The Hypothalamic-Hypophyseal-Ovarian Axis and the Menstrual Cycle

The menstrual cycle requires precise coordination between several processes in the body.^{1, 2} The major components of this control system include the hypothalamic gonadotropin-releasing hormone (GnRH) pulse generator, the pituitary gonadotropes, the ovaries, and the uterus. The gonadotropes respond to GnRH pulses by releasing the gonadotropins, follicle-stimulating hormone (FSH), and luteinizing hormone (LH), which stimulate folliculogenesis and steroid and peptidergic hormone secretion from the ovaries. Hypothalamic and pituitary activities are strictly controlled by ovarian hormone feedback loops, whereas the GnRH pulse generator is also modulated by a variety of inputs from other neural centers.

The menstrual cycle is divided into two successive phases. The follicular phase represents the process whereby a follicle is selected and becomes a mature follicle destined to ovulate. This first phase is dominated by estradiol secretion. Ovulation in response to a large release of gonadotropins signals the beginning of the second phase, the luteal phase, in which the ovulated follicle is transformed into a corpus luteum. The dominant ovarian hormone secreted in this phase is progesterone, supplemented by estradiol. Changing cyclic ovarian steroid hormone patterns prepare the uterus for implantation, were fertilization to occur in that cycle. At the end of the luteal phase, ovarian steroid hormone secretion collapses. This terminates the support of the endometrium, and menstruation occurs. The menstrual cycle lasts for 25–30 days in most women. By convention, the day of menstruation is designated as day 1 of the menstrual cycle, although the FSH signal initiating the cycle (see later) may occur 2–3 days before this. Although the follicular phase lasts 14 days in the typical 28-day cycle, in reality its length is variable. In contrast, the life span of the corpus luteum is remarkably constant, and the luteal phase lasts 13–15 days.

There are four major circulating hormones that can easily be monitored during the menstrual cycle: FSH, LH, estradiol, and progesterone. Although the concentrations of these hormones in blood can vary substantially on an hourly basis, their daily profiles provide characteristic changes during the cycle (Fig. 1). The important event in regard to FSH is a small but significant rise in its levels at the end of the preceding cycle. This rise heralds the recruitment of secondary early antral follicles and the process whereby a follicle will be selected for ovulation. The most striking change in LH secretion occurs at the end of the follicular phase, when there is an abrupt rise in its concentration. This is the preovulatory gonadotropin surge that initiates the ovulatory process. Estradiol reflects the secretory activity of the growing follicle, and its concentrations rise in parallel with follicular growth, reaching highest levels when the follicle achieves maturation. Progesterone reflects the secretory activity of the corpus luteum. It rises and falls in a characteristic 14-day bell-shaped curve representative of the finite life span of the corpus luteum. Estradiol is also secreted by the corpus luteum in a similar rise-and-fall fashion. Other hormones also are released in a cyclic fashion (e.g., the inhibins) (see below).

The hypothalamic unit and the gonadotropin-releasing hormone (GnRH) pulse generator

The GnRH pulse generator is the primary structure that drives the menstrual cycle. In the absence of a functional GnRH pulse generator, the gonadotropes remain unstimulated and the ovaries dormant. The GnRH pulse generator is the hypothalamic structure that releases GnRH, a decapeptide that stimulates LH and FSH synthesis and release from the pituitary. GnRH is synthesized in specialized neurons as part of a larger precursor molecule (prohormone) containing 92 amino acids that is processed before release during axonal transport from the hypothalamic neuron to the median eminence.³ Although GnRH neurons in the hypothalamus appear as a dispersed population spread over several classic architectonic areas,⁴ the primary GnRH neurons controlling the pituitary gland in primates appear to be located within the arcuate nucleus (a structure that contains several other small peptides, neurotransmitters, and endogenous opiates).⁵ Upon reaching the median eminence, GnRH axons end on capillary loops that give rise to long portal veins that descend along the pituitary stalk to terminate in anterior pituitary sinusoids. This anatomic arrangement is essential because it allows for the rapid and undiluted transport of the neurohormone, which has a half-life of only a few minutes, to its target receptor on the gonadotrope in the anterior pituitary gland.

