| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Reproductive Medicine and Gynecology (N.S.M., B.C.J.M.F.), University Medical Center Utrecht, 3508 GA Utrecht, The Netherlands; Division of Reproductive Sciences (R.L.S.), Oregon National Primate Research Center, Oregon Health & Science University, Portland, Oregon 97239-3098; and Department of Obstetrics and Gynecology (L.C.G.), Stanford University School of Medicine, Stanford, California 94305
Correspondence: Address all correspondence and requests for reprints to: N. S. Macklon, Department of Reproductive Medicine and Gynecology, University Medical Center Utrecht, P.O. Box 85500, 3508 GA Utrecht, The Netherlands. E-mail: n.s.macklon{at}umcutrecht.nl
 |
Abstract |
|---|
|
I. Introduction |
|---|
). This allows for inefficiencies in subsequent oocyte maturation, fertilization in vitro, embryo culture, embryo selection for transfer, and implantation (5). Multiple embryos can be transferred in the great majority of patients, and spare embryos may be cryopreserved to allow for subsequent chances of pregnancy without the need for repeated ovarian stimulation and oocyte retrieval (6, 7).
|
|
II. Physiology of Ovarian Function Relevant to Ovarian Stimulation |
|---|
from theca cells, growth differentiation factor 9, and bone morphogenetic protein 15 produced by the oocyte may limit the effects of FSH on granulosa cell differentiation and follicle development at this early stage (14). Only at more advanced stages of development do follicles become responsive to FSH and obtain the capacity to convert the theca-cell derived substrate androstenedione to estradiol (E2) by the induction of the aromatase enzyme activity (4, 15).
Due to the demise of the corpus luteum during the late luteal phase of the menstrual cycle, E2, inhibin A, and progesterone (P) levels fall. This results in an increased frequency of pulsatile GnRH secretion, inducing rising FSH levels at the end of the luteal phase (16, 17). Although each growing follicle may initially have an equal potential to reach full maturation, only those antral follicles that happen to be at a more advanced stage of maturation during this intercycle rise in FSH (levels surpassing the so-called threshold for ovarian stimulation) gain gonadotropin dependence and continue to grow (4) (Fig. 1
). This process is referred to as cyclic, gonadotropin-dependent or "secondary" recruitment, as opposed to the initial gonadotropin-independent "primary" recruitment of primordial follicles (9). Based on indirect observations, it is believed that the cohort size of healthy early antral follicles recruited during the luteo-follicular transition is around 10 per ovary (8, 18, 19)
In the subsequent follicular phase, FSH levels plateau during the initial days (20, 21) and are gradually suppressed thereafter by ovarian inhibin B (22) and E2 (23) negative feedback. Gonadotropin withdrawal studies have demonstrated the association between FSH, LH, and inhibin production (16, 24, 25). Administration of recombinant inhibin A during the early follicular phase to nonhuman primates results in a gradual decrease in FSH (26). On the contrary, administration of recombinant inhibin A during the luteal phase prevents the subsequent rise of FSH during the luteo-follicular transition (26). These experiments suggest a direct endocrine role for inhibin A in the negative feedback on pituitary FSH production, whereas inhibin B does not contribute to the dynamic changes within a menstrual cycle (26, 27, 28, 29).
Decremental follicular phase FSH levels (effectively restricting the time where FSH levels remain above the threshold, referred to as the FSH window) (Fig. 1) appear to be crucial for selection of a single dominant follicle from the recruited cohort (20). As FSH levels fall, all but the dominant follicle (with its increased sensitivity to FSH) lose the stimulus to further development and become atretic (4, 30). The important concept of increased sensitivity of the dominant follicle for FSH has been confirmed by human studies showing developing follicles to exhibit variable tolerance to GnRH antagonist-induced gonadotropin withdrawal (31, 32). Recent evidence also points to a central role for LH in monofollicular selection and dominance in the normal ovulatory cycle (33, 34). Although granulosa cells from early antral follicles respond only to FSH, those from mature follicles (exhibiting receptors to both gonadotropins) are responsive to both FSH and LH. The maturing dominant follicle may become less dependent on FSH because of the ability to respond to LH (33, 34, 35).
Cyclic variation in the expressed isoforms of FSH (differing in oligosaccharide structure, the degree of terminal sialylation and sulfation) has been described (36). A greater proportion of less acidic circulating FSH isoforms are observed during the late follicular phase and midcycle (37, 38, 39). The half-life of human FSH secreted 2–3 d before ovulation is considerably shorter than during the early follicular phase (40). It has been suggested that the preferential secretion of short-lived isoforms during the periovulatory period indicates the existence of regulatory mechanisms that control the intensity and duration of the FSH signal delivered to the ovary (36).
B. Intraovarian modulators of steroidogenesis
Gonadotropins are the primary regulators of follicular development, cytodifferentation, and sex-steroid production in the ovary. However, a large number of intraovarian regulators modulate the response to gonadotropin stimulation. The principal regulatory systems in the human involve the IGF system (41), the epidermal growth factor (EGF) system (42), and the TGF- and -ß systems (43). In addition to their primary endocrine and paracrine functions of suppressing FSH secretion by the pituitary (44), inhibins and activins also exhibit local actions in the ovary (45).
Although IGFs and their receptors are known to be present in developing human follicles, uncertainty remains regarding the individual role of the different IGFs, their receptors, and binding proteins in vivo. In vitro studies have identified the effects of IGFs on granulosa and theca cell function. IGF-I has been shown to stimulate proliferation and aromatase activity of granulosa cells both alone or synergistically with FSH (46, 47, 48, 49). At the theca cell, IGF-I stimulates production of 17 -hydroxyprogesterone (50), whereas both IGF-I and -II can alone or together with LH stimulate androgen synthesis (51, 52). For review, see Poretsky et al. (41).
The activity of the IGFs is modulated by their degree of binding to IGF binding proteins. GH is the primary regulator of serum IGF-levels (41, 53). However, GH does not affect IGF-I and IGF-II expression in granulosa cells in vitro. It has therefore been proposed that GH may indirectly modulate the follicle by stimulating hepatic production of IGF-I.
