Special Features. REVIEW: Aromatase Inhibitors for Ovulation Induction. Página 1 de 23. Abbreviations:

Transcription

1 Página 1 de 23 Use of this content is subject to the Terms and Conditions of the MD Consult web site. Journal of Clinical Endocrinology and Metabolism Volume 91 Number 3 March 2006 Copyright 2006 The Endocrine Society Special Features 760 REVIEW: Aromatase Inhibitors for Ovulation Induction Robert F. Casper 1 Mohamed F. M. Mitwally 2 1 Division of Reproductive Sciences (R.F.C.), Fran and Lawrence Bloomberg Department of Obstetrics and Gynecology, Samuel Lunenfeld Research Institute, Mount Sinai Hospital and The University of Toronto, Toronto, Ontario, Canada M5S 2X9; 2 and Division of Reproductive Endocrinology and Infertility (M.F.M.M.), Department of Obstetrics and Gynecology, Wayne State University, Detroit, Michigan Received August 29, Accepted November 29, Address all correspondence and requests for reprints to: Robert F. Casper, M.D., F.R.C.S.(C), 150 Bloor Street West, Suite 210, Toronto, Ontario, Canada M5S 2X9. rfcasper@aol.com. This work was supported by the Canadian Institutes of Health Research, Ottawa, Ontario, Canada. The authors have no conflict of interest. Copyright 2006 by The Endocrine Society F Abbreviations: AI Aromatase inhibitor CC clomiphene citrate COH controlled ovarian stimulation E2 estradiol ER estrogen receptor hcg human chorionic gonadotropin IUI

2 Página 2 de 23 intrauterine insemination IVF in vitro fertilization OHSS ovarian hyperstimulation syndrome PCOS polycystic ovary syndrome Context: For the last 40 yr, the first line of treatment for anovulation in infertile women has been clomiphene citrate (CC). CC is a safe, effective oral agent but is known to have relatively common antiestrogenic endometrial and cervical mucous side effects that could prevent pregnancy in the face of successful ovulation. In addition, there is a significant risk of multiple pregnancy with CC, compared with natural cycles. Because of these problems, we proposed the concept of aromatase inhibition as a new method of ovulation induction that could avoid many of the adverse effects of CC. The objective of this review was to describe the different physiological mechanisms of action for CC and aromatase inhibitors (AIs) and compare studies of efficacy for both agents for ovulation induction. Evidence Acquisition: We conducted a systematic review of all the published studies, both controlled and noncontrolled, comparing CC and AI treatment, either alone or in combination with gonadotropins, for ovulation induction or augmentation, identified through the Entrez-PubMed search engine. Evidence Synthesis: Because of the recent acceptance of the concept of using AIs for ovulation induction, few controlled studies were identified, and the rest of the studies were pilot or preliminary comparisons. Based on these studies, it appears that AIs are as effective as CC in inducing ovulation, are devoid of any antiestrogenic side effects, result in lower serum estrogen concentrations, and are associated with good pregnancy rates with a lower incidence of multiple pregnancy than CC. When combined with gonadotropins for assisted reproductive technologies, AIs reduce the dose of FSH required for optimal follicle recruitment and improve the response to FSH in poor responders. Conclusions: Preliminary evidence suggests that AIs may replace CC in the future because of similar efficacy with a reduced side effect profile. Although worldwide experience with AIs for ovulation induction is increasing, at present, definitive studies in the form of randomized controlled trials comparing CC with AIs are lacking. ( J Clin Endocrinol Metab 91: , 2006) FOR THE LAST 40 yr, the first line of treatment for anovulation in infertile women has been clomiphene citrate (CC). The choice of CC was appropriate because the drug was highly effective in inducing ovulation in selected patients with the advantages of being orally administered, relatively safe, and inexpensive. In contrast, alternative treatments usually involved parenteral gonadotropins that were significantly more complicated and uncomfortable to administer, expensive, and associated with more frequent and serious complications. However, CC was also found to have adverse effects, especially in the form of common antiestrogenic endometrial and cervical mucous changes that could prevent pregnancy in the face of successfully induced ovulation. We believed it was time to develop a more effective ovulation induction agent, and we have demonstrated that aromatase inhibitors (AIs) seem to be a promising alternative to CC for ovulation induction. The purpose of this review was to outline the data collected by us and others, over the past 5 yr, regarding the use of AIs for ovulation induction. We will start with a brief description of the physiology of the normal ovulatory cycle, information required for the reader to understand the mechanism of action and problems with CC as well as the mechanism of action and advantages of AIs. Physiological Basis of Ovulation The ovarian cortex at puberty contains hundreds of thousands of primordial follicles [1]. In response to unknown signals, independent of gonadotropins, a cohort (hundreds) of primordial follicles is recruited to grow [2]. During this early follicle development, the oocyte enlarges and the granulosa cells proliferate to form a preantral follicle. Over 3 6 months, the follicle develops FSH receptors in the granulosa cells and LH receptors in the theca cells, and the follicle forms a fluid-filled space called an antrum [1]. At this stage, antral