GnRH neurons, in contrast to other neurons, do not originate within the brain.⁶ They derive from progenitor cells in the embryonic olfactory placode and during fetal development the GnRH neurons migrate into the brain to reach the locations that they will occupy during adult life.⁶ Functional connections between these neurons and the hypophyseal portal system are established by 16 weeks of fetal life.⁴ Completed migration failure and lack of establishment of functional connections are characteristic of patients with hypogonadotropic hypogonadism, which when accompanied by anosmia fullfills the diagnosis of Kallmann’s syndrome.⁷ In the 19-week fetus with X-linked Kallmann’s syndrome, the GnRH neurons accompanying the olfactory nerves are shown to be arrested in their voyage within the meninges, and contact with the brain and the hypophyseal portal system cannot be established.⁸ The migration of GnRH neurons over long distances and through changing molecular environments suggest that numerous factors, local and possibly external, influence this process at its different stages.^{9, 10} Factors that mediate the adhesion of GnRH neurons to changing surfaces along their voyage, that promote cytoskeletal remodeling and modulate axonal guidance, may play critical roles.^{11, 12}

GnRH causes the release of both gonadotropins from the anterior pituitary gland. One essential aspect of gonadotropin secretion is that LH and FSH are released in a pulsatile rather than a continuous fashion. Each pulse of LH consists of the abrupt release of the hormone from the gonadotrope into the peripheral circulation, followed by an exponential decline representative of the half-life of the hormone. As is now well established in several animal species, pulsatile gonadotropin release is not the result of an intrinsic property of the anterior pituitary gland but is caused directly by the pulsatile release of GnRH (the GnRH pulse generator).⁵ Pulses of GnRH have been measured in hypophyseal portal blood and cerebrospinal fluid of nonhuman primates and sheep, and synchrony between pulses of GnRH in portal blood and of LH in peripheral blood is well shown.^{13, 14} Left to itself and in the absence of endogenous feedback systems and of exogenous influences, the pulse generator releases a GnRH pulse at about hourly intervals (Fig. 2). Pulsatility seems to be an intrinsic property of the GnRH neuron early on, because cultured neurons obtained from the olfactory placode and from migratory pathways in fetal life already release GnRH in a pulsatile manner.¹⁵ What accounts for the synchronous discharge of widely scattered GnRH neurons remains to be determined. One recently identified peptide, kisspeptin, may, among others, play a signaling role in this process: several studies have shown that kisspeptin stimulates the release of GnRH by activating receptors expressed by GnRH neurons.¹⁶

Fig. 2. Hourly pulsatile GnRH release during two periods of 8 hours in an ovariectomized monkey (upper levels). Note the concordance of LH (lower levels) and GnRH pulses. (Xia L, Van Vugt D, Alston EJ, et al: A surge of gonadotropin-releasing hormone accompanies the estradiol-induced gonadotropin surge in the rhesus monkey. Endocrinology 131: 2812, 1992; used by permission of The Endocrine Society.)

The hypothalamic GnRH pulse generator is operational at birth; significant pituitary and gonadal secretory activity occurs at that time and for a while thereafter. Later on, however, and throughout infancy until the prepubertal period, pulsatile gonadotropin secretion is significantly dampened.¹⁷ This is seen even in the absence of the gonads, as shown in patients with gonadal dysgenesis.¹⁸ Initiation of puberty is heralded by the resumption of a frank pulsatile pattern which is at first only seen in conjunction with deep sleep. This sleep-related pattern disappears in the later stages of puberty and is not apparent in adulthood when pulsatile secretion occurs throughout the 24-hour period. There are reports, however, of a nocturnal slowing of pulsatile LH release in a small number of women during the follicular phase of the menstrual cycle.¹⁹

The pituitary unit

The gonadotropins, LH and FSH, are synthesized within the gonadotrope in the anterior pituitary gland. These two hormones are glycoproteins containing two subunits (α and β). The β subunits of LH and FSH differ in their sequence. The β subunit confers hormone specificity. The α subunit is common to LH and FSH (as well as thyroid-stimulating hormone and human chorionic gonadotropin). Biologic activity of the hormone requires the association of both subunits.²⁰