EGF and TGF- are structurally similar polypeptides that bind to a common receptor, and both have been detected in human follicles (54). They would appear to stimulate granulosa cell proliferation (42) but inhibit FSH-induced aromatase expression and E2 synthesis (55). TGF-ß differs in structure and function from EGF and TGF-, being a homodimeric polypeptide with no clear direct inhibitory function on granulosa cell aromatase activity. Both EGF and TGF-ß synergize with FSH to stimulate granulosa cell proliferation in hamster preantral follicles (56).
In addition to regulating pituitary FSH, inhibin and activin also act as paracrine and autocrine modulators of ovarian follicle growth and maturation (57).
Activin acts on small follicles to stimulate proliferation of granulosa cells (58, 59), up-regulates FSH receptor expression in granulosa cells, and increases aromatase expression, resulting in increased E2 production (60). Inhibin augments LH-stimulated androgen production by thecal cell cultures (61).
C. Control of corpus luteum function
As reviewed recently (62, 63), it is generally believed that the predominant hormonal regulators of the corpus luteum in women and many nonhuman primates are LH-like gonadotropins. Unlike in some species, notably rodents (64), the luteotropic process in humans does not include a principal role for prolactin-like hormones, and luteal regression does not involve a uterine signal (prostaglandin F2) (63). Instead, it is: 1) the midcycle surge of gonadotropins (notably, LH) secreted by the anterior pituitary that stimulates resumption of meiosis and oocyte maturation in the preovulatory follicle, rupture of the ovulatory follicle and release of the expanded cumulus-oocyte complex, and conversion of the follicle wall into the corpus luteum (i.e. luteinization); 2) the pulsatile secretion of pituitary LH during the luteal phase of the menstrual cycle that promotes the continued development and normal functional lifespan of the corpus luteum; 3) the exponential rise in circulating levels of the LH-like hormone, chorionic gonadotropin (CG), secreted by the implanting blastocyst and syncytiotrophoblast of the developing placenta, that extends the functional lifespan of the corpus luteum in early pregnancy until luteal activities are assumed by the placenta, i.e., at the luteal-placental shift.
Elegant studies during the 1970s and 1980s (for reviews, see Refs. 65 and 66) using techniques such as gonadotropin ablation and GnRH infusion to control LH secretion established the critical role of LH/CG in regulating primate luteal structure-function. More recent experiments using GnRH antagonists and pure recombinant human LH or human CG (hCG) have strengthened this concept (67, 68, 69, 70) and clarified a number of issues.
First, although the maturing dominant follicle may be less sensitive to acute LH withdrawal at midcycle, a GnRH-induced LH surge of substantial length is required for ovulation and development of normal luteal function. Of considerable interest is how the duration and/or amplitude of the midcycle LH surge influences periovulatory events. Although more research is needed, initial monkey and human studies on GnRH-induced LH surges or administering exogenous LH/CG suggest that surges of lesser duration (<24 h) and amplitude are sufficient to reinitiate oocyte meiosis and granulosa cell luteinization, but surges of greater duration (>24 h) and amplitude improve oocyte recovery, fertilization, and corpus luteum development (71, 72). Although the LH surge is believed to be the physiological signal for periovulatory events, studies in rodents (73) showed that a midcycle bolus of FSH can replace LH and elicit oocyte maturation, ovulation, and successful pregnancy. Likewise, an FSH bolus will induce certain periovulatory events in macaque follicles (74) after ovarian stimulation, including oocyte meiotic maturation, fertilization, and early luteinization of granulosa cells.
Secondly, although luteinizing tissue or cells appear less responsive to exogenous gonadotropin around the time of ovulation (presumably due to LH/CG receptor desensitization, down-regulation, or occupancy by gonadotropins from the surge interval), the developing corpus luteum in the early luteal phase (70) is comparable to the developed corpus luteum by midluteal phase (32) in its critical dependence on circulating LH for continued function. Several reports confirm that suppression of LH support for 72 h results in irreversible loss of luteal structure-function (69, 70), but LH or CG (not FSH) replacement sustains luteal structure-function (75, 76). Attempts to titrate the amount of LH required to maintain the normal functional lifespan of the corpus luteum in GnRH antagonist-treated monkeys showed that increasing the dose from mid-to-late luteal phase was critical (77, 78). Such results support the concept that the primate corpus luteum becomes less sensitive to LH as the luteal lifespan progresses. Although the frequency of LH pulses declines during the luteal phase (79), prevention of this phenomenon (via pulsatile GnRH infusion or LH injections (80) does not prevent timely luteal regression. Rather, decreasing luteal sensitivity to gonadotropin could be a critical factor in timely luteolysis near the end of the menstrual cycle (66).
Collectively, the data suggest that rescue of the corpus luteum from impending regression and continuation of its functional lifespan in early pregnancy are not likely due to inherent differences in LH vs. CG bioactivity or to a change from pulsatile (LH) to continuous (CG) gonadotropin exposure in the fertile cycle. Rather, it appears that a more robust luteotropic stimulus, in the form of rising levels of LH/CG, is required to extend the functional lifespan of the primate corpus luteum.
Local modulating factors may include the steroid hormones produced by the corpus luteum. There is considerable evidence that another action of the midcycle LH surge, which complements the promotion of P production, is the induction of P receptors (PRs) in luteinizing granulosa cells of the follicle (for review, see Ref. 62). The hypothesis that estrogen acts locally as a luteolytic signal (81) has renewed credence with the discovery of estrogen receptor (ER)-ß in the primate corpus luteum (82). Although there are reports of androgen receptors in primate luteal tissue (83, 84), there has been little consideration of local androgen action in the corpus luteum.
LH/CG also regulates the expression of angiogenic and angiolytic factors that likely control the expanding vasculature in the ovulatory follicle and developing corpus luteum. The LH-stimulated luteinization of granulosa cells around ovulation includes enhanced vascular endothelial growth factor (VEGF) production (85), which is likely essential for the angiogenic process within the primate corpus luteum (86, 87). With respect to hCG, a midcycle bolus in ovarian stimulation cycles increased expression of the endogenous angiopoietin agonist, Ang-1, without altering that of the endogenous antagonist, Ang-2 (88), in macaque granulosa cells. It is important to recognize that these factors control not only the development or maintenance of the vasculature in developing tissue beds, but also vascular integrity, maturity, and permeability. It has been proposed (89, 90) that overexpression, increased bioavailability, or a change in the ratio of angiogenic factors, notably VEGF-A, is a cause of ovarian hyperstimulation syndrome (OHSS) (91), a serious side effect of ovarian stimulation characterized by intravascular volume loss and extravascular fluid accumulation. The early or late occurrence of OHSS in ovarian stimulation cycles has been linked to the ovulatory hCG bolus and endogenous CG production at pregnancy recognition, respectively.
D. Control of endometrial function
1. Steroid hormone receptors and actions in endometrium.
Ovarian-derived steroid hormones have profound effects on the endometrium that result in proliferation and differentiation of the tissue, receptivity to embryonic implantation, and shedding in the absence of a pregnancy. During the follicular phase, E2 secreted by growing follicles stimulates ER expression, with highest levels observed in glandular epithelium during the late follicular phase (92, 93, 94, 95, 96, 97). Two forms of ER are now appreciated: ER- and ER-ß, which are two distinct gene products (98). They are expressed in both glands and stroma (with ER- predominating), whereas ER-ß is only expressed in endothelium (99). ER()is significantly down-regulated in epithelium in the luteal phase, a universal response in all mammalian species (100).
With regard to PR, peak expression in human endometrium induced by E2 is observed at the time of ovulation (92, 97, 101, 102). PR is most prominent in glandular epithelium in the proliferative phase and is undetectable in the midluteal phase in this cell type (92). In contrast, stromal cells have high levels of PR in the follicular phase and throughout the luteal phase. Similar observations have been made in nonhuman primate endometrium (93, 95, 96, 103). The human PR has two functionally distinct isoforms, PR-A and PR-B, encoded by a single gene (104). In endometrial glands, PR-A and PR-B are expressed in the follicular phase, but only PR-B persists during the mid- and late luteal phase in this cell type (102, 105, 106). In endometrial stroma, PR-A predominates throughout the cycle, suggesting that it is important in P action in the luteal phase (102, 105, 106). Overall, these results support the view that PR-A and PR-B mediate distinct pathways of P action in the glandular epithelium and stroma throughout the menstrual cycle. It should be noted that the timely down-regulation of epithelial PR coincides with the opening of the window of implantation and uterine receptivity for embryonic implantation (see Section II.D.3), and histological delay of the endometrium (a clinically abnormal state) is associated with a failure of such PR down-regulation (107).
The major roles of E2 are for endometrial growth and for enabling P to act on the tissue. To accomplish these goals, E2 induces PR expression and promotes cellular proliferation in the tissue—directly through its cognate receptors, and indirectly by induction of growth factors that act as autocrine and/or paracrine modulators (108, 109, 110).
2. Endometrial morphological changes in response to ovarian-derived steroid hormones.
The endometrium demonstrates day-by-day morphological changes, extensively described by Noyes et al. (111) who analyzed several thousand endometrial biopsies to develop the criteria for assessment that are still considered the gold standard. The initial Noyes’ criteria correlated the results of the biopsy with the basal body temperature and with the subsequent menstrual period. Extensive ultrastructural changes also occur during the cycle that underscore the magnitude of effects of ovarian-derived steroid hormones on this tissue (112).
Morphological features correlating with endometrial maturity have been identified by scanning electron microscopy. Pedunculated extrusions of the luminal epithelial cell membrane, termed pinopodes, can be identified near the lateral cell border, rising above the plane of the normal microvilli, but not occupying the entire cell surface (113). They are P-dependent and inhibited by E2 (114, 115, 116). These structures last for 1–2 d, and their numbers positively correlate with implantation sites (117). Although they likely play a role in the early stages of implantation, their precise functions remain to be clarified.
3. The window of implantation and the effects of ovarian-derived steroid hormones.
Ovarian steroid hormone actions during normal ovulatory and hyperstimulation cycles result in a temporally and spatially restricted period ("window of implantation") in which the tissue is receptive to embryonic implantation (118). Available evidence supports the discrete time in the cycle between 6 and 10 d after the LH surge that defines the window of implantation. The window is advanced in clomiphene citrate (CC) or gonadotropin-stimulated cycles (114, 119) and is delayed in steroid hormone replacement cycles for donor recipients (120), underscoring its plasticity and the significant effects that steroid hormones have on this tissue. When embryo transfers were performed in IVF cycles between cycle d 17 and 19, 40.5% conceptions occurred, compared with no conceptions in cycles where embryo transfers occurred after cycle d 20 (121). These observations collectively support a distinct and narrow period of endometrial specialization that coincides with the window of implantation.
A major challenge is to define the molecular events occurring during the window of implantation that render the endometrium receptive to implantation and the interactions that occur between the maternal endometrium during pregnancy (i.e., the decidua) and the implanting conceptus. Immunohistochemical techniques have characterized the expression of receptors, adhesion molecules, and other markers of receptivity. The presence of ER and PR is most pronounced during ovulation (122). These nuclear receptors are especially induced by ovarian estrogens and are present in the glandular and stromal compartment (92).
The expression of cell adhesion molecules such as integrins is also under endocrine and paracrine control (92). Flow cytometry studies have shown E2 and P to decrease Vß3 integrin expression. Down-regulation of this integrin by E2 and P indicates that implantation and receptivity may arise as a result of a down-regulation of E2 and PRs during the midluteal phase (92, 123). E2 and P may therefore have a suppressive role on integrins and other critical endometrial proteins such as cytokines, which may only be expressed when this inhibitory signal is removed.
Leukemia inhibitory factor (LIF) is the first cytokine that appeared to be critically involved in embryonic development and implantation in mice. LIF is a secreted pleiotropic cytokine with a glycoprotein structure of 180 amino acids (33). High serum P levels coincide with the presence of LIF, and glandular LIF expression can be blocked by antiprogestins (124). The biological action of LIF in human endometrium is still unclear, but it probably has a function in human implantation at the stage of embryonic invasion. Coexpression of LIF and pinopodes has been found in fertile women (125).
Functional genomic approaches have been used to defined the molecular events occurring during the implantation window that contribute to the interactions between the endometrium and an implanting conceptus (126, 127, 128, 129). Moreover, new light has been shed on the impact of ovarian hyperstimulation on endometrial gene expression. Putative molecular players in the endometrium for uterine receptivity and their roles in the early stages of implantation have been reviewed in the mouse (130), the nonhuman primate (131), and the human (118, 131, 132, 133).
E. Ovarian aging
By a process of mitosis, the pool of germ cells undergoes rapid expansion, reaching a maximum of approximately 7 x 106 oogonia by the fifth month of intrauterine life (134). The oogonia then enter meiotic prophase, marking the completion of germ cell production. From here on, attrition in germ cell numbers occurs such that at birth each ovary contains between 25 x 104 and 50 x 104 resting follicles (135, 136). Depletion of these primordial follicles, already begun before birth, continues throughout childhood so that by the menarche a total of approximately 3 x 104 remain (137). During reproductive life, follicle depletion occurs at a rate of approximately 1000 per month by either atresia or entry into the growth phase, and this rate increases after the age of 35 yr (138) until the menopause when the stock of resting follicles falls to less than 1000 per ovary (135, 138).
The vast majority of follicles are removed from this stock by apoptosis (139). During fetal life and childhood, follicle development occurs up to the early antral stage (140). From puberty until the menopause, full maturation and ovulation occur, but only approximately 400 follicles are destined to achieve full maturation. As reproductive age advances to the menopause, the menstrual cycle decreases in length predominantly due to a shortening of the follicular phase (141). The shortened follicular phase in older ovulatory women is due to advanced follicle growth and earlier dominant follicle selection (142). Depletion of the ovarian follicular pool (138) leads to a diminished production of E2 and inhibin-B (143) and a gradual rise in FSH concentrations (144). Major individual variability exists in the rate of follicle pool depletion within the normal range of menopausal age of 40 to 60 yr (145) (Fig. 2). Hence, chronological age is only loosely associated with the extent of follicle depletion (ovarian age).
|
The identification of sensitive and specific markers of ovarian aging may enable prediction of individual response to ovarian stimulation and outcome of IVF. The most widely used endocrine marker for ovarian reserve is the early follicular phase FSH level (144). Although baseline FSH levels predict poor response to ovarian stimulation (148, 149), age appears to be more closely related to the chance of implantation and ongoing pregnancy (150, 151). Young women with high FSH levels demonstrate lower numbers of growing follicles but can achieve good ongoing pregnancy rates if oocytes and embryos are obtained (152, 153). Inconsistencies may arise from the wide interindividual variation that exists in follicular phase FSH concentrations in the normo-ovulatory cycle (21, 154) and from discrepancies between quantity (follicle number) and quality (competence) of oocytes.
Elevated E2 levels on cycle d 3 may also predict poor response to ovarian stimulation for ART, even when baseline FSH levels are normal (155). Because FSH levels are not always correlated with E2 concentrations, the rising FSH levels are partly attributed to lower inhibin levels produced by the aging ovary (143). A decrease in serum inhibin B precedes both the fall in inhibin A levels (156) and the perimenopausal rise in basal FSH (157). Falling inhibin B concentrations may also provide a more direct assessment of ovarian reserve, because it is directly produced by developing early antral follicles (22, 158). Studies concerning the value of assessing inhibin B have shown discordant results (22, 157, 159, 160, 161, 162). Recently, it was demonstrated in a multivariate analysis that addition of basal FSH and inhibin B levels to a logistic model including ultrasound characteristics can improve the performance of the prediction model for ovarian response to stimulation (163).
The age-related decrease in number of antral follicles present in the ovary at the start of the cycle is considered to correlate with the number of primordial follicles remaining in the ovary (164). It should be emphasized, however, that direct evidence to support this contention is lacking. The antral follicle number assessed by ultrasound during the early follicular phase has been shown to correlate with ovarian response to stimulation (163, 165), and to predict the number of immature oocytes retrieved from unstimulated ovaries before in vitro maturation (IVM) (166). Ovarian volume, which partly reflects the number of ovarian follicles, has also been shown to decrease with age (167), and a number of studies have suggested a role for this parameter as a marker of ovarian reserve (168, 169, 170).
Recently, anti-Mullerian hormone (AMH), also referred to as Mullerian-inhibiting substance, has also been studied as a marker of ovarian aging. A member of the TGF-ß superfamily, AMH is produced by ovarian granulosa cells from about 36 wk gestation to the menopause (171). Expression of AMH is highest in granulosa cells of growing preantral and small antral follicles. The role of AMH in follicle development and function has been elucidated in studies of AMH-deficient mice. In the absence of AMH, ovaries are more quickly depleted of primordial follicles due to increased recruitment (172). Additional studies have suggested that AMH may also influence the sensitivity of growing follicles to FSH (173). In vitro studies have shown AMH to reduce expression of aromatase mRNA, and decrease the number of LH receptors in granulosa cells (174). In vivo studies in AMH null mice have supported an inhibitory role for AMH in the cyclic recruitment of follicles by lowering sensitivity to FSH (175). Because AMH is produced by growing follicles, it has been proposed as a marker of ovarian reserve. Indeed, serum concentrations of AMH decrease over time in young normo-ovulatory women (176). AMH concentrations correlate well with the number of antral follicles and age, and less strongly with inhibin B and FSH levels (176, 177). In contrast to inhibin B and FSH, serum AMH levels are relatively constant throughout the menstrual cycle. Taken together with recent clinical studies showing high correlations between low AMH levels and ovarian response to stimulation (178, 179), AMH may represent an important marker of ovarian reserve.
|
III. The Development of Ovarian Stimulation Agents |
|---|
B. The discovery of clomiphene citrate
In the late 1950s, the first nonsteroidal estrogen antagonist (MER-25) was tested for the treatment of breast cancer and endometrial hyperplasia. The administration of CC in women with endometrial hyperplasia suffering from secondary amenorrhea was followed by the recommencement of menstrual cycles (185). Shortly thereafter, the ovulation-inducing capacity of a closely related antiestrogen (MRL/41) was recognized (186). CC was originally developed for clinical use by the Merrel company in 1956. Nearly 50 yr later, it is still considered to represent the first line treatment strategy in most anovulatory infertility and is still the most applied drug for infertility therapies worldwide.
CC is an oral antiestrogen consisting of a racemic mixture of two stereoisomeres. The enclomiphene isomere has a relatively short half-life, whereas the zuclomiphene isomere has an extended clearance and may accumulate over consecutive cycles. The two isomers demonstrate different patterns of agonistic and antagonistic activity in vitro (187, 188). Stimulation of ovarian function is elicited by raised pituitary FSH secretion due to blockage of E2 steroid feedback by CC. Overall, a 50–60% increase of serum FSH levels above baseline has been described (189, 190, 191). The exact nature of the mechanism of action of CC is still uncertain (189, 192), but induced changes in the IGF system may also be important (191). CC for ovulation induction in anovulatory women is considered to be relatively safe because steroid negative feedback remains intact. The oral route of administration and low costs represent additional advantages of this preparation.
In addition to its desirable central action of stimulating a transient increase in gonadotropin secretion, CC may have other potentially detrimental effects on peripheral reproductive functions. In vitro studies have revealed inhibition of human granulosa or luteal cell steroidogenesis (188). However, in the context of higher E2 levels as a result of dominant follicle growth, this is probably not of clinical importance. Antiestrogenic effects at the uterine level (cervical mucus production and endometrial receptivity) are believed to underlie the observed discrepancy between achieved ovulation and pregnancy rates (193, 194). CC does not appear to be associated with preterm birth and congenital abnormalities (195, 196). However, data from well-designed prospective studies are lacking. In vitro animal studies only reveal effects on oocytes or embryos when exposed to levels far higher than those attained in vivo. The putative increased risk of ovarian cancer reported to be associated with the use of CC for more than 12 months (197) has led CC to be licensed for just 6 months of use in some countries.
After the first IVF baby born in a natural cycle (198), four normal IVF pregnancies were subsequently reported after ovarian stimulation with CC (199). In the following years, many groups reported IVF results after CC, with or without gonadotropin cotreatment (200, 201).
C. Gonadotropins
Human menopausal gonadotropins first became widely used for IVF in the United States (202, 203). For over two decades, gonadotropin preparations have also been extensively applied for ovarian stimulation in ovulatory women for empirical treatment of unexplained subfertility. The aim is to increase the number of oocytes available for fertilization in vivo (for review, see Ref. 204).
The FSH to LH bioactivity ratio of registered hMG preparations is 1:1. As purity improved, it was necessary to add hCG to maintain this ratio of bioactivity (205). The initial preparations were very impure with many contaminating proteins; less than 5% of the proteins present were bioactive. Bioactivity of gonadotropin preparations continues to be assessed by the crude in vivo rat ovarian weight gain Steehlman and Pohley assay (206). This rather anachronistic technique has the disadvantage of allowing considerable batch to batch inconsistency in bioactivity. However, improved protein purification technology allowed for the production of hMG with reduced amounts of contaminating nonactive proteins and eventually the development of purified urinary FSH (uFSH) preparations by using monoclonal antibodies since the early 1980s (Fig. 3) (for review, see Ref. 184). The currently available pure products allow for less hypersensitive reactions and less painful sc administration. Due to the worldwide increased need for gonadotropin preparations, demands for postmenopausal urine increased tremendously, and adequate supplies could no longer be guaranteed. In addition, concern regarding the limited batch-to-batch consistency along with possibilities of urine contaminants emerged (207).
|
- and hormone-specific ß-subunit of the glycoprotein hormone into Chinese hamster ovary cell lines (208), the large scale in vitro production of human recombinant FSH (recFSH) has been realized (209, 210). The first pregnancies using this novel preparation in ovulation induction (211) and in IVF (212, 213) were reported in 1992. Since then, numerous, large-scale, multicenter studies have been undertaken (214) demonstrating their efficacy and safety. The recombinant products offer improved purity, consistency, and large-scale availability. Because of its purity, recFSH can now be administered by protein weight rather than bioactivity, and so-called "filled-by-mass" preparations (215) are now available for clinical use. During recent years, recombinant LH (recLH) (216, 217) and hCG (rechCG) (218, 219) have also been introduced for clinical application.
The first report on the design of a long-acting FSH agonist was by Boime and co-workers (220), who used site-directed mutagenesis and gene transfer techniques to manufacture FSH-carboxy-terminal peptide (CTP). This molecule is also produced by a Chinese hamster ovary cell line and contains four N-linked carbohydrate chains (52, 78, ß7, and ß24) and four O-linked carbohydrate chains at the CTP (ß115, ß121, ß126, and ß32). The latter group causes a 3- to 4-fold increase in half-life in vivo compared with wild-type recFSH (221). FSH-CTP has recently been subject to clinical studies as discussed in Section IV.B.
D. GnRH agonists
In 1971, the small decapeptide GnRH was isolated, and its structure was elucidated (222, 223). Amino acid substitutions have revealed the significance of specific regions for its stability, receptor binding, and activation of the pituitary gonadotroph cells (224). This decapeptide is secreted by the hypothalamus into the portal circulation in an intermittent fashion stimulating the pituitary gonadotropes to synthesize and secrete LH and FSH (for review, see Ref. 225). In addition to this long-established central role, recent studies suggested that GnRH also acts as a local autocrine and/or paracrine factor in the human ovary by regulating steroidogenesis (226), cell proliferation (227), and apoptosis (228). However, the current therapeutic applications of GnRH analogs are derived from their proven role in regulating gonadotropin secretion.
Clinically safe GnRH agonists were developed by replacing one or two amino acids. An increased potency could be achieved by replacing glycine for D-amino acids at position 6 and by replacing Gly-NH2 at position 10 by ethylamide (229). Such simple structural changes render these compounds more hydrophobic and more resistant to enzymatic degradation. In 1978, it was discovered that repeated administration of GnRH agonists produced a transient increase in gonadal function followed by a decrease in gonadal function and a significant fall in sex steroids (230, 231). Although initial binding to GnRH receptors results in activation, continuous occupation leads to desensitization due to the clustering and internalization of pituitary GnRH receptors, resulting in falling LH and FSH levels (232). If the agonists are administered for a period of several months, LH levels remain suppressed, but FSH levels return to normal and eventually rise to supraphysiological levels (233).
Pulsatile administration of GnRH was established as an effective and safe means of treating hypogonadotropic hypogonadal anovulation (231, 234). The first reports concerning its use for the prevention of a premature LH rise during ovarian stimulation appeared in the early 1980s (235, 236). During initial studies with hMG stimulation of multiple follicle development for IVF, it became apparent that a premature LH peak occurred in 20–25% of cycles due to positive feedback activity by high serum E2 levels during the midfollicular phase of the stimulation cycle (237). This advanced exposure to high LH was associated with premature luteinization of follicles and either cycle cancellation due to follicle maturation arrest or severely compromised IVF outcomes. The clinical development of GnRH agonists (for reviews, see Refs. 225 and 229) allowed for the complete suppression of pituitary gonadotropin release during ovarian stimulation protocols for IVF (235, 238, 239, 240). Induced pituitary down-regulation indeed resulted in significantly reduced cancellation rates and improved overall IVF outcome (241). Moreover, the approach of GnRH agonist cotreatment facilitated scheduling of IVF and timing of oocyte retrieval.
Recently, a second form of GnRH agonist (GnRH-II) has been identified. This differs from the mammalian GnRH-I by three amino acid residues. In addition to expression in the brain, GnRH-I and GnRH-II transcripts are expressed in various cells within the ovary (242). The physiological significance of GnRH-II in the human remains unclear.
E. GnRH antagonists
Immediate suppression and recovery of pituitary function rendered GnRH antagonists particularly appropriate for short-term use in IVF. However, it has taken almost three decades to develop such compounds with acceptable safety and pharmacokinetic characteristics. The first generation antagonists were developed by replacing "his" at position 2 and "trp" at position 3, but these compounds suffered from low potency. In second-generation compounds, the activity was increased by incorporating a D-amino acid at position 6. However, the widespread clinical application of these compounds was hampered by frequent anaphylactic responses due to histamine release. By introducing further replacements at position 10, third generation compounds were developed (225). Subsequently, both the compounds ganirelix (developed by Syntex Research, Palo Alto, CA) and Cetrotide (developed by Asta Medica, Frankfurt, Germany) were shown to be safe and efficacious in IVF. These third-generation GnRH antagonists were registered in 2001 for use in IVF (243).
The expression of GnRH and GnRH receptors in developing mouse embryos at the mRNA and protein levels raises issues of safety for the embryo. In one study, the incubation of mouse embryos with a GnRH agonist enhanced the preimplantation embryonic development in a dose-dependent way, whereas GnRH antagonist blocked this development (244). Moreover, GnRH mRNA and GnRH proteins are produced in the human fallopian tube during the luteal phase of the menstrual cycle (245). Studies are still required to demonstrate convincingly clinically relevant direct effects of GnRH analogs on fertilization, early embryonic development, and implantation in humans. Follow-up data on pregnancy, birth, and neonatal outcome of 227 children born after IVF or intracytoplasmic sperm injection cycles in which cetrorelix was used showed no abnormal results in comparison to outcome after commonly used long GnRH agonist protocols (246).
|
IV. Ovarian Stimulation Regimens |
|---|
CC usually induces the development of at least two follicles, which may sometimes elicit a premature LH rise. By virtue of the fact that CC is therapeutically active through interference with estrogen feedback (requiring an intact pituitary-ovarian axis), this compound cannot be combined with GnRH agonist cotreatment for prevention of a premature LH surge. Moreover, undesired antiestrogenic effects of CC at the level of the endometrium have been implicated by some in the observed discrepancy between relatively low embryo implantation rates coinciding with successful ovarian stimulation (204). CC administration is usually initiated on cycle d 2, 3, or 5, and given daily for 5 subsequent days, with doses varying between 100 and 150 mg/d. In most applied regimens, exogenous gonadotropin medication (150 IU/d) is initiated after cessation of CC. It seems that CC alone induces a limited but dose-dependent increase in the number of developing follicles. However, the addition of gonadotropins elicits increased ovarian response as manifest by more follicles. Sufficiently powered randomized comparative trials to support one approach over the other are lacking.
Reported outcomes with CC alone are variable, but in general pregnancy rates appear higher compared with natural cycle IVF, but lower compared with conventional gonadotropin/GnRH agonist protocols. Again, most studies are uncontrolled, but an extensive summary of almost 40,000 cycles reported in the literature suggests pregnancy rates of 6% per started cycle and up to 20% per embryo transfer (253). Apart from hot flushes, which may occur in up to 10% of women taking CC, side effects are rare. Nausea, vomiting, mild skin reactions, breast tenderness, dizziness, and reversible hair loss have been reported, but less than 2% of women are affected. The mydriatic action of CC may cause reversible blurred vision in a similar number of women. Overall side effects are CC dose related and are completely reversible once medication is stopped.
Tamoxifen, like CC, is a nonsteroidal selective ER modulator. Primarily developed for and used in the treatment of breast cancer, it has also been used in ovulation induction for many years. In contrast to CC, tamoxifen only contains the zu-isomer and appears to be less antiestrogenic at the uterine level. The possible advantages of tamoxifen over CC include beneficial effects on cervical mucus (254) and an agonistic effect at the endometrium. However, although endometrial thickness may increase on ultrasound monitoring, histological studies indicate that this may be due to edema and enlargement of stromal cells, rather than a purely estrogenic proliferative effect (for review, see Ref. 255). In recent years, tamoxifen has been proposed as an alternative means of ovarian stimulation for ART in women who have had breast cancer (256, 257), while protecting the breasts from concomitant high serum estrogen levels. Additional follow-up studies should be carried out before this drug is widely applied in these patients.
B. Gonadotropins
Gonadotropin preparations still constitute the principal agents for ovarian stimulation in IVF. The daily administration of these preparations is usually efficacious in the maintenance of growth of multiple antral follicles, allowing for the retrieval of many oocytes for IVF. Preparations initially used were hMG (containing both LH and FSH bioactivity), followed by purified uFSH and more recently recFSH and recLH.
Starting doses vary between 100 and 300 IU/d and are often adjusted depending on the observed individual ovarian response. However, there is little evidence to support dose adjustments midcycle (258). Several randomized clinical trials employing GnRH agonist cotreatment have failed to demonstrate improvements in outcome when higher doses of FSH are employed, even in older patients (259, 260, 261, 262, 263). A single-center study comparing 150–225 IU recFSH with GnRH antagonist comedication (264) showed similar results. The widely applied practice of increasing gonadotropin to ameliorate low response to stimulation is not supported by published evidence (265). This is not surprising when the pathophysiology of ovarian aging (i.e., follicle pool depletion) is taken into consideration.
When excessive follicle development raises the concern of imminent OHSS, gonadotropins are often reduced or temporarily withheld, a practice known as "coasting" (266). Studies of the efficacy of this approach have been inconclusive (267). Major individual differences in body weight may determine response (268), as may the cotreatment employed to prevent premature luteinization. Because endogenous gonadotropins are suppressed by GnRH antagonists for a limited period of time, less exogenous FSH is required. The ideal day of initiation of gonadotropin therapy is another variable that has been poorly characterized so far (243). The approach of starting exogenous FSH early during the luteal phase of the preceding cycle recognizes the physiological principle of early recruitment of a cohort of follicles for the next cycle (4). However, this protocol did not result in improved ovarian response in women with a low oocyte yield during previous IVF attempts (269).
To allow for the clinical introduction of recFSH, large scale, multicenter, comparative trials in IVF sponsored by pharmaceutical companies were published from 1995 onward (214). Several independent comparative trials have since been published, but sample sizes of these single-center studies were usually insufficient to allow for the detection of small differences. Meta-analyses (270) and health economics studies (271, 272) indicated a slightly improved outcome for recFSH compared with uFSH. A meta-analysis comparing recFSH vs. hMG suggested comparable outcomes (273). Subsequent multicenter trials also reported similar clinical outcomes comparing uFSH vs. recFSH (274) or hMG vs. recFSH (275). A meta-analysis comparing urinary-derived FSH with recFSH showed no significant difference in pregnancy rates (276). Finally, a meta-analysis comparing clinical pregnancy rates per started cycle after recFSH, uFSH, and hMG concluded that there is no evidence of clinical superiority for recFSH over different urinary-derived FSH gonadotropins (277). The data from the principle meta-analyses are summarized in Fig. 4. The continuing debate relating to the relative efficacy and effectiveness of different gonadotropin preparations in IVF is largely driven by commercial rather than scientific imperatives. When selecting a gonadotropin regimen, other factors should therefore be taken into account when selecting a gonadotropin regimen. In terms of tolerance, recFSH preparations showed some improvement over urinary-derived preparations, allowing for safe sc administration (278). The use of recFSH also reduces the theoretical risk of transmission of prion proteins, which have been identified in human urine (279). Although infections by urine prions in humans and animals have not been reported, the risk of prion disease such as new variant Creutzfeldt-Jakob disease has been deemed by some to be sufficient to advise against the use of uFSH, or urinary hCG (uhCG) (280). However, in a recent study of the 143 cases of Creutzfeld-Jakob to date registered in the United Kingdom, 63 were females and only one of these had undergone an infertility treatment from 1998–1999 (281). Although this may suggest low risk in association with infertility treatments, the long incubation period of this condition may continue to mask the real risk.
|
C. The role of LH
A number of recent studies indicated that excessively suppressed LH concentrations in the late follicular phase may be detrimental for clinical IVF outcome (286, 287). Under these circumstances, the use of urinary preparations containing both LH and FSH activity or the addition of recLH or rechCG next to exogenous FSH may be useful. However, several recent studies failed to confirm these findings and questioned the need for exogenous LH (288, 289, 290, 291, 292, 293, 294, 295). It remains unclear for which patients this approach may be beneficial. Supplementation of LH activity may offer advantages in some patients by hastening large follicle development and therefore shortening the duration of treatment (296). Moreover, the work of Zeleznik and co-workers (33) referred to a potential therapeutic role for LH in effecting monofollicular stimulation as part of a sequential ovarian stimulation protocol after initiation with recFSH. This concept was supported in a recent study in which anovulatory women with a hyperresponse to recFSH were randomized to continue treatment with the addition of either placebo or recLH (297). In those in whom LH was administered, a trend toward fewer preovulatory follicles was observed. However, in a parallel study, treatment with recLH alone in the late follicular phase was found to be detrimental to preovulatory follicle development (298, 299).
Recently, the concept that exogenous LH is capable of selectively stimulating the development of the more mature dominant follicles has been developed (300). A shift from FSH to LH preparations during stimulation may therefore be useful to stimulate a more homogeneous cohort of follicles for IVF (33, 34). However, opposing views have also been published suggesting no added value of LH supplementation (301). In accordance with the reported association between low LH levels (<0.5 IU/liter) and lower ongoing pregnancy rates in IVF cycles (286), LH levels were proposed to have a role in the lower pregnancy rates in GnRH antagonist cycles, because these cycles often lead to extensive suppression of endogenous LH activity during the late follicular phase if combined with rFSH administration. However, several recent studies reported conflicting results with regard to a possible association between serum LH levels during ovarian stimulation and IVF outcomes (302, 303, 304, 305). Recently, it was proposed that it might be more appropriate to look at a LH "window" instead of a single LH cutoff level, because there seems to be a "threshold" LH level, below which E2 production is not adequate, and a "ceiling" level, above which LH may be detrimental to follicular development (290).
The debate as to the optimal LH exposure for successful IVF outcomes continues. A novel approach recently proposed aims to improve outcomes by reducing the incidence of premature LH rises, as has been observed to occur in a small proportion of patients (306). It is suggested that earlier administration of GnRH antagonist could eliminate this problem. In a recent randomized study, neither follicular development nor the number of mature oocytes obtained was adversely affected by commencing GnRH antagonist on d 1 vs. d 6 of stimulation. However, LH and E2 exposure in the follicular phase was reduced in d 1 administration compared with initiation on d 6 of stimulation (307). Previous studies have suggested that prevention of high LH levels at the commencement of stimulation may improve endometrial receptivity (303, 307). Additional studies are required to ascertain the effects of the different GnRH antagonist protocols on endometrial maturation and implantation.
D. GnRH agonists
As outlined earlier, the introduction of GnRH agonists to prevent a premature rise in LH, premature oocyte maturation, and luteinization had a considerable impact on outcomes in IVF. They have now been in use for some 20 yr, yet surprisingly few dose finding studies have been performed (308), and randomized studies comparing different GnRH agonists are scarce. However, much attention has been given to discerning the optimal protocol for their use.
In the long protocol, GnRH agonist treatment usually commences in the luteal phase in the preceding cycle and is continued until hCG administration. Due to the intrinsic agonist activity of the compound, pituitary down-regulation is preceded by an initial stimulatory phase (referred to as the "flare" effect). This flare effect renders the approach of GnRH agonist long protocol for ovarian stimulation time consuming, because ovarian stimulation can only commence when pituitary quiescence has occurred, usually around 2 wk after commencing treatment (309). It is uncertain whether ovarian response to exogenous stimulation is affected by GnRH agonist cotreatment (310), and some women suffer from serious hypoestrogenic side effects, such as mood changes, sweating, and flushes. The "short" or "flare-up" protocol combines GnRH agonist therapy, started at cycle d 2, with gonadotropins initiated 1 d later (311). The immediate stimulatory action of the GnRH agonist serves as the initial stimulus for follicular recruitment. Adequate follicular maturation is on average reached in 12 d, which should allow enough time for sufficient pituitary desensitization to prevent any premature LH surges (312).
Several investigators have tried to shorten the duration of GnRH agonist administration by early cessation, because pituitary recovery after cessation takes around 14 d (313). The GnRH agonist is started in the midluteal phase of the preceding cycle and discontinued during or even before the FSH treatment is started. Several prospective randomized controlled studies have been performed comparing this approach with the long protocol (314, 315, 316, 317). Although premature rises in LH did not occur (confirming delayed pituitary recovery from desensitization), no clear clinical benefit has been demonstrated by this approach.
A meta-analysis comparing short and long IVF protocols showed a higher number of oocytes retrieved and higher pregnancy rates in the long protocol, although more units of gonadotropin were needed (318). In terms of gonadotropin suppression and number of retrieved oocytes, the midluteal phase of the preceding cycle is the optimal moment for the initiation of the GnRH agonist, in comparison to the follicular, early, or late luteal phase (319, 320). A major clinical advantage of the long protocol of GnRH agonist administration is the contribution to the planning of the oocyte retrieval because the initiation of exogenous gonadotropins after pituitary desensitization can be delayed, without a detrimental effect on IVF outcome (321, 322). A potential disadvantage with the luteal phase initiation of GnRH agonist is that spontaneous pregnancy present at the time of commencing treatment cannot be excluded with certainty. The extensive evidence supporting the long protocol has led to its widespread adoption as the standard of care (318). However, the recent clinical introduction of GnRH antagonists may ultimately lead to a new standard of care in IVF practice.
E. GnRH antagonists
GnRH antagonists may be administered at any time during the early or midfollicular phase of a treatment cycle to prevent a premature LH rise. Several studies have been performed to determine the minimal effective dose and treatment schedule in IVF patients (323, 324, 325). Two general approaches have emerged. In the single-dose protocol, one injection of 3 mg cetrorelix (ganirelix is not provided in this depot formulation) is administered in the late follicular phase on stimulation d 8 or 9. This is sufficient to prevent a LH surge in 80% of women (324). In the multiple-dose GnRH antagonist protocol, 0.25 mg cetrorelix or ganirelix is given daily from the sixth day of gonadotropin stimulation onward (323, 325). The rationale behind starting GnRH antagonist at least 5 d after commencing stimulation with gonadotropins is based on the reduced possibility of observing a premature LH rise in the early follicular phase (326).
Four large, industry-sponsored, prospective multicenter clinical trials comparing daily GnRH antagonist injections with long GnRH agonist protocols in IVF patients undergoing ovarian stimulation have been reported (327, 328, 329, 330). With a GnRH antagonist, the duration of gonadotropin treatment is shortened by 1–2 d, and slightly fewer follicles are seen at the time of hCG injection compared with a GnRH agonist. Therefore, the number of recovered oocytes tends to be lower. In these studies, no significant difference was found with respect to percentages of metaphase II oocytes, fertilization rates, and number of good quality embryos. Pregnancy rates were adequate in both groups in all four studies, but in every one the absolute rate was lower in the GnRH antagonist group. A meta-analysis of five large randomized trials showed an overall decrement in pregnancy rate of 5% (odds ratio, 0.75; 95% confidence interval, 0.62–0.97) (331) (Table 1). It has been hypothesized that the lower observed pregnancy rates may be a consequence of the currently advised treatment regimen. It has been suggested that the larger numbers of oocytes and embryos with agonists allow better selection, although the numbers of good quality embryos do not seem to be different. The GnRH antagonist was started on a fixed day of stimulation (d 6) in these studies, which may be too early for some patients and may lead to a diminished number (and quality) of oocytes.
|
. This novel approach promises a cost-effective and patient-friendly alternative to standard stimulation regimens. Based on the inverse association between implantation rates and ganirelix dose in the higher dosage groups in the large dose-finding study (337), direct effects of GnRH antagonists on human embryos have been suggested. Adverse effects were not observed on the freeze-thaw embryos of these cycles (338). Moreover, retrospective comparison of pregnancy rates after transfer of frozen-thawed two-pronucleate oocytes obtained in either a long GnRH agonist protocol (n = 286) or a GnRH antagonist protocol (n = 56) showed no differences in implantation, pregnancy, or miscarriage rates (339).
|
V. Adjuvant Therapies |
|---|
 TopAbstractI. IntroductionII. Physiology of Ovarian...III. The Development of... |
|---|
Abstract
I. Introduction