3 Página 3 de 23 follicles become acutely dependent on FSH for further development [1]. In natural cycles just before menses, falling estrogen levels result in withdrawal of negative feedback centrally leading to increased gonadotropins levels [3]. FSH stimulates granulosa cell proliferation and differentiation, with the development of more FSH receptors and the production of aromatase [4]. LH stimulates androstenedione production by theca cells that diffuses into the granulosa cell providing substrate for estrogen secretion. This step is catalyzed by the aromatase enzyme, which is induced by FSH. The so-called two-cell, two-gonadotropin theory [5] [6] postulated that FSH concentrations must exceed a certain level (FSH threshold) before follicular development will proceed. The duration of this period in which the threshold is exceeded (the FSH window) is limited in the normal cycle by a gradual decrease in FSH, occurring in the early-midfollicular First Published Online December 29, phase as a response to negative feedback from rising estrogen levels produced by the larger follicles [7]. Smaller follicles, with fewer FSH receptors, are no longer stimulated to grow by FSH levels below the FSH threshold and undergo atresia [8]. Therefore, generally only one follicle reaches the stage of ovulation each cycle, despite the fact that hundreds of primordial follicles, the number of which varies depending on a woman's age, may have begun development in the same cohort 3 6 months earlier. Administering exogenous FSH prolongs the time FSH levels are above the FSH threshold and extends the FSH window. This allows multiple ovulation by rescuing smaller follicles that would otherwise have undergone atresia. In addition, FSH also induces LH receptors in larger antral follicles above 1.0 cm in diameter [9]. At this point, LH can substitute for FSH in stimulating follicle growth and aromatase activity [10]. This observation led Sullivan et al. [11] and Filicori et al. [12] to postulate that exogenous LH/human chorionic gonadotropin (hcg) could be used as a substitute for FSH to continue follicular maturation before ovulation. Rapidly increasing levels of estradiol produced by the mature preovulatory follicle precede the midcycle LH and FSH surge that will initiate ovulation. The duration and circulating level of estradiol seem to be the determinant of the timing of the LH surge. Although the actual trigger of the surge is unclear, there is the suggestion that it may be a response to a rise in both estrogen and progesterone [13], with a role for diurnal and estradiol-dependent changes in GnRH neuron firing activity [14]. Angiogenesis and luteinization of granulosa and theca cells occur to form the corpus luteum in which an alteration of the steroidogenic pathway results in progesterone as the primary steroid hormone produced after luteinization. The corpus luteum retains the ability to produce estrogen. In addition, the demonstration of multiple waves of follicular growth in the majority of cycles [15] also suggests that some estrogen in the luteal phase may be contributed by growing follicles that will subsequently undergo atresia. With implantation and pregnancy, hcg production by the trophoblast results in maintenance of corpus luteum function. Luteal progesterone is required until about 8 wk gestation when the placenta takes over progesterone production [16]. In the absence of pregnancy, LH levels likely become too low to sustain the corpus luteum, as a result of very infrequent LH pulses [17]. The corpus luteum undergoes regression with a fall in progesterone and estrogen and the onset of menses. FSH levels rise with withdrawal of estrogen negative feedback and the next cohort of follicles begins to develop. The laboratory of Yen and colleagues [18] demonstrated, using naloxone (narcotic antagonist) infusions, that increased central endogenous opioid peptide activity is the cause of the decreased pulse frequency in the luteal phase. Such endogenous opioid activity is induced by elevated levels of progesterone. In summary, normal follicular development culminates in ovulation of a mature oocyte, followed by the development of a corpus luteum producing adequate amount of progesterone. This sequence of events is orchestrated by the interaction of local ovarian factors and endocrine factors from the pituitary and hypothalamus. The presence of subtle abnormalities despite the occurrence of ovulation may be responsible, at least in part, for unexplained infertility in some women, and perhaps for endometriosis-related infertility. On the other hand, overt anovulation or oligoovulation has been traditionally classified into three groups: World Health Organization (WHO) types I, II, and III anovulation. WHO Classification of Anovulation Women in WHO group I suffer from a defect at the level of the hypothalamus/pituitary. They are estrogen

4 Página 4 de 23 deficient with normal or low FSH or prolactin levels. They typically have amenorrhea and do not bleed in response to progestin challenge. They generally require pulsatile GnRH infusions or injectable gonadotropins to ovulate. Women in WHO group II are not estrogen deficient. Their FSH and prolactin levels are normal. They typically experience oligomenorrhea, but they may have anovulatory cycles or amenorrhea with bleeding in response to a progestin challenge. This is the most common type of anovulation and includes women with polycystic ovary syndrome (PCOS). For this group, oral agents such as insulin sensitizers, CC, or more recently AIs are useful for ovulation induction. Women in WHO group III include those with elevated gonadotropins secondary to primary ovarian failure mainly due to diminished ovarian reserve and loss of ovarian follicles. They are resistant to various methods of ovarian stimulation, and the best approach for their infertility is oocyte donation. The present review will focus on the use of the oral agents, CC and AIs, to induce ovulation in women with WHO group II anovulation. Clomiphene Citrate For more than 40 yr, CC has been the most commonly used oral agent for induction of ovulation. In 1961 Greenblatt et al. [19] published the first results of ovulation induction achieved by CC (at that time known as MRL/41). Since the early 1960s, pregnancy rates with CC treatment have not changed appreciably, despite advances in cycle monitoring. It is interesting to note that CC is considered pregnancy risk category X. This is particularly important when considering its relatively long clearance half-life of about 5 d to 3 wk (depending on the isomer) and that CC may be stored in body fat [20] [21]. Chemical structure and pharmacokinetics Chemically, CC is a nonsteroidal triphenylethylene derivative that exhibits both estrogen agonist and antagonist properties, i.e. selective estrogen receptor modulating activity. CC is a racemic mixture of two distinct stereoisomers, enclomiphene and zuclomiphene, having different properties. Available evidence indicates that enclomiphene is the more potent antiestrogenic isomer primarily responsible for the ovulationinducing actions of CC [20] [21]. Currently available clomiphene compounds are skewed toward enclomiphene predominance. Levels of enclomiphene rise rapidly after administration and fall to undetectable concentrations after a few days. Zuclomiphene is cleared far more slowly; levels of this less active isomer remain detectable in the circulation for more than 1 month after treatment and may actually accumulate over consecutive cycles of treatment [21]. Mechanism of action of CC 762 CC binds to estrogen receptors (ERs) throughout the body due to its structural similarity to estrogen. CC binding to ERs occurs for an extended period of time, i.e. weeks rather than hours as with natural estrogen. Such extended binding ultimately depletes ER concentrations by interfering with the normal process of ER replenishment [19] [22]. The antiestrogenic effect on the hypothalamus, and the pituitary, is believed to be the main mechanism of action for ovarian stimulation. Depletion of hypothalamic ERs prevents correct interpretation of circulating estrogen levels; estrogen concentrations are falsely perceived as low leading to reduced estrogen-negative feedback on GnRH production and subsequent increased gonadotropin (FSH and LH) secretion ( Fig. 1 ). The rise of FSH promotes growth of ovarian follicles and ovulation in anovulatory women. It is believed that the hypothalamus is the main site of action because in normally ovulatory women, CC treatment was found to increase GnRH pulse frequency [23]. However, actions at the pituitary level may also be involved because CC treatment increased pulse amplitude but not frequency in anovulatory women with PCOS, in whom the GnRH pulse frequency is already abnormally high [24]. During CC treatment, levels of both LH and FSH undergo a prolonged rise, compared with a natural cycle