The relevance of pulsatile GnRH release to the reproductive process is illustrated by observations that only pulsatile GnRH administration, provided at an appropriate frequency, restores normal gonadotropin secretion in individuals lacking endogenous GnRH secretion.²¹ In contrast, continuous (nonpulsatile) exposure to GnRH, even to high doses, produces a desensitization of its receptor on the gonadotrope, resulting with time in a lowering of LH and FSH release.^{22, 23} The cellular mechanism of desensitization (also referred to as down-regulation) remains to be elucidated and may implicate receptor and postreceptor mechanisms.²⁴

Although GnRH is known to release both LH and FSH, the relative amounts of each gonadotropin released may vary in response to ovarian feedback input (see later). In addition, it is now clear that not only is a pulsatile GnRH signal required for the stimulation of gonadotropin subunit gene transcription, but also that changes in the profile of pulsatile GnRH secretion differentially influence the synthesis and release of each gonadotropin. In particular, the frequency of GnRH pulse release selectively regulates gonadotropin subunit gene transcription with a faster frequency favoring LH β subunit gene expression and a lower frequency favoring FSH β subunit.²⁵ In concurrence with this finding, studies in monkeys in which the GnRH pulse generator was lesioned and in GnRH-deficient patients show that a high GnRH pulse frequency infusion favors LH release, whereas a low GnRH pulse frequency favors FSH secretion^{26, 27, 28} (Fig. 3).

The ovarian unit and ovarian-hypothalamic and pituitary feedback loops

Both gonadotropins act on the ovaries to induce morphologic changes and ovarian steroid secretion. Morphologic processes include folliculogenesis (i.e., the cyclic recruitment of a pool of follicles to produce a mature follicle ready for ovulation) and the formation of a corpus luteum. These processes occur in sequence conferring a monthly rhythm to the reproductive cycle (see later). Granulosa and thecal cells within the follicle and luteal cells respond to LH by synthesizing and releasing ovarian steroids, mainly estradiol 17β and progesterone. The type and amount of hormone released depend on the status of the follicle and the corpus luteum.

Feedback communication between the ovaries and the hypothalamic-pituitary unit is an essential component to the physiology of the reproductive cycle. It is important for the brain and pituitary gland to modulate their secretion in response to the activity status of the ovary. Estradiol and progesterone play a major role in these feedback communications. As in other endocrine systems, the major feedback loop is inhibitory (the negative feedback loop), whereby the steroid secreted by the target organ (the ovary) regulates the hypothalamic-hypophyseal unit to adjust GnRH and gonadotropin secretion (Fig. 4).

Fig. 4. In the primate, the gonadotropin-releasing hormone (GnRH) pulse generator is probably located within the arcuate nucleus of the hypothalamus. GnRH pulses are released in hypophyseal portal veins, which transport GnRH to the pituitary to stimulate pulsatile luteinizing hormone (LH) and follicle-stimulating hormone (FSH) release. These two gonadotropins stimulate folliculogenesis and steroid and peptidergic hormone secretion from the ovaries. Hypothalamic (GnRH) and pituitary (LH and FSH) activities are strictly controlled by ovarian hormone (estradiol and progesterone) feedback loops. MBH, mediobasal hypothalamus; ME, median eminence.

Estradiol-17β is a potent physiologic inhibitor of GnRH and of gonadotropin secretion.²⁹ The threshold for the negative feedback action of estradiol is such that even small increases in the levels of the hormone induce a decrease in gonadotropins. Levels of LH and FSH during the follicular phase vary in accord with the changes in estradiol concentrations that accompany maturation of the follicle. As circulating estradiol levels increase during the follicular phase, gonadotropin concentrations decrease. In menopausal women or women who have undergone ovariectomy, in whom estradiol secretion is deficient, sustained increases in LH and FSH release occur. In these conditions, the administration of physiologic doses of estradiol results in a rapid and sustained decrease in LH and FSH to levels equivalent to those seen during the menstrual cycle (Fig. 5A).^{30, 31} The estradiol negative feedback loop acts to decrease LH secretion rapidly, mainly by controlling the amplitude of the LH pulse. As the follicular phase progresses, LH pulse amplitude declines. LH pulse frequency during the follicular phase (at 60–100-minute intervals) approximates that observed at menopause or after ovariectomy, suggesting that estradiol does not particularly affect LH pulse frequency. (At higher physiologic concentrations, estradiol also exerts a separate stimulatory [positive feedback loop] effect on gonadotropin secretion. This is described later.) Progesterone, at high concentrations such as those observed during the luteal phase of the cycle, also exerts an inhibitory effect on gonadotropin secretion. In contrast to estradiol, progesterone affects mainly LH pulse frequency. Experimentally a significant decrease in the frequency of the LH pulse can be induced by treating women in the follicular phase with progesterone.³²

