that HCG does. The separation of the ovulatory signal from the stimulation of the early corpus luteum by the use of the GnRHa trigger for the first time allowed studies that directly addressed either one of these two events separately. Indeed, despite good oocyte and embryo quality, an initial substantial effort was required to develop support for the luteal phase after GnRHa trigger in order to obtain results similar to the use of the HCG trigger (Humaidan et al., 2005, Humaidan et al., 2006, Humaidan et al., 2010, Humaidan et al., 2013 and Kolibianakis et al., 2005).

Importantly, these direct studies of the luteal-phase support have offered new insights as to how the current luteal-phase support may be improved. The aim of the present review is to discuss these issues and to suggest modes of possibly optimizing the luteal phase after ovarian stimulation.

Hormone concentrations during the normal luteal phase and after ovarian stimulation: the functions of progesterone
In response to the mid-cycle surge of gonadotrophins or to an exogenous HCG bolus administration, the corpus luteum is created by the transformation of granulosa and theca cells into luteal cells. The most important function of the corpus luteum is progesterone secretion, as progesterone is necessary to obtain a secretory transformation of the endometrium, crucial for successful implantation. Further and equally important, progesterone is indispensable for the maintenance of early pregnancy. Thus, an insufficient progesterone concentration at the time of implantation or during early pregnancy may cause early pregnancy loss (EPL) or lack of implantation. The incidence of luteal-phase deficiency has been reported to be in the range of 3.5–20% of infertile patients (Balasch et al., 1995).

In addition to supporting endometrial development, progesterone is thought to facilitate implantation by promoting the immune system to produce noninflammatory T-helper-2 cytokines (Druckmann and Druckmann, 2005 and Szekeres-Bartho et al., 2008). Moreover, improvement of blood flow and oxygen to the endometrium is also accomplished through progesterone actions by increasing nitric oxide production (Simoncini et al., 2006 and Sladek et al., 1997). Finally, progesterone reduces the contractility of the myometrium at the time of implantation, which is believed to promote implantation (Hill et al., 1990).

Throughout the luteal phase of the normal menstrual cycle, LH concentration is confined to a relatively narrow range of about 4–10 IU/l. This concentration is sufficient to induce a peak of progesterone during the mid-luteal phase, normally occurring approximately 7 or 8 days after the mid-cycle surge (unless pregnancy occurs). A progesterone concentration of 25 nmol/l during the mid-luteal phase of the natural cycle is considered to reflect ovulation and a normally functioning corpus luteum. However, progesterone concentration often varies and rises to considerably more than this, even after monofollicular ovulation. Moreover, peripheral progesterone concentration starts to increase in connection with ovulation, exceeding 20 nmol/l about 3 or 4 days after the mid-cycle surge (Groome et al., 1996).

Ovarian stimulation typically involves pituitary desensitization either in the form of GnRHa or GnRH antagonist administration. In the early luteal phase, this often results in reduced endogenous release of gonadotrophins (Messinis, 2006), and the HCG activity from the HCG bolus is necessary to maintain functional corpus luteum. Furthermore, due to the multiple corpora lutea, supraphysiological concentrations of oestradiol and progesterone are observed during the luteal phase following ovarian stimulation, which, by negative feedback, reduce the amount of LH released from the pituitary. Therefore, the continuous release of progesterone from the corpora lutea is dependent on exogenous LH-like activity (or HCG produced by an implanting embryo). Due to the multiple corpora lutea and the use of a large bolus of HCG for final follicular maturation, progesterone concentration during the mid-luteal phase often rises to 100–250 nmol/l or more, and its concentration is positively associated with the dose of HCG used for luteal-phase support (Humaidan et al., 2010 and Humaidan et al., 2013).

Luteal-phase support after GnRHa trigger

This study group has performed three randomized controlled trials in which a GnRHa trigger was compared with HCG for final follicular maturation (Humaidan et al., 2005, Humaidan et al., 2010 and Humaidan et al., 2013; Table 1). In all trials, a standard stimulation protocol was used with GnRH antagonist cotreatment. In addition to keeping ovarian stimulation and the follicular phase constant, the type and dose of GnRHa for trigger was identical for the three trials. The demographic characteristics of patients included were similar and the same laboratories handled the in-vitro procedures. Thus, one might assume that the embryos deriving from these trials were of a similar quality. However, modifications of the luteal-phase support were performed between the studies. Notably, the differences in luteal-phase support had a profound impact on the mid-luteal progesterone concentrations measured 7 days after oocyte retrieval (Table 1) and on the reproductive outcome. Importantly, the early pregnancy losses, defined as positive pregnancy tests which did not result in a viable intrauterine pregnancy during the first trimester, were inversely correlated with the mid-luteal serum progesterone concentration (Figure 1). Thus, it appears that there is a sharp decline in the EPL rate from about 80% to 10% as the mid-luteal-phase progesterone concentration increases from about 40 nmol/l to about 80–100 nmol/l. In contrast, increasing the mid-luteal progesterone concentration beyond 100 nmol/l does not appear to reduce the EPL rate further; it maintains it at a low rate of approximately 10%. Assuming that embryos in these studies were of a similar quality, these observations infer a reduced capability of the endometrium to sustain the early implantation when mid-luteal progesterone concentrations are below approximately 80 nmol/l.




 

Interestingly, similar results have been obtained previously. (Liu et al. (1995) demonstrated in 341 patients undergoing ovarian stimulation for IVF that patients with viable pregnancies had significantly higher mean serum progesterone concentrations during the preimplantation period of the early luteal phase than patients not pregnant or with early miscarriage. On the day of implantation, 73% of viable pregnancies, 42% of clinical abortions and 20% of preclinical abortions had a progesterone concentration exceeding 95 nmol/l (Liu et al., 1995). Further, Mitwally et al. (2010) in a study of 544 women undergoing IVF treatment also found a significant positive association between the mean serum progesterone concentration during the luteal phase and the clinical pregnancy rate. Ellenbogen et al. (2004) found by assessing 224 IVF cycles that mid-luteal progesterone concentrations were significantly higher (15.3 ± 4.9 ng/ml versus 12.9 ± 4.9 ng/ml, P < 0.002) in patients that became pregnant compared with those who did not. In pregnant patients, 53% of them had mid-luteal progesterone exceeding 15 ng/ml (vs. 31% in nonpregnant patients, P < 0.01). The group of patients with progesterone <15 ng/ml (47%) had the exogenous progesterone dose increased in the second half of luteal phase and the study concluded that increasing the dosage of progesterone administration in these patients was beneficial.

Thus, it appears that the EPL rate may act as a quality marker for the luteal-phase support. Depending on the type of patient – mainly the age of the woman – the EPL rate is approximately 20–30% of patients having a positive pregnancy test. This loss is usually considered unavoidable, as it has until now been assumed to be linked to chromosomally abnormal embryos. However, an EPL may result not only from chromosomally abnormal embryos, but also from malfunctions of the endometrium or both. This fact then raises the question as to whether EPL, despite the transfer of high-quality embryos, is related to the embryo or the uterus and, importantly, what impact the mid-luteal-phase progesterone concentration has on endometrial receptivity. Indeed, one might ask whether there is a minimum progesterone threshold concentration in IVF above which the chance of an ongoing pregnancy is increased.

Although some early studies suggested an association between conception and the mid-luteal-phase progesterone concentration and the fact that low luteal-phase progesterone concentrations may compromise fertility (Yovich et al., 1985a, Yovich et al., 1985b and Hammond and Talbert, 1982) other studies have failed to find an association (Laufer et al., 1982 and Sallam et al., 1999). However, a recent large intrauterine insemination study found an association between live birth and mid-luteal progesterone concentration (Arce et al., 2011). In a group of WHO-II women undergoing ovarian stimulation and ovulation induction with a bolus of HCG prior to intrauterine insemination, the authors reported a positive correlation between live birth and mid-luteal-phase progesterone concentration. Further, this study developed a mathematical model to calculate the chances of a live birth in relation to the mid-luteal progesterone concentration, which was valid also for women who developed no more than one follicle, indicating that the number of corpora lutea was not the decisive parameter (Arce et al., 2011).

In addition, this study group recently evaluated the reproductive outcome in patients receiving frozen–thawed embryo transfer before and after doubling vaginal progesterone gel supplementation. The vaginal progesterone dose (Crinone) was doubled from 90 mg to 180 mg, resulting in a significant decrease in the EPL rate (67% versus 44%, respectively; P = 0.014) and a significant increase in the delivery rate (9% versus 21%, respectively; P = 0.002) ( Alsbjerg et al., 2013). Although the mid-luteal progesterone concentration was not measured, a higher exogenous dose of progesterone is likely to increase the endogenous progesterone exposure.

Taken together, the most recent results suggest that the EPL rate might be reduced and that the ongoing pregnancy rate could be increased by securing a higher serum progesterone concentration around the time of implantation. The minimal progesterone concentration during the mid-luteal phase associated with establishment of a pregnancy is not absolute, since concentrations exceeding 25 nmol/l seem sufficient during the natural menstrual cycle, whereas concentrations need to surpass 80–100 nmol/l after ovarian stimulation and GnRH antagonist cotreatment, followed by a GnRHa trigger.

These findings fit nicely with original studies from Hull and coworkers, who reported a mid-luteal-phase progesterone concentration of 30 nmol/l as the lower limit for conception in natural menstrual cycles, whereas this concentration was raised by a factor of about 3 (i.e. ∼90 nmol/l) in gonadotrophin-stimulated cycles (Hull et al., 1982). Moreover Yovich et al. (1986) also found progesterone concentrations to be 2–3-times higher during conception cycles when unstimulated and stimulated cycles were compared. Finally, the progesterone concentration to secure implantation and ongoing pregnancy in patients receiving frozen–thawed embryo transfer may be different from that of patients receiving ovarian stimulation.

The variable low threshold concentration of progesterone during the luteal phase is not readily explained, but may relate to: (i) the oestradiol concentration during the follicular phase; and (ii) the profile and amplitude of the progesterone curve in the early luteal phase. The oestradiol concentration during the follicular phase may have an impact on the progesterone concentration necessary to prepare a receptive endometrium during the luteal phase (Balasch et al., 1995, Molo et al., 1995 and Radwanska et al., 1980). In support of this theory, Wilcox and coworkers, in a group of volunteers attempting to conceive, found that when comparing nonconception cycles to conception cycles, conception cycles had a steeper early luteal-phase progesterone rise and higher mid-luteal oestrogen and progesterone concentrations, suggesting that these hormonal characteristics could be better markers of ‘quality cycles’ with a higher chance of conception; however, the authors were unable to more closely define such cycles (Baird et al., 1997). As will be discussed further, the profile and amplitude of the progesterone curve in the early luteal phase of stimulated cycles is very different from that of natural cycles and may therefore have different impact on the maturation of the endometrium.

Whether a ceiling concentration of progesterone exists during the mid-luteal phase is currently unknown. However, if it exists it is likely to be very high. As a consequence of the highly variable HCG concentration during early pregnancy, progesterone concentration will mirror concentrations of HCG and, thus, will also be highly variable; however, in contrast to HCG, progesterone concentration does not continue to increase during early pregnancy (Yovich et al., 1985a and Yovich et al., 1985b). Currently, it is probably difficult to predict which women will experience a too-low concentration of progesterone during the mid-luteal phase, and since there does not appear to be an upper concentration, this review suggests increasing progesterone concentration, especially during the mid-luteal phase, in all patients.

HCG concentration during the mid-luteal phase after HCG trigger
The relatively long half-life of HCG as compared with LH allows HCG to act as a strong luteotrophic signal during the first half of the luteal phase. Based on the reported half-life of HCG (Trinchard-Lugan et al., 2002), the shapes of the HCG curves during the luteal phase after administration of 5000, 6500 and 10,000 IU HCG are depicted in Figure 2. Further, the concentration of HCG secreted by the implanting embryo during the second half of the luteal phase is also depicted.



 

The concentration profile of HCG demonstrates that its serum concentration exceeds 100–150 IU/l 1 or 2 days following the trigger injection (Figure 2). This peak is followed by a slow clearance, and on days 5 or 6 after oocyte retrieval the concentration is reduced to a one that stimulates progesterone production only to a limited extent.

The implanting embryo embarks on HCG production during the second half of the luteal phase. However, the timing of HCG production is variable and differs from embryo to embryo and it also differs between embryos resulting in an ongoing pregnancy and those resulting in an EPL (Shapiro et al., 2012), being significantly slower in the latter. Irrespective of whether the embryo is a fast or slow implanter or whether embryos fail to result in an ongoing pregnancy with the current trigger and luteal-phase support policy, there is a considerable mid-luteal reduction in LH-like activity, during which the corpus luteum does not receive sufficient stimulation (Mitwally et al., 2010). This gap in HCG/LH takes place exactly during the period during which implantation occurs and when progesterone concentration would normally peak during the natural menstrual cycle. In order to alleviate the effects of too-low HCG/LH concentrations during the mid-luteal-phase, exogenous progesterone is usually administered in ovarian stimulation. However, the circulating progesterone concentration obtained after exogenous administration alone is relatively modest, being about 30–40 nmol/l (Levine and Watson, 2000) and whereas the administration of exogenous progesterone undoubtedly exerts a positive effect on the reproductive outcome after IVF/intracytoplasmic sperm injection (Van Steirteghem et al., 1998), it is an open question whether this supplementation is sufficient to secure the most optimal reproductive outcome in all patients, especially nowadays when the trend is to reduce the dose of HCG for ovulation induction (e.g. to 6500 or 5000 IU) in order to reduce the risk of ovarian hyperstimulation syndrome (OHSS).

In rhesus monkeys, it has been shown that a 3-day withdrawal of gonadotrophin support during the early and mid-luteal phases does not result in an irreversible loss of corpus luteum responsiveness to further gonadotrophic stimulation (Hutchison and Zeleznik, 1985). However, these authors also showed that the disappearance of circulating LH was followed by a rapid fall in plasma progesterone concentration, regardless of the stage of the luteal phase. Further, it is unknown to what extent the subsequent progesterone production following a temporary reduction in HCG/LH activity is sufficient to support optimal implantation or whether this will increase the EPL rate.

Collectively, this review suggests that an insufficient corpus luteum stimulation followed by low endogenous progesterone production, especially during the crucial mid-luteal phase, may be linked to a poorer reproductive outcome in subgroups of women who undergo fertility treatment.

Profile and amplitude of serum progesterone concentration after administration of an HCG bolus and exogenous progesterone supplementation
The sustained luteotropic effect of the HCG bolus exerts a profound stimulation on the corpus luteum and progesterone concentrations have been reported to exceed 150 nmol/l on the day of embryo transfer (Fauser et al., 2009). Further, the maximal concentration of progesterone is reached at embryo transfer and starts to decline after 1 week (Fauser et al., 2002 and Mochtar et al., 1996). This clearly shows that the circulatory luteal-phase progesterone profile in women receiving an HCG bolus differs from that of the natural menstrual cycle, during which the progesterone concentration rises relatively modestly during the early luteal phase, only to peak during the mid-luteal period. One negative effect of this massive output of progesterone during the early luteal phase after an HCG trigger could be an advancement of the endometrium, resulting in asynchrony between the implanting embryo and the endometrium. Further, this may explain why better pregnancy rates appear to be achieved when frozen embryos are transferred in a natural cycle as compared with after an HCG bolus (Shapiro et al., 2011). Not all clinics obtain similar good cryopreservation results and it is an open question as to whether a freeze-all approach should be used, with the exception of patients with an imminent risk of OHSS since these patients are obviously anxious to pursue their chances of becoming pregnant.

Taken together, this review suggests that the difference in shape and height of the progesterone curve during the luteal phase of a stimulated cycle triggered with a bolus of HCG, as compared with the natural cycle, may affect the receptivity of the endometrium.

Is it possible to improve implantation rates by the administration of low-dose HCG for luteal-phase support?

The use of HCG for luteal-phase support is widely accepted, and it was recognized early on that it was necessary to reach acceptable pregnancy rates in WHO group I patients (Messinis et al., 1988). However, HCG administration is associated with an increased risk of OHSS: the higher the HCG trigger bolus, the higher the risk of OHSS development. Thus, the current use of HCG as a trigger is a trade off between the necessary early luteal-phase support and the risk of OHSS development. However, irrespective of the HCG dose used, there is a period of HCG deficiency during the mid-luteal phase (Figure 2). There are no drugs registered for covering this gap, but as an off-label use it can be suggested to administer low-dose HCG activity to achieve LH activity in the physiological range. This may be achieved in the form of low-dose HCG administration (e.g. 100 IU daily) resulting in a steady-state HCG concentration of 6–7 IU/l during the luteal phase (Thuesen et al., 2012), preferentially in combination with the use of 5000 IU HCG or GnRHa trigger. In contrast to that observed when using, for instance, 1500 IU HCG once or twice during the luteal phase, which results in circulating HCG concentrations of about 50–100 IU/and therefore poses a higher risk of OHSS, such a dosing regime will result in a low but sufficient concentration of HCG to stimulate progesterone production by the corpus luteum. Luteal-phase stimulation with a HCG concentration of LH-like activity in the physiological range may potentially result in both a reduced risk of OHSS development and a sufficient luteal-phase support, especially during the critical time of implantation which could optimize the reproductive outcome.

Alternatively, recombinant LH may be used for luteal-phase support in connection with the use of an agonist trigger. Papanikolaou et al. (2011) demonstrated that 300 IU recombinant LH every other day during the luteal phase provided reproductive outcomes equal that those with a conventional HCG trigger protocol, without any examples of OHSS. However, the financial cost of recombinant LH makes this option less attractive as compared with the use of low-dose HCG administration.

Conclusions

Collectively the present review suggests that: (i) a minimum serum progesterone threshold exists in ovarian stimulation, which needs to be surpassed in order to obtain the most optimal reproductive outcome: this threshold is proposed to be 80–100 nmol/l during the mid-luteal phase in connection with ovarian stimulation; (ii) after a standard bolus of HCG used for trigger in ovarian stimulation, an early–mid-luteal-phase HCG/LH deficiency exists during which the corpus luteum lack an optimal stimulation: at present it is unknown whether the corpus luteum will function optimally following an LH-like activity deprivation period of several days; and (iii) the early/mid luteal-phase progesterone profile following a large bolus of HCG differs markedly from that of the natural menstrual cycle and that of GnRHa trigger: the impact of supraphysiological progesterone concentrations on the endometrium is currently unknown.

It is suggested that the luteal-phase support could be optimized in parallel with a reduction in OHSS risk by the use of either GnRHa or a lower dose of HCG for triggering final oocyte maturation in combination with low-dose HCG or recombinant LH luteal supplementation. Most recent research seems to support the idea that optimization of the luteal-phase progesterone profile could result in a higher ongoing pregnancy rate and, thus, a reduced EPL rate.

Acknowledgements

The Novo Nordic Foundation, Sophus Carl Emil Friis and wife Olga Doris Friis’ foundation, the Lundbeck Foundation, and the University Hospital of Copenhagen are gratefully acknowledged.