5 Página 5 de 23 ( Fig. 2A ) [25] because of the prolonged ER depletion in the brain. Supraphysiological levels of estrogen can occur without central suppression of FSH because the normal ER-mediated feedback mechanisms are blocked as a result of prolonged ER depletion ( Fig. 1 ). This pertubation of normal negative feedback is reflected in the FSH window being extended, leading to multiple follicle growth and a higher multiple pregnancy rate with CC than occurs in natural cycles. Indications The two main indications for CC treatment are induction of ovulation in anovulatory infertility and stimulating multifollicular ovulation or enhancing ovulation in ovulatory infertile women, e.g. unexplained infertility. For anovulatory infertility, CC treatment will likely be effective only in conditions in which adequate levels of circulating estrogen exist to exert estrogen-negative feedback on gonadotropin production, which in turn could be antagonized by the antiestrogenic effect of CC as explained above. Adequate estrogen levels for a response to CC are hard to determine in absolute values. However, a withdrawal bleed after progesterone administration is usually a reasonable clinical evidence for adequate levels of circulating estrogen. CC is the initial treatment of choice for most anovulatory or oligoovulatory infertile women who are euthyroid and euprolactinemic but having adequate circulating levels of estrogen (WHO type II anovulation, e.g. PCOS. On the other hand, women with very low circulating estrogen levels such as (WHO types I and III) or women with a defective hypothalamic/pituitary axis such Figure 1. CC (d 5). Administration of CC from d 3 to 7 results in ER depletion at the level of the pituitary and mediobasal hypothalamus. As a result, estrogen-negative feedback centrally is interrupted and FSH secretion increases from the anterior pituitary leading to multiple follicular growth. D 10, By the late follicular phase, because of the long tissue retention of CC, there continues to be ER depletion centrally, and increased E2 secretion from the ovary is not capable of normal negative feedback on FSH. The result is multiple dominant follicle growth and multiple ovulation. AI (d 5), Administration of an AI from d 3 to 7 results in suppression of ovarian E2 secretion and reduction in estrogen-negative feedback at the pituitary and mediobasal hypothalamus. Increased FSH secretion from the anterior pituitary results in stimulation of multiple ovarian follicle growth. D 10, Later in the follicular phase, the effect of the AI is reduced and E2 levels increase as a result of follicular growth. Because AIs do not affect ERs centrally, the increased E2 levels result in normal negative feedback on FSH secretion, and follicles less than dominant follicle size undergo atresia, with resultant monofollicular ovulation in most cases. [Reproduced with permission from R. F. Casper: Fertil Steril 80: , 2003 [106]. The American Society for Reproductive Medicine.] as Sheehan's syndrome and Kallmann's syndrome will not respond to CC treatment. For ovulatory infertility, CC is believed to enhance chances of achieving pregnancy by stimulating multifollicular development as well as alleviating possible subtle ovulation dysfunction [26].

6 Página 6 de 23 Regimens of CC administration CC is administered orally, usually starting on the second to fifth day of spontaneous or progestin-induced menses. Treatment typically begins with a 50-mg tablet daily for 5 consecutive days, increasing by 50-mg increments in subsequent cycles until ovulation is induced. After ovulation is achieved, the same dose is repeated until pregnancy is achieved or a maximum number of around six cycles is reached. Once the effective dose of CC is established, there is no indication for further increases in dose unless the ovulatory response is lost. 763

7 Página 7 de 23 Figure 2. A, Mean serum FSH levels in natural cycles ( solid circles) and CC-treated cycles ( open circles) in 10 normal young women. Data are normalized to the day of the LH surge. Prolonged elevation if serum FSH is seen during CC treatment consistent with absent estrogennegative feedback because of prolonged ER depletion. B, Mean serum FSH levels in natural cycles ( solid circles) and letrozole-treated cycles ( open triangles) in nine normal young women. Data are normalized to the day of the LH surge. Serum FSH declines normally during the letrozole treatment consistent with an intact central feedback mechanism, in which estrogen suppresses FSH secretion and results in monofollicular ovulation. [Reproduced with permission from S. A. Fisher et al.: Fertil Steril 78: , 2002 [25]. The American Society for Reproductive Medicine.] Outcome of CC treatment Classically, CC treatment has been reported to induce ovulation in 60 80% of properly selected candidates. More than 70% of those who ovulate respond at the 50- or 100-mg dosage level. Cumulative conception rates up to 70% were observed after up to three successfully induced ovulatory cycles, [27] [28] [29]. In another study,