In view of these feedback loops, it is not surprising that the characteristics of pulsatile LH secretion vary greatly with the stage of the menstrual cycle (Fig. 6). During the estrogenic stage or follicular phase (Fig. 6A), pulses of high frequency but of low amplitude are seen, whereas during the progesterone stage or luteal phase (Fig. 6B), there is a progressive reduction in the frequency of the LH pulse, with pulse intervals occurring every 200 minutes or more by the end of the luteal phase. This decreased pulse frequency is accompanied by a significant increase in pulse amplitude.

Inhibins, a family of peptides of ovarian origin, are glycoproteins that consist of a dimer, with two dissimilar α and β subunits.³³ The two subunits are coded by different genes. Two forms of the β subunit have been identified and thus inhibin can exist as α-βA (inhibin A) and as α-βB (inhibin B). When measured throughout the menstrual cycle, plasma concentrations of inhibin B rise rapidly on the day after the intercycle FSH rise, remain elevated for a few days, then fall progressively during the remainder of the follicular phase. After a short-lived peak following the ovulatory gonadotropin surge, inhibin B falls to a low concentration during the luteal phase. In contrast, inhibin A concentrations rise only in the later part of the follicular phase and are maximal during the midluteal phase (Fig. 7).³⁴ These different patterns of circulating inhibin B and inhibin A during the human menstrual cycle suggest different physiologic roles (see later).

The inhibins influence FSH release specifically through their own negative feedback loop.³⁵ This negative feedback loop, however, functions at a slower rate than that of the above-mentioned ovarian steroid negative feedback loop which is activated within minutes, and is directed mainly at the pituitary gland. It is believed that the decline in FSH after its peak in the early follicular phase of the normal cycle results from a negative feedback action of inhibin B at the pituitary level. This action has been shown only under experimental conditions, however, for example after antiestrogen therapy in the midfollicular phase.³⁶ At menopause or in premature ovarian failure, data show a decreased secretion of inhibin with reproductive aging,³⁷ suggesting that inhibin B negative feedback may be an important factor controlling the early monotropic increase in FSH with aging.^{38, 39} Other dimers of the β subunit, called activins, have also been reported to play a role in FSH release.³⁵

The follicular phase

CYCLIC RECRUITMENT OF A COHORT

Although mechanisms of initial follicular growth in the ovary are difficult to investigate because of its protracted process, there is good evidence that dormant primordial follicles are recruited continuously into growing, a process that transforms them into primary, secondary, and early antral follicles, at which stage they become atretic. This process is initiated already before birth and continues until menopause and seems to proceed under the control of still unknown local ovarian factors, with gonadotropins perhaps playing a minor modulatory role.⁴⁰ After puberty and in each menstrual cycle, one cohort of early antral follicles is rescued from going into the atretic process and recruited for further growth. This 'cyclic' antral follicle recruitment requires an appropriate signal from the pituitary gland (Fig. 8).⁴¹ This signal is FSH as represented by its small but selective increase starting at the end of the preceding luteal phase (see Fig. 1).⁴² The size of the cohort is dictated by the amount of FSH present at recruitment: a greater sized cohort can be recruited by increasing FSH amounts, as shown in gonadotropin-stimulated cycles in which hormones usually are administered at supraphysiologic levels.

FSH stimulates mitosis and activates aromatase production in granulosa cells, allowing for growth and increased local estradiol production. Data also show that FSH is the major stimulus for inhibin B secretion, which increases at that time.⁴³ Experimental treatment with a GnRH antagonist at the end of the luteal phase, which abolishes the early follicular phase rise in FSH, also prevents the associated increase in inhibin B, whereas the same treatment followed by exogenous FSH restores the secretion of inhibin B.⁴⁴ The measurement of circulating inhibin B levels in the early follicular phase or in the early days of a stimulated cycle provides an early indicator of the number of recruited follicles and of their activity.

SELECTION OF A DOMINANT FOLLICLE

Although several follicles are recruited in the early follicular phase as part of the cohort, in mono-ovular species usually only one continues to grow into a mature preovulatory follicle.⁴⁵ This is referred to as the selection process, which occurs early in the midfollicular phase.⁴⁶ Experimental cauterization of the largest follicle at that time in the monkey results in a delay in the midcycle gonadotropin surge, a reflection of the absence of any surrogate follicle able to immediately take the place of the destroyed selected follicle and of the need to start anew and recruit another cohort of follicles.

The precise mechanism by which one follicle of the cohort is selected in primates remains to be elucidated. The selected dominant follicle has a distinguishable structure, however, in terms of its cellular development and especially of its vascularization it possesses a denser microvascular network than that of lesser developed follicles.⁴⁷ Experimental evidence has shown an active angiogenic process in this follicle destined for ovulation, suggesting an important role for regulators of angiogenesis in cyclic folliculogenesis. Vascular endothelial growth factor (VEGF), a major angiogenic factor, is detected in the developing follicle, with the most intense signal in the mature follicle.⁴⁸ VEGF activity is well correlated with proliferating activity markers in vascular endothelial cells and is a sign of active vasculogenesis.⁴⁹ Experimental data in nonhuman primates showed that VEGF receptor inactivation in the early follicular phase results in a rapid decrease in inhibin B secretion, suggesting an arrest in the development of the cohort of recruited antral follicles and in a delay in the rise in estradiol (Fig. 9).⁵⁰ These data directly show that normal follicular development cannot occur in the absence of adequate VEGF stimulation to ensure proper local angiogenesis. Data in the literature also suggest a role for the gonadotropins in VEGF production by the follicle. Studies with monkey granulosa cells show that not only large amounts of gonadotropins representative of the midcycle ovulatory surge but also smaller amounts more typical of tonic secretion both enhance local VEGF production.⁵¹

GROWTH OF THE DOMINANT FOLLICLE

The morphologic hallmark of the selected follicle is the acquisition of LH receptors in response to FSH action⁵² and the differentiation of an endocrinologically active theca layer capable of synthesizing androgens in response to LH stimulation. These androgens are aromatized to estradiol in the granulosa cell layer, which contains aromatase but which itself does not possess the full complement of steroid biosynthetic enzymes required for the synthesis of androgens.⁵³ Serum estradiol levels begin to rise as a result of the emergence of the dominant follicle. The rising estradiol, through the negative feedback loop, suppresses FSH levels (see Fig. 1) to concentrations that are too low to sustain maturation of the other follicles in the cohort with the consequence that these undergo final atresia.⁵⁴ This process can be prevented experimentally by neutralizing the rising estradiol levels or by providing a moderate and continued elevation of FSH during the midfollicular phase in these instances, the physiologic process of single follicle dominance is interfered with, and ongoing growth of multiple follicles is fostered.⁵⁵

By the midfollicular phase, there is an incremental increase in LH receptors, in aromatizable androgens, and in estrogens, which induces FSH receptors in the granulosa cells of the dominant follicle in a self-propagating mechanism. The developing dominant follicle generates its own estradiol microenvironment, which promotes further granulosa cell growth directly through its mitogenic activity or indirectly through the stimulation of local growth factors, such as insulin growth factor, members of the transforming growth factor superfamily (inhibins, activins), and others.^{56, 57}

The pattern of pulsatile GnRH-driven LH secretion changes dynamically across the human menstrual cycle. During the follicular phase, a pulse frequency of about 1 pulse/90 minutes is the norm. Adherence to a specific regimen of pulse amplitude and frequency is crucial for a normal follicular phase, menstrual cyclicity, and reproductive function, and derangements of episodic LH secretion are associated with reduced rates of ovulation in humans and nonhuman primates.^{58, 59, 60}

LATE FOLLICULAR PHASE

In the late follicular phase, the diameter of the selected follicle increases exponentially to a final size of 15–20 mm, and as a result secretion of estradiol increases exponentially. Vascular development in the dominant follicle plays a crucial role in this process because short-term inhibition of angiogenesis after anti-VEGF antibody administration during the later growth phase of the dominant follicle interferes with its normal development, interrupts the characteristic rise in estradiol secretion, and results in a significant lengthening of the follicular phase.⁶¹ Under thess conditions, access to peripheral factors needed to support follicle growth, such as the gonadotropins, is most probably limited.

The increased estrogen milieu in the late follicular phase modifies the genital tract.⁶² The glandular endometrium proliferates, the characteristics of cervical mucus change (increased secretion, decreased viscosity, and increased pH), and cornification of the vaginal epithelium occurs.

As the dominant follicle approaches maturity, estradiol secretion reaches its peak. This peak acts as the crucial ovarian signal that triggers the ovulatory gonadotropin surge. Experimental neutralization of this estradiol signal results in a suppression of the gonadotropin surge.⁶³ In humans, an LH surge can be induced experimentally during the early follicular phase after the administration of estradiol in amounts mimicking those seen in the late follicular phase (see Fig. 5B). The process by which estrogen transients stimulate the release of LH and FSH is referred to as the positive estradiol feedback loop and represents the crucial process that synchronizes follicle maturity and ovulation.

Whether gonadotropins are released after an estradiol challenge depends on the strength and duration of the estrogen signal. Minimal requirements are an increase in circulating estradiol concentrations to a threshold level close to that seen spontaneously in the late follicular phase and for a period of at least 34 hours. Subthreshold (even maintained for a prolonged period) or short-lived (even larger than threshold) rises in estradiol levels do not elicit an LH surge experimentally.^{64, 65}

As shown in nonhuman primates and in sheep, the preovulatory gonadotropin surge is preceded by a dramatic and sustained rise in GnRH (Fig. 10).^{66, 67, 68} Increased GnRH secretion drives the preovulatory LH surge in a dose-dependent fashion. These data suggest a substantial central action of estradiol on the hypothalamus, possibly on different neuronal cell populations than those modulating the estradiol negative feedback loop.⁶⁹ Experimental evidence indicates that the preovulatory LH surge depends on GnRH stimulation throughout its entire course, but that the GnRH surge persists many hours beyond the termination of the LH surge (Fig. 11).⁷⁰ The LH surge terminates even though there seems to be no change in the biologic activity of GnRH.⁷¹ The functional significance of this excess of GnRH to midcycle events remains to be determined. Suggesting a similar role of GnRH in the induction of the preovulatory gonadotropin surge in humans is the observation that administration of a GnRH antagonist Nal-Glu acutely inhibits the LH surge and ovulation.⁷² Because LH responses to identical GnRH stimuli increase during the follicular phase in parallel with increasing estradiol concentrations, estradiol also may act to augment the gonadotrope’s responsiveness to GnRH.⁷³ It is logical to postulate that during the normal menstrual cycle, central and pituitary sites respond to the estradiol positive feedback loop signal to ensure a timely gonadotropin surge.

Although progesterone secretion is minimal during the follicular phase, there is a small but significant rise in this hormone at the time of initiation of the gonadotropin surge (see below). In low concentrations, progesterone facilitates LH release and it is believed that this increase in progesterone may be required for the full expression of the gonadotropin surge because administration of a progesterone antagonist at midcycle in humans results in a delay or in the abolition of the gonadotropin surge.^{74, 75}

OVULATORY PERIOD

The estradiol-induced gonadotropin surge initiates a chain of events that includes a series of finely orchestrated biochemical events, many of which are still poorly understood, and that culminates in follicular rupture and ovulation. Hormonally, drastic changes in ovarian steroid profiles occur after alterations of enzymatic activity. The large increase in LH inhibits androgen production, and as a result estradiol concentrations decrease drastically from the preovulatory peak. Granulosa cells become 'luteinized', and consequently a small preovulatory rise in progesterone occurs within one hour of the LH surge (see Fig. 1).

Mechanically, ovulