Abstract
The advent of embryo and oocyte vitrification today gives reproductive specialists an opportunity to consider new strategies for improving the practice and results of IVF attempts. As the freezing of entire cohorts does not compromise, and may even improve, the results of IVF attempts, it is possible to break away from the standard sequence of stimulation–retrieval–transfer. The constraints associated with ovarian stimulation in relation to the potential harmful effects of the hormonal environment on endometrial receptivity can be avoided.
This review will look at the new stimulation protocols where progesterone is used to block the LH surge. Thanks to ‘freeze all’ strategies, the increase in progesterone could actually be no longer a cause for concern. There are two ways of using progesterone, whether it be endogenous, as in luteal phase stimulation, or exogenous, as in the use of progesterone in the follicular phase i.e. progestin primed ovarian stimulation.
A literature search was carried out (until September 2016) on MEDLINE. The following text words were utilized to generate the list of citations: progestin primed ovarian stimulation, luteal phase stimulation, luteal stimulation, duostim, double stimulation, random start. Articles and their references were then examined in order to identify other potential studies. All of the articles are reported in this review.
The use of progesterone during ovarian stimulation is effective in blocking the LH surge, whether endogenous or exogenous, and it does not affect the number of oocytes collected or the quality of the embryos obtained. Its main constraint is that it requires total freezing and delayed transfer. A variety of stimulation protocols can be derived from these two methods, and their implications are discussed, from fertility preservation to ovarian response profiles to organization for the patients and clincs. These new regimens enable more flexibility and are of emerging interest in daily practice. However, their medical and economic significance remains to be demonstrated.
The use of luteal phase or follicular phase protocols with progestins could rapidly develop in the context of oocyte donation and fertility preservation not related to oncology. Their place could develop even more in the general population of patients in IVF programs. The strategy of total freezing continues to develop, thanks to technical improvements, in particular vitrification and PGS on blastocysts, and thanks to studies showing improvements in embryo implantation when the transfer take place far removed from the hormonal changes caused by ovarian stimulation.
Introduction
Improvements in techniques for freezing embryos and oocytes for ART mean that reproductive physicians are able, on the one hand, to respond to needs where previously it was not possible and, on the other, to consider new strategies for improving the practice and results of IVF attempts.
Until recently, the freezing technique used, called slow freezing, gave disappointing results for oocyte cryopreservation (Bernard and Fuller, 1996) and had a low yield for embryo freezing, giving an average survival rate of around 52% (Debrock et al., 2015). The first birth following oocyte vitrification was reported by a team from Australia and Italy in 1999 (Kuleshova et al., 1999), around 10 years after the first pregnancies resulting from embryo vitrification (Gordts et al., 1990; Feichtinger et al., 1991). These techniques became widespread throughout the world from 2004, and represented a real breakthrough in reproductive biology. They enabled oocyte and embryo cryopreservation to be achieved with an excellent survival rate of over 90%, and pregnancy chances equivalent to those with fresh transfers (Cobo et al., 2008, 2010; Rienzi et al., 2010) or even sometimes better (Roque et al., 2015). The practice of oocyte vitrification, which is effective and high-yielding, is also a major step forward in female fertility preservation, regardless of the context, compared to previously when freezing embryos was the only valid option apart from ovarian cortex preservation.
The second revolution we are witnessing with the development of a cryopreservation technique where the quality of embryo or oocyte can be kept the same as fresh, is the freedom to move away from the standard sequence of ovarian stimulation–retrieval–transfer, to which we were previously bound in order to provide the best chances of success. Today we are seeing the advent of so-called ‘freeze all’ strategies, that is the freezing of the entire cohort of embryos or oocytes, or a mixture. We can therefore consider ovarian stimulation without the constraints associated with the potential harmful effects of the hormonal environment on endometrial receptivity, on the one hand, and those associated with the risk of ovarian hyperstimulation syndrome (OHSS), on the other.
This second point will not be discussed here but briefly, the cumulative live birth rate increases with the number of oocytes collected (Drakopoulos et al., 2016). Thanks to freeze all strategies, stronger stimulations with agonist triggers can be used to suppress the risk of OHSS, without the fear of damage in the luteal phase or the risk of secondary hyperstimulation associated with early pregnancy.
This review will look at the new stimulation protocols where, thanks to freeze all strategies, progesterone is actually used as part of the protocol, instead of an increase in progesterone being a cause for concern. There are two ways of using progesterone, whether it be endogenous, as with luteal stimulation, or exogenous, as with the use of progesterone in the follicular phase (progestin primed ovarian stimulation (PPOS)). A variety of stimulation protocols can be derived from these two methods, and their potential in clinical practice remains to be determined: fertility preservation, ovarian response profile and organization for the patients and hospital centres.
Methods
A literature search was conducted using PubMed until September 2016. The following keywords were utilized to generate the list of citations: progestin primed ovarian stimulation, luteal phase stimulation, luteal stimulation, duostim, double stimulation, random start. A systematic review of English-language publications was carried out. Articles and their references were examined in order to identify other potential studies. All of the articles are reported in this review. The goal of this review is to provide an overview of the current data on the use of progesterone to block the LH surge during ovarian stimulation and its potential application in IVF.
Physiological principles
Use of progesterone to block the LH surge
Action of progestin on the LH surge. According to Heikinheimo et al., administration of the progestin levonorgestrel (LNG) at the beginning of the cycle in monkeys, blocks the occurrence of the LH surge. This action is reversible on discontinuation. Reprinted from Heikinheimo et al. (1996) with permission.
Waves of folliculogenesis during the menstrual cycle
Waves of follicle development. The graphs represent, for each day of the cycle, the average number (±SD) of follicles ≥ 5 mm (▪ ·) seen on ultrasound and the average diameter of the largest follicle (○ ·) for women with 2 (A) or 3 (B) waves of follicular growth. We can see that only the wave emerging in the early follicular phase provides a follicle which will exceed 11 mm in size and reach the ovulatory stage. Reprinted from Baerwald et al. (2003a) with permission.
Follicular phase ovarian stimulation
Usual ovarian stimulation regimens use gonadotrophins to promote multifollicular development associated with an analog of the GnRH to prevent the LH surge and premature ovulation. The analog improves the success of IVF cycles by optimizing oocyte retrieval and synchronization of the endometrium. Ovarian stimulation in these protocols starts in the early follicular phase or equivalent, either in short protocols using agonist or antagonist or after pituitary desensitization in long agonist protocols.
Luteal phase ovarian stimulation
Schematic representation of the protocols reported in this review. For each bibliographic reference cited in the review, the study protocol (progesterone block regimen) is represented and compared to the control protocol (conventional regimen); the table reports outcomes for each protocol.
The first European feasibility study (Buendgen et al., 2013) was published in 2013 by G. Griesinger's team in Luebeck, and it was not very optimistic. It found a lack of significance for this strategy because of the higher dose of FSH required for luteal phase stimulation. However, comparison of the two groups was flawed because of a higher starting dose for luteal stimulation. It was a prospective case-control study of good prognosis patients with ten in each arm. Luteal phase stimulation was started between Day 19 and Day 21 of a spontaneous menstrual cycle, after verification of progesterone levels >7 ng/ml, with 300 IU of urinary FSH and a daily antagonist, while in the control group the starting dose was between 150 and 225 IU. It is interesting to note that rates of maturity, fertilization and embryo development were identical in the two groups.
The model of oocyte donation is still a reference in IVF for differentiating between impact on the endometrium and impact on oocytes in ovarian stimulation and for overcoming the problem of poor quality oocytes. It is therefore interesting to see that Martínez et al. (2014) report a pilot study on luteal stimulation in this model. They compared the clinical pregnancy rate per transfer in recipients of embryos obtained after ovarian stimulation initiated either on Day 2 or Day 15 of the cycle, and showed that there is no difference in pregnancy rates (62.5 vs 58.3% respectively) or implantation rates (41.7 vs 45.0%). The study included nine donors who were given the two types of stimulation with a protocol combining rFSH and antagonist (started in a flexible manner with follicular stimulation and from the first day for luteal phase stimulation), with the trigger being a GnRH agonist and the oocytes being vitrified. The data for follicular and luteal phase stimulation did not show any difference regarding the number of mature oocytes, (14.0 ± 7.0 vs 16.9 ± 7.5) or the fertilization rate (77.3 vs 76.5%). This data is reassuring and allows us to be optimistic regarding urgent fertility preservation in the context of oncology. For the oocyte donors (Martínez et al., 2014), the number of days for FSH stimulation was not increased in the follicular vs luteal phase (10.4 vs 9.9), and nor was the total dose of FSH (2261 IU vs 2147 IU), but the number of days for antagonist was doubled depending on the protocol applied (5.1 vs 9.9). However, we will see with the Chinese studies that use of a GnRH agonist is not necessary in the luteal phase.
One reassuring argument for the competence of oocytes obtained in the luteal phase, compared with those obtained by standard stimulation, is the absence of an increased risk of aneuploidy reported by Ubaldi et al. (2016) in 43 patients with a genetic diagnosis of aneuploidy by trophectoderm biopsy. The rate of euploid blastocysts biopsied was 46.9% in the follicular phase compared to 44.8% in the luteal phase.
The largest study by far of luteal stimulation in normal-responding patients is a Chinese study (Wang et al., 2016a). As with most studies on the subject, it is a retrospective study. It should be noted that, in general, Asian clinics have used vitrification for longer and practise freeze all almost routinely. This is partly due to the use of clomiphene citrate, to reduce the cost of attempts with gonadotropins, and this has a well-known anti-oestrogen impact on the endometrium. In their study among normal responders, Wang et al. compared three types of protocol: luteal stimulation with 225 IU of HMG combined with 5 days of clomiphene (or letrozole) vs mild stimulation with clomiphene or letrozole for 5 days followed by 225 IU of HMG and the introduction of a flexible antagonist vs the short agonist protocol. The first two protocols used an agonist trigger and for luteal stimulation, and progestin was added at the end of stimulation in the first protocol to avoid the arrival of menses and retrieval under these conditions. Embryo vitrification was automatically used in the three groups, including in the short agonist protocol. Under these conditions, it is difficult to compare the duration of stimulation and total dose, as according to the protocol both are higher with luteal stimulation but are accompanied by a higher number of mature oocytes and top quality embryos. However, the study shows on a large scale that the use of an agonist or antagonist is not necessary in the luteal phase and that endogenous progesterone alone is sufficient to block the LH surge. In this retrospective study with a large number of patients with a good prognosis, Wang shows that implantation rates are identical with luteal stimulation (35.5%, N = 727) and mild stimulation (34.8%, N = 830), but significantly lower with the standard short agonist protocol (31.8%, N = 1385, P = 0.012). Birth rates were also significantly lower with the short agonist protocol and the levels of miscarriage were identical in the three groups. The same group reports reassuring but partial data (the actual number is lower than the study cited as a source) on the future of pregnancies and the health of children born following luteal phase stimulation compared with the two follicular phase protocols (Chen et al., 2015); no statistical differences were observed for gestational age, birth weight, neonatal anomalies or number of malformations.
Although an economic study remains to be conducted, it is not evident that luteal stimulation provides a benefit to a population with a good prognosis. The same team has more recently shown a possible advantage for women who are poor responders according to the Bologna criteria (Li et al., 2016). It was another retrospective study, using the mild protocol combining clomiphene then HMG at 75 or 150 IU/d (N = 98) either beginning at Day 3 with a flexible antagonist or beginning at Day 1 or Day 2 post-ovulation (N = 33). In this study, the trigger was 10 000 IU of HCG with fresh transfer for follicular stimulation and delayed transfer after embryo vitrification for luteal stimulation. The authors showed that the duration for luteal stimulation was longer than for follicular stimulation (10.3 vs 7.8 days) but this difference was not significant in contrast to the difference concerning the total dose of HMG (1085 vs 703 IU, P < 0.05). Furthermore the luteal stimulation was associated with a significantly higher number of oocytes collected and top quality embryos obtained, 2.8 vs 2.0 (P < 0.05) and 0.9 vs 0.4 (P < 0.05), respectively. The cycle cancellation rate was also lower in the luteal phase: 12.1 vs 30.6% (P < 0.05). As with the previous studies, the implantation rates were similar regardless of the origin of the embryos, but the difference is in the more frequent possibility of embryo transfer with luteal stimulation in poor responders.
Still with this same group of patients and to increase the chances of transfer and pregnancy, the authors experimented with the sequence of two stimulation protocols in double stimulation or the so-called Shanghai protocol (Kuang et al., 2014). After the first stimulation with the Asian mild protocol already described with the agonist trigger, stimulation with HMG is given from the evening of oocyte retrieval or the following day if at least two antral follicles are present. The patients included (N = 38) met the Bologna criteria and 30 of them had two retrievals in the same cycle. Kuang et al. showed that the number of oocytes collected in this second retrieval is higher than with the first retrieval (3.5 vs 1.7, P = 0.001) and that the levels of maturity and fertilization are identical. The number of embryos vitrified at Day 3 was higher with the second retrieval but not significantly different (1.3 vs 0.9). In summary, this double stimulation protocol enabled 21 patients out of 38 (55%) to have an embryo transfer from the first oocyte retrieval, with an ongoing pregnancy rate of 47.8% per transfer, which is 29% per patient in a population with a poor prognosis.
Ubaldi et al. (2016) reported the same benefit with ‘duostim’ and show, using genetic analysis, an increase in chances among patients with a poor prognosis of having a transfer of an euploid blastocyst, in the same menstrual cycle: 41.9% in the follicular phase, 53.5% in the luteal phase and overall 69.8% of patients with at least one euploid blastocyst in the two phases, in the same menstrual cycle.
This protocol uses the physiological principles of waves of folliculogenesis, and raises the question of the impact of FSH of the first stimulation on the small follicles of the following follicular waves. The action of exogenous FSH given in the follicular stimulation could have a positive effect on following waves, aspecially the one responding in the luteal phase stimulation. Small antral follicles may be sensitive to FSH, earlier than expected in the folliculogenesis process. It has been suggested that the follicle is sensitive to FSH at a very early stage and that the repetition of cycles leads to an improvement in response (Barad et al., 2014).
Progestin primed ovarian stimulation
If endogenous progesterone in the luteal phase is sufficient to block the LH surge and does not compromise oocyte competence, in cycles followed by oocyte or embryo cryopreservation, could exogenous progesterone replace the use of an agonist or antagonist in the follicular phase, with the advantage of oral administration and potential cost reduction?
Kuang et al. (2015) report the first randomized study with PPOS: they added medroxyprogesterone acetate (MPA) to gonadotrophins in a follicular phase stimulation. They used MPA because it is progestative, slightly androgenic, and does not interfere with the dosage of endogenous progesterone. MPA was started on Day 3 after spontaneous menses at the same time as HMG at 150 or 225 IU, and this protocol was compared with a conventional short protocol. Embryo vitrification was automatically used in both groups and embryos were transferred in a subsequent cycle. The calculation of the number of patients was based on the primary objective of non-inferiority of the number of oocytes collected (N = 150/group). Randomization was based on the odd and even numbers of the patients. The authors showed that the duration of stimulation was significantly longer by one day and the total dose of HMG was higher (~+400 IU) with MPA, but the number of mature oocytes was not significantly different in a population of normal responders (9.9 vs 9.0). The number of frozen embryos with MPA was higher but not significantly different (4.3 vs 3.7, P = 0.09) and the rates for implantation, pregnancy and miscarriage were not significantly different. The authors also noted that for the children born, birth weight was the same in the two groups and no congenital malformations were reported. MPA is teratogenic in animals but in this protocol with delayed transfer, there is no embryo exposure because the therapeutic window is around four weeks and the life duration of MPA is short. In contrast, there was a query over the use of an agonist trigger with MPA. The authors routinely did an LH assay at around 12 h from triggering to check the response to the agonist, with a fixed threshold of LH > 20 IU/L. Out of the first 50 patients, three were judged to have a non-optimal LH response, and for the subsequent patients 1000 IU of HCG was added to the agonist, while in the short agonist protocol group the triggering was conducted with 2000 or 5000 IU of HCG. There was a failure with one patient in the MPA group who had an LH surge with an increase in endogenous progesterone. In her case, stimulation was nevertheless continued and oocyte retrieval enabled collection of a number of oocytes that was similar to the group average. The hormone profiles (FSH, LH, oestradiol and progesterone) were superimposable in the two groups. It can also be concluded that the progestin in the follicular phase was effective at blocking the LH surge during stimulation. The authors stated, on personal data, that to be effective, the progestin has to be started before oestradiol rises (<50–70 pg/ml) but no reference was provided. The dose of MPA used was 10 mg, based on studies on contraception, as the the anti-gonadotropic effect of a 5 mg dose was insufficient.
The same MPA protocol was compared to the natural cycle in poor responders (FSH between 10 and 30 IU/L or antral follicle count (AFC) < 5) by the same team in a prospective cohorts (non-randomized) study, the results of which were presented by Kuang (ESHRE Lisbon 2015, not published). The average age of the patients was over 37 with an AFC of around 2.5. LH was completely controlled in the MPA group while 50% of patients had an LH surge in the natural cycle, and 11% ovulated before retrieval compared with 2% in the MPA group (P = 0.01). The number of mature oocytes and frozen embryos was significantly higher in the MPA group, 0.9 vs 0.7 (P = 0.001) and 0.5 vs 0.3 (P = 0.009) respectively, which allowed a transfer in 50% of cycles in the MPA group compared with 38% in the natural cycle group. The authors also reported an advantage in the MPA group, as a result of the LH being controlled, which is a reduction in restrictions regarding the organization of oocyte retrieval.
This same protocol has been applied among patients with polycystic ovary syndrome (PCOS) in a pilot study of 60 patients per group comparing the MPA protocol to the short agonist protocol (Wang et al., 2016b). The authors showed no difference in the data for stimulation and the results thereof, in particular in the number of oocytes collected and the ongoing pregnancy rates, however, there was a higher total dose of HMG in the MPA group. Regarding the specific risk of OHSS, two cases were reported in the short protocol group vs none in the MPA group (NS, P = 0.154). It is important to note that the short protocol is not the protocol currently recommended in this at risk population.
Alternatively, MPA could be replaced with micronised progesterone, as shown in a study, conducted under the same conditions as with MPA, which demonstrated the absence of an LH surge (Zhu et al., 2015). This was a retrospective study with 187 patients in each group comparing Utrogestan taken orally (100 mg twice a day) with HMG from Day 3 and a short agonist protocol. In the Utrogestan group, half of the patients were triggered with HCG and the other half were triggered with an agonist, with the same results. The total dose of HMG was significantly higher and the duration of stimulation significantly longer with Utrogestan, as with MPA and luteal phase stimulation studies. Despite the higher amount of HMG, the numbers of mature oocytes were not significantly different in these groups of normal responders. In contrast, the number of viable embryos was significantly higher in the Utrogestan group compared with the short protocol (5.0 vs 4.5, P = 0.029), however, there was no significant difference in the ongoing pregnancy rate.
‘Random start’ ovarian stimulation
Finally, whether it be luteal stimulation or stimulation with progesterone or a progestin, we are moving towards the idea that thanks to freeze all strategies, it is no longer necessary to use the menstrual cycle as a reference point for the collection of competent oocytes. This is where the idea of a ‘random start’ comes from, meaning starting stimulation on any day during the menstrual cycle.
This principle has been applied among normally responding patients at the Shanghai Ninth People's Hospital (Qin et al., 2016) which manages about 10 000 cycles per year. For reasons of practical organization (due to the number of women with significant travel constraints) stimulation is started when they arrive, whatever day it is in their cycle. For each group, 50 patients were reported and the inclusion criteria were quite broad: age < 42, AFC > 3 and FSH < 12. The first group was the ‘conventional’ group with stimulation beginning in the early follicular phase with a combination of HMG, clomiphene and MPA. The second group was the ‘late follicular’ group where an agonist injection was given to prevent the possibility of a dominant follicle, with the introduction of clomiphene and MPA on the same day, and of HMG on the following day. Finally in the third ‘luteal’ group, stimulation was started with clomiphene and HMG, without MPA, after checking the level of endogenous progesterone. Qin et al. showed in this retrospective study that the duration of stimulation was significantly shorter by around two days in the conventional group compared with the late follicular and luteal groups, but that the number of oocytes collected and the number of metaphase II oocytes as well as the average number of frozen embryos were identical in the three groups. However, the cancellation rate was lower (although not significantly, P = 0.388) in the conventional group (10 vs 22 and 16%). The reported ongoing pregnancy rates per transfer in these three groups were not statistically different, but unfortunately they did not include all of the patients who benefited from embryo cryopreservation. If there is an advantage to be seen in being able to begin stimulation at any point in the cycle, we should not forget that the freeze all strategy is also restrictive as the patient to has to return in the future for her embryo transfer, around 4 weeks later, to allow for adequate endometrial preparation time.
Practical applications
Figure 3 shows a schematic presentation and the results of all of the protocols using progesterone to block the LH surge compared with conventional protocols. Can these new approaches find a place in clinical practice? Regarding fertility preservation in the urgent context of oncology, the data are very reassuring althoughmore perspectives are required for its evaluation (Decanter and Robin, 2013). For subfertile patients undergoing IVF, the perspectives are discussed below.
Normal responders
In any of the situations where clinics wish to facilitate logistical aspects for the patients, luteal stimulation or the use of a progestin or progesterone in the follicular phase may be of great interest. These protocols on the one hand eliminate restrictions imposed by the dates of the menstrual cycle and on the other hand limit the ovarian stimulation agents togonadotrophin and trigger injections. It is easy to imagine a role for these new protocols in oocyte donors, with for example the use of long-acting FSH. Nevertheless it remains to be investigated whether the total dose and the duration of stimulation are increased, as seems to have been shown by the Asian retrospective studies. Situations of normal response with the absence of a fresh transfer are the preferred target, with the number and competence of the oocytes being equivalent to conventional protocols. At the forefront of these situations are oocyte donation and fertility preservation. The significance of the protocols using progesterone in the general population with a good prognosis is not clear and requires medical and economic analyses, especially under the hypothesis of a necessary total increase in gonadotrophin dose and the obligation to offer fresh transfer. The systematic use of the freeze all strategy is also coming into question (Blockeel et al., 2016). However evidence is growing on the improvement in chances of success in a more receptive endometrium, far removed from negative hormonal impact of ovarian stimulation. As reported for example by Almeida Ferreira Braga et al. (2016) in an oocyte sharing programme, implantation rates were 37.2% with fresh transfer compared to 67.3% with transfer of frozen embryos. The protocols using progesterone to block the LH surge would clearly compete with the conventional protocols in normally responding patients and those at risk of ovarian hyperstimulation.
Poor responders
Could these protocols bring a better response in the very heterogeneous group of poor responders undergoing IVF? The data for comparison with conventional protocols is not able to provide conclusions because it is based on retrospective studies with few patients. We can say that the use of a progestin in the follicular phase has allowed in this population very good hormonal control of the LH surge (Kuang et al., 2015). What seems to be making progress in this population is ‘duostim’ described by the Italian team of Ubaldi et al. (2016) and the ‘Shanghai protocol’ described by Kuang et al. (2014), which shows that carrying out a second stimulation immediately following the initial retrieval (within three days) significantly increases the chances of embryo transfer in a single menstrual cycle in this population with a poor prognosis. It is also possible that the number of oocytes retrieved following luteal stimulation may be slightly higher than following follicular stimulation, but this remains to be demonstrated. Physiologically speaking, it could be connected to the action of FSH which improves its own sensitivity by the stimulation of its own receptors and those of LH on the granulosa cells (Monniaux et al., 1993, 2009; Messinis et al., 2014). However, there needs to be an evaluation as to whether ‘banking’ of oocytes and embryos has a real advantage compared to immediate transfer, because as Cobo et al. (2012) suggest in their study involving a sequence of three stimulations in three different cycles, it is the drop out rate, i.e. the anticipated number of patients giving up the treatment, which is the main factor in the loss of potential pregnancies in the group without banking.
Conclusion
SWOT analysis of protocols using progesterone to block the LH surge. FET: frozen embryo transfer, RCT: randomized controlled study, OHSS: ovarian hyperstimulation syndrome, PCOS: polycystic ovary syndrome.
However, more research is needed with randomized controlled trials and medico-economic studies, with a larger number of patients, to confirm the absolute control of LH surge with progestins or endogenous progesterone and to confirm the outcomes. Finally, the progesterone block protocols also highlights the concept of follicular waves and the interest in synchronizing stimulation with the emergence of the wave to maximize the number of oocytes retrieved needs further investigation.
Acknowledgements
I am extremely grateful to Professor Clément Jimenez for his contribution to the writing of this review, in particular for his expertise in vitrification, as well as his careful proofreading.
Author's role
Nathalie Massin: medline search, analysis and article writing.
Funding
None.
Conflict of interest
None.