8 Página 8 de 23 cumulative conception rate of 73% was achieved within nine CC-induced ovulatory cycles [28]. Overall, cycle fecundity is approximately 15% in women who ovulate in response to treatment, with higher chance of pregnancy in the first cycles [27]. It is important to realize that these figures apply to young women in whom anovulation is the sole reason preventing them from conceiving. In the reality of daily clinical practice, such a group of patients does not frequently exist, particularly in a subspecialty referral infertility practice, in which much lower pregnancy rates are observed with CC induction of ovulation. Age, presence of other infertility factors, treatment history, and duration of infertility in addition to androgen levels are important factors affecting treatment outcome [28]. Amenorrheic women are more likely to conceive than oligomenorrheic women, probably because those who already ovulate, albeit inconsistently, are more likely to have other coexisting infertility factors. Generally speaking, failure to conceive within six CC-induced ovulatory cycles should be regarded as a clear indication to expand the diagnostic evaluation to exclude other factors or change the overall treatment strategy when evaluation is already complete [29]. Adverse antiestrogenic effects associated with CC treatment Because of the relatively long half-life of CC isomers, CC may exert unavoidable antiestrogenic effects on peripheral estrogen targets (endocervix and endometrium) that likely explain the absence of pregnancy despite ovulation observed in some CC-treated patients. Numerous studies in women and in various model systems have described adverse effects on the quality or quantity of cervical mucus, and endometrial growth and maturation [30] [31] [32] [33] [34] [35]. It is generally believed that these effects are most apparent at higher doses or after longer durations of treatment [31] [32] [33]. The endometrium is believed to be one of the most important targets of the antiestrogenic effect of CC and may explain a large part of the lower pregnancy rate and the possible higher miscarriage rate with CC. A reduction in endometrial thickness below the level thought to be needed to sustain implantation was found in up to 30% of women receiving CC [32]. This observation has been confirmed by other studies [34] [35]. It is important to mention that the absolute value for endometrial thickness, below which pregnancy seems unlikely, may be different from one center to another. This is obviously due to the inter- and intraobserver differences in measuring the endometrial thickness. However, in general, an endometrium that is thinner that 5 or 6 mm is usually associated with significant likelihood of failure to conceive [32] [34] [35]. In addition, successful implantation requires a receptive endometrium, with synchronous development of glands and stroma. Despite conflicting reports of CC effects on the endometrium [32] [33] [34] [35] [36] [37] [38], possibly due to different methodology used for endometrial assessment, a recent study prospectively applied a quantitative and objective technique to study the effect of CC on the endometrium in a group of normal women. This study found CC to have a deleterious effect on the endometrium, demonstrated by a reduction in glandular density and an increase in the number of vacuolated cells [38]. Decreased uterine blood flow during the early luteal phase and the periimplantation stage may also explain, at least in part, the poor outcome of CC treatment [39]. To summarize, the exact significance of adverse antiestrogenic 764 effects on pregnancy outcome in CC-treated women is difficult to quantify. Individual variability exists, likely because of the complexity of ER replenishment and activation and individual differences in the pharmacokinetics of CC. The discrepancy in the success rates in achieving pregnancy between women with unexplained infertility and women with PCOS in response to CC treatment (higher rates in PCOS women) suggests that PCOS women may be less vulnerable to the antiestrogenic effects of CC on peripheral tissues. A New Method of Ovulation Induction In our tertiary referral center for infertility, we frequently identified women, especially with unexplained infertility, who had failed to conceive with up to a year of CC treatment cycles before referral. Almost universally, we discovered an endometrial thickness less than 6 mm on transvaginal ultrasound. As a result of the antiestrogenic effects of CC on the endometrium in these and other women, we attempted development of a novel oral ovulation method that would be free of these adverse side effects. Aromatase was identified as a potential target for suppressing estrogen-negative feedback centrally and the availability of new specific AIs led to initial studies with these drugs.

9 Página 9 de 23 Aromatase Inhibitors Chemical structure and pharmacokinetics Aromatase is a microsomal cytochrome P450 hemoprotein-containing enzyme (the product of the CYP19 gene) and catalyzes the rate-limiting step in the production of estrogens, i.e. the conversion of androstenedione and testosterone to estrone and estradiol, respectively [40]. Aromatase activity is present in many tissues, e.g. ovaries, brain, adipose tissue, muscle, liver, breast tissue, and malignant breast tumors. The main sources of circulating estrogens are the ovaries in premenopausal women and adipose tissue in postmenopausal women [41]. Aromatase is a good target for selective inhibition because estrogen production is a terminal step in the biosynthetic sequence. A large number of AIs have been developed over the last 30 yr with the most recent, third-generation AIs licensed mainly for breast cancer treatment in postmenopausal women. The thirdgeneration AIs have been developed over the last 10 yr after the clinical failure of the second(fadrozole and Formestane, developed around 15 yr ago) and first-generation (aminoglutethimide, developed more than 30 yr ago) AIs. The clinical failure of the first two generations was mainly due to significant side effects associated with their use and the lack of satisfactory potency or specificity in inhibiting the aromatase enzyme [42] [43] [44]. The third-generation AIs commercially available include two nonsteroidal preparations, anastrozole [ZN 1033 (Arimidex)] and letrozole [CGS (Femara)], and a steroidal agent, exemestane (Aromasin), and are available for clinical use in North America, Europe, and other parts of the world for treatment of postmenopausal breast cancer. Letrozole and anastrozole are reversible, competitive AIs with considerably greater potency than aminoglutethimide (>1000 times) and, at doses of 1 5 mg/d, reduce estrogen levels by 97% to more than 99% down to concentrations below detection by most sensitive immunoassays. AIs are completely absorbed after oral administration with mean terminal halflife of approximately 45 h (range h) with clearance mainly by the liver. Mild gastrointestinal disturbances account for most of the adverse events, although these have seldom limited therapy [45] [46] [47]. Exemestane is a steroidal, suicide inhibitor of aromatase (aromatase inactivators) with a circulating half-life of approximately 9 h but potentially a longer effect to inhibit aromatase because it is irreversible [48]. Induction of ovulation with AIs Mechanism of action of AIs. We postulated that it would be possible to block estrogen-negative feedback, without depletion of ERs as occurs with CC, by administration of an AI in the early part of the menstrual cycle. Both circulating estrogen (produced mainly by the ovarian follicles and peripheral conversion of androgens in fat and other tissues) and locally produced estrogen in the brain exert negative feedback on gonadotropin release [49] [50] [51] [52]. Inhibition of aromatization will block estrogen production from all sources and release the hypothalamic/pituitary axis from estrogenic negative feedback ( Fig. 1 ). The resultant increase in gonadotropin secretion will stimulate growth of ovarian follicles. Withdrawal of estrogen centrally also increases activins, which are produced by a wide variety of tissues including the pituitary gland [53] and will stimulate synthesis of FSH by a direct action on the gonadotropes [54]. The selective nonsteroidal AIs have a relatively short halflife (~45 h), compared with CC, and would be ideal for this purpose because they are eliminated from the body rapidly [55] [56]. Because AIs do not deplete ERs, as does CC, normal central feedback mechanisms remain intact. As the dominant follicle grows and estrogen levels rise, normal negative feedback occurs centrally, resulting in suppression of FSH ( Fig. 2B ) and atresia of the smaller growing follicles. A single dominant follicle, and monoovulation, should occur in most cases ( Fig. 1 ). In women with PCOS, relative oversuppression of FSH may be the result of excessive androgen produced from the ovary being converted to estrogen by aromatization in the brain. The AIs suppress estrogen production in both the ovaries and brain. In the case of PCOS, therefore, AIs should result in a robust increase in FSH release and subsequent follicle stimulation and ovulation. The actual FSH release is likely to be blunted by the high levels of circulating inhibin found in PCOS patients [57] [58] [59] [60] that would not be altered by aromatase inhibition. In addition, as pointed out above, aromatase inhibition does not antagonize ERs in the brain, and the initiation of follicle growth accompanied by increasing concentrations of both

10 Página 10 de 23 estradiol and inhibin results in a normal negative feedback loop that limits FSH response, thereby avoiding the risk of high multiple ovulation and ovarian hyperstimulation syndrome (OHSS). Peripheral mechanism of action. A second hypothesis that may add to the mechanism of action of the AIs in ovarian stimulation involves an increased follicular sensitivity to FSH. 765 This could result from temporary accumulation of intraovarian androgens because conversion of androgen substrate to estrogen is blocked by aromatase inhibition. Recent data support a stimulatory role for androgens in early follicular growth in primates [61]. Testosterone was found to augment follicular FSH receptor expression in primates, suggesting that androgens promote follicular growth and estrogen biosynthesis indirectly by amplifying FSH effects [62] [63]. Also, androgen accumulation in the follicle stimulates IGF-I, which may synergize with FSH to promote folliculogenesis [64] [65] [66] [67]. It is likely that women with PCOS already have a relative aromatase deficiency in the ovary, leading to increased intraovarian androgens [68] [69] that leads to the development of multiple small follicles responsible for the polycystic morphology of the ovaries. The androgens, as described above, may also increase FSH receptors making these PCOS ovaries exquisitely sensitive to an increase in FSH through either exogenous administration of gonadotropins (hence the high risk of OHSS) or endogenous increases in FSH as a result of decreased central estrogen feedback induced by aromatase inhibition. In the latter case, a relatively small rise in FSH, because of a normal estrogen feedback loop as described above, generally leads to monofollicular ovulation, thus avoiding the occurrence of OHSS. Another part of the peripheral hypothesis involves ERs in the endometrium. It is possible that aromatase inhibition, with suppression of estrogen concentrations in the circulation and peripheral target tissues, results in up-regulation of ERs in the endometrium, leading to rapid endometrial growth once estrogen secretion is restored. Estrogen has been shown to decrease the level of its own receptor by stimulating ubiquitination of ERs, resulting in rapid degradation of the receptors. In the absence of estrogen, ubiquitination is decreased allowing up-regulation of the ER and increasing sensitivity to subsequent estrogen administration [70]. This could increase endometrial sensitivity to estrogen resulting in more rapid proliferation of endometrial epithelium and stroma and improved blood flow to the uterus and endometrium [71]. As a result, normal endometrial development and thickness should occur by the time of follicular maturation, even in the face of the observed lower-than-normal estradiol concentrations in Al-treated cycles. Indications for AIs in induction of ovulation. As a result of the mechanisms of action of the AIs described above, we proposed that AIs could be used alone for induction of ovulation or as an adjuvant in conjunction with exogenous FSH or other medications to improve the outcome of ovulation induction. A major advantage of an AI used alone is the ability to achieve restoration of monofollicular ovulation in anovulatory infertility, e.g. PCOS as a result of the intact estrogennegative feedback loop as described above. An AI could also be used in conjunction with FSH injections to increase the number of preovulatory follicles that develop and improve the outcome of treatment. An increase in intraovarian androgen concentrations during aromatase inhibition could improve ovarian response to exogenous gonadotropins by increasing ovarian sensitivity to FSH (peripheral mechanism of action described above). To summarize, the AI when used alone should result in a predictable response with the development of one or two mature follicles and a significantly reduced risk for OHSS and multiple gestation. To achieve multiple ovulation, the addition of FSH to the AI is likely necessary. Clinical studies of AIs for ovulation induction Induction of ovulation after CC failure.

11 Página 11 de 23 The first study we performed was a proof-of-concept trial in a group of 22 women who had failed to respond to CC and were about to move on to injectable gonadotropin treatment with its associated costs and risks [72] [73]. We defined inadequate response to CC as failed ovulation induction or ovulation with a very thin midcycle endometrium ( 0.5 cm). Patients who failed to respond adequately to CC were offered the option of trying the AI, letrozole, as an alternative treatment for induction of ovulation. In the first study group, 12 women with PCOS received letrozole 2.5 mg daily from d 3 to 7 after a progestinwithdrawal bleed. During letrozole treatment, ovulation occurred in nine patients (75%), including three of the four patients who were anovulatory with CC. The mean endometrial thickness in the women receiving letrozole for ovulation induction was 0.81 ± 0.14 cm. Pregnancy was achieved in three cycles (25%), two of which were singleton clinical pregnancies and one was a chemical pregnancy. Patients in the second group were ovulatory women with unexplained or mild male factor infertility and endometrial thickness 0.5 cm or less on CC. Each patient subsequently received one letrozole treatment cycle of 2.5 mg from cycle d 3 to 7. During letrozole treatment, all women ovulated with endometrial thickness greater than 7 mm. A singleton clinical pregnancy (10%) resulted from timed intrauterine insemination (IUI) in one unexplained infertility couple who developed two follicles greater than 1.5 cm. A prospective, randomized, controlled trial of CC vs. an AI for augmentation of ovulation in women with unexplained infertility was carried out at about the same time by Biljan and colleagues [74] in Montreal. Twentyfour patients were randomized to receive CC 100 mg daily from cycle d 3 to 7, and 26 patients received letrozole 2.5 mg daily from d 3 to 7. Using standard hormonal and ultrasound monitoring, the investigators showed that the CC group had a mean of two preovulatory follicles, whereas the letrozole group produced a mean of one. Estradiol levels on the day of hcg were significantly elevated in the CC group (2300 pmol/liter), compared with the letrozole group (600 pmol/liter). Endometrial thickness and blood flow were significantly less with CC (6.9 mm and pulsatility index of 3.6), compared with letrozole (8.6 mm and pulsatility index of 3.1) on the day of hcg. The pregnancy rate was 5.6% in the CC group vs. 16.7% in the letrozole group (nonsignificant). We concluded from these preliminary studies that AIs were an effective alternative to CC, particularly in cases with recurrent CC failure. However, it is important to point out that in cases of CC resistance (failure to ovulate) due to severe insulin resistance or the use of CC for inappropriate indications ( e.g. hypothalamic amenorrhea or ovarian failure), the use of an AI is also unlikely to be successful. The correction of insulin resistance with an insulin sensitizer is the logical 766 approach in patients with insulin resistance. Alternative treatments should be considered for other problems such as exogenous gonadotropin injection in patients with hypogonadotropic hypogonadism and oocyte donation for cases with ovarian failure. Since this first study, evidence supporting the success of AIs in ovulation induction for infertility treatment has been accumulating [74] [75] [76] [77] [78]. Most of the studies used the AI, letrozole. However, anastrozole, another third-generation AI similar to letrozole, was used in other studies [78] [79]. It is currently not known whether there are any clinically significant pharmacological differences between letrozole and anastrozole [80], especially regarding efficacy of ovulation induction [81]. AIs plus gonadotropins Reducing FSH dose required for optimum controlled ovarian hyperstimulation. We investigated the idea of combining an AI with FSH injection to reduce the dose of FSH required to achieve optimum controlled ovarian stimulation (COH), without adverse antiestrogenic effects. We [82] [83] [84] and others [85] [86] found a significant reduction in the FSH dose required (from 45 to 55%). We also compared the use of an AI in conjunction FSH with the use of CC in conjunction with FSH or FSH alone [83]. The study included women with unexplained infertility and mild male factor infertility. Thirty-six women received an AI with FSH in 42 treatment cycles, 18 women received CC with FSH in 19 cycles, and 56 women received FSH only in 91 cycles. In this study, we found that cotreatment with an AI significantly reduced the FSH dose required during COH to the same extent as did CC. The AI was not associated with

12 Página 12 de 23 antiestrogenic effects as demonstrated by the significantly lower endometrial thickness noted with CC treatment despite the significantly higher estradiol (E2) level. In addition, the pregnancy rate with an AI and FSH was equivalent to FSH alone and almost twice the level seen with CC and FSH treatment [83]. It is known that the cost of FSH injections constitutes a significant part of the expense of infertility treatment, especially during assisted reproduction. We believe that the AIs will markedly reduce the cost of infertility treatment by decreasing the FSH dose required for optimum ovarian stimulation. This could make assisted reproductive technology available to a larger group of infertile couples. Improving ovarian response to FSH stimulation in poor responders. The reduction in the dose of FSH required observed in conjunction with an AI encouraged us to explore the use of an AI to improve response to ovarian stimulation with FSH in poor responders. In an observational cohort study [82], 12 patients with unexplained infertility and a poor response to ovarian stimulation with FSH in at least two cycles (total of 25 cycles of FSH-only stimulation) were studied. These women underwent COH for IUI and poor response defined as less than three follicles greater than 1.8 cm in diameter on the day of the LH surge or hcg administration. In the experimental cycles, letrozole was given at a dose of 2.5 mg from d 3 to 7 after onset of menses, and FSH injection was started on d 7 of the menstrual cycle at a dose of IU/d, depending on the number of developing follicles seen on ultrasound. The aim was to achieve three preovulatory follicles on the day of hcg injection. During FSH plus letrozole stimulation cycles, the mean number of mature follicles was 3.3, which was significantly higher than in FSH-only cycles (1.9 follicles). The amount of FSH required was significantly lower in the letrozole plus FSH cycles than the FSH-only cycles (616 ± 454 vs ± 708 IU, respectively). There was no significant difference between FSH only or letrozole plus FSH on the day of hcg administration or endometrial thickness on the day of hcg administration. Although there was no difference in the level of E2 on the day of hcg administration, the amount of estrogen per mature follicle was significantly lower with letrozole treatment, consistent with previous studies of using an AI for ovulation induction. Three of the women conceived a clinical pregnancy with the combined letrozole and FSH treatment. In this clinical trial, we demonstrated a benefit of aromatase inhibition in improving ovarian response to FSH stimulation in poor responders. The improved response was clearly shown by the significantly higher number of mature follicles and significantly lower amount of FSH needed to achieve an adequate number of preovulatory follicles. In addition, the endometrium sustained implantation as demonstrated by three clinical pregnancies in the letrozole and FSH cycles. Optimal dose of AIs for repeated administration The optimal dose of each AI is not yet clear. In most of the studies to date, the dose of letrozole (2.5 mg) or anastrozole (1.0 mg) typically used for breast cancer treatment in postmenopausal women has been chosen. Biljan et al. [85] in a randomized study comparing 2.5 and 5.0 mg of letrozole in women with unexplained infertility suggested that the higher dose might be associated with more follicles developing. However, the study was not large enough to demonstrate a significant advantage. A more recent study by Healey et al. [86] used a dose of 5.0 mg of letrozole together with FSH in an overlapping regimen for superovulation in women undergoing IUI. Compared with FSH treatment alone, the combination of letrozole and FSH resulted in a reduction of gonadotropin dose, similar pregnancy rates, but a slight negative effect on endometrial thickness, likely because the overlapping regimen advanced follicle growth and ovulation and did not allow time for clearance of the larger dose of letrozole by the time of hcg administration. Similarly, a study using 7.5 mg letrozole from cycle d 3 to 7 showed, for the first time, a thinning of the endometrium similar to CC [76]. Based on current data, it is likely that the optimal dose of letrozole for a 5-d course of treatment is between 2.5 and 5.0 mg, with higher doses resulting in persistence of aromatase inhibition and estrogen levels too low for normal endometrial development by the time of ovulation. For anastrozole, there are not yet enough data to determine the preferred dose, although it appears that the standard 1-mg dose is too low for optimal follicle recruitment and ovulation [78]. Adverse effects and concerns about using AIs for induction of ovulation Side effects of AIs. 767

13 Página 13 de 23 In clinical use, nonsteroidal AIs are generally well tolerated. The main side effects are hot flushes, headaches, and leg cramps [87] [88]. These adverse effects were observed in older women with advanced breast cancer who were given the AIs on a daily basis over several months. Fewer adverse effects would be expected in healthy young women administered a short course of AI for induction of ovulation. In addition, our clinical experience with ovulation induction has been fewer side effects such as hot flushes and premenstrual syndrome-type symptoms with AIs, compared with CC. Aromatase inhibition is associated with significantly lower serum estrogen levels at midcycle and per mature follicle than found with CC [73] [83]. The question whether low or very low intrafollicular estrogen is compatible with follicular development, ovulation, and corpus luteum formation has been reviewed before [89]. Markedly reduced to absent intrafollicular concentrations of estrogen are known to be compatible with follicular expansion, retrieval of fertilizable oocytes, and apparently normal embryo development [90] [91] [92] [93]. The rapid clearance of the AIs, the reversible nature of enzyme inhibition, and elevated levels of FSH, which induces new expression of aromatase enzyme, are factors that limit accumulation of androgens and likely result in increasing estrogen production that should be relatively normal at the time of ovulation. This conclusion has now been confirmed by the use of AIs in in vitro fertilization (IVF) reviewed below. Pregnancy outcome with AIs We recently reported the clinical outcome of pregnancies obtained through the use of AIs for ovulation induction or COH for IUI [94]. We described a cohort study comparing the outcome of pregnancies achieved after letrozole and other ovarian stimulation treatments with a control group of pregnancies spontaneously conceived without ovarian stimulation. In three tertiary referral centers over a 2-yr period, there were 394 pregnancy cycles in 345 infertile couples (133 pregnancies with 2.5 or 5 mg letrozole alone or gonadotropins, 113 pregnancies with CC alone or with gonadotropins, 110 pregnancies with gonadotropins alone, and 38 pregnancies achieved without ovarian stimulation). Pregnancy rates in the various treatment groups are shown in Fig. 3A. Pregnancies conceived after AI treatment was associated with comparable miscarriage and ectopic pregnancy rates, compared with all other groups including the spontaneous conceptions. In addition, letrozole use was associated with a significantly lower rate of multiple gestation, compared with CC (4.3 vs. 22%, respectively) ( Fig. 3B ), consistent with our hypothesis of an intact negative feedback loop centrally with aromatase inhibition. Recent safety studies have found anastrozole to have no teratogenic or clastogenic effects in animal embryo development, whereas there have been some concerns regarding teratogenic effects of letrozole if administered unintentionally during pregnancy [95]. Because of the short half-life of each drug, administration in the early follicular phase should result in clearance of the agents before implantation takes

14 Página 14 de 23 Figure 3. A, Pregnancy outcome in 3748 cycles of timed intercourse or IUI in women undergoing cycle monitoring with serial transvaginal ultrasound scans and serum E2 and LH determinations. The column labeled Natural refers to women undergoing cycle monitoring and timed intercourse or IUI with no drug treatment. B, Multiple gestation rates associated with the different ovarian stimulation treatments. There were no multiple gestation cases among pregnancies that occurred spontaneously. Letrozole (Let) treatment, at both the 2.5 and 5.0 mg/d doses, was associated with significantly lower multiple gestation rates, compared with all other methods of ovarian stimulation. CC treatment (alone or with gonadotropins) was associated with significantly higher multiple gestation rates, compared with all other ovarian stimulation treatments. [Reproduced with permission from M. F. Mitwally et al.:am J Obstet Gynecol 192: , 2005 [95]. Elsevier Inc.] place. Nevertheless, care should be taken in all cases of ovulation induction with these drugs to ensure that the patient is not pregnant before administration. We recommend that

15 Página 15 de 23 a blood βhcg level be ascertained before using AIs in premenopausal women. 768 Recently, Health Canada and Novartis Pharmaceuticals issued a warning that letrozole should not be used for ovulation induction because of the potential for fetal toxicity and malformations. This warning was based on an abstract by Biljan et al. [96] in which six congenital abnormalities and one case of hepatocellular carcinoma were reported among 150 births resulting from the use of letrozole for ovulation induction. These 150 babies were compared with a database of over 36,000 spontaneously conceived babies born in a lowrisk community hospital in Montreal. Although the overall anomaly rate was not increased, the authors reported a significant increase in locomotor and cardiac abnormalities in the letrozole-treated pregnancies compared to the controls. We believe a larger group of letrozole babies needs to be assembled and compared to a more suitable control group of infertile women before any conclusion of an increase in congenital anomalies with letrozole can be confirmed. We are now in the process of collecting such data. AIs for IVF Despite recent developments in ovarian stimulation and assisted reproductive technologies, there has not been a corresponding increase in implantation rates. Supraphysiological levels of estrogen may explain some of the adverse effects of ovarian stimulation on infertility treatment, including deleterious effects on the endometrium [97] [98] and embryo [99] [100]. Basir et al. [101] have shown that high response to COH for IVF results in endometrial glandular and stromal dyssynchrony as assessed by morphometric analysis. A stepdown protocol has been proposed for IVF to lower E2 concentrations and improve successful implantation [102]. An alternative approach is using an AI to significantly suppress E2 levels around d 3 7 of the menstrual cycle. After stopping the AI, E2 levels increase steadily to reach levels high enough to trigger an LH surge around d of the cycle. However, preovulatory levels of estrogen remained about half to one third the levels seen in CC- or FSH-stimulated cycles [83]. This strategy may work similarly to a step-down protocol without jeopardizing the potential number of oocytes retrieved. Clinical studies of AIs for ovarian stimulation in IVF In a randomized, controlled study, Goswami et al. [103] compared an AI plus FSH protocol with a standard GnRH agonist and FSH protocol in poor responders undergoing IVF. Although this was a small pilot study including only 38 patients, the results suggested that the addition of an AI to a small dose of FSH (150 IU; two injections of 75 IU on cycle d 3 and 8) resulted in a similar number of oocytes retrieved, embryos transferred, and pregnancy rate as observed in the women on the standard protocol receiving a mean (± SEM) of 2865 ± 228 IU FSH. The authors concluded that AIs could be a low-cost alternative to natural-cycle IVF in patients who are poor responders to FSH and that larger randomized studies are required to confirm their data. A recent, larger study by Garcia-Velasco et al. [104] evaluated the use of an AI as an adjuvant to FSH treatment in IVF cycles of poor-responder patients. In this study, 147 low responders, with at least one previous canceled IVF cycle, were enrolled. The study was prospective but not randomized. The women were divided into a control group of 76 patients treated with high-dose gonadotropins in a GnRHantagonist regimen. The experimental group of 71 patients received letrozole 2.5 mg plus gonadotropins for the first 5 d of stimulation followed by the same gonadotropin/antagonist regimen. The study demonstrated that patients receiving an AI had higher numbers of oocytes retrieved and had a higher implantation rate despite receiving the same doses of FSH/human menopausal gonadotropin as the control group. Both testosterone and androstenedione were significantly increased in concentration in the follicular fluid in the experimental group, compared with the controls. Interestingly, follicular fluid E2 levels were similar to controls. These findings are consistent with the hypothesis that aromatase inhibition, by blocking androgen to estrogen conversion, increases intraovarian androgens and follicular FSH receptor expression and sensitivity to FSH administration and is rapidly reversible. Another reason to avoid elevated estrogen levels during ovarian stimulation is fertility treatment in women with breast cancer. Treatment for this cancer often involves chemotherapy with alkylating agents that can damage ovarian follicle reserve leading to premature ovarian failure. With the recent success of oocyte cryopreservation, some women are opting to freeze oocytes or embryos for later use by themselves or a gestational carrier. Oktay et al. [105] studied 60 women (age range yr) with breast cancer. Twenty-nine women requested IVF before chemotherapy, and 31 women did not undergo IVF and served as the control group. The 29 patients underwent 33 ovarian stimulation cycles with either tamoxifen 60 mg/d alone or in combination with low-dose FSH (TamFSH-IVF) or letrozole 5 mg in combination with FSH (letrozole-ivf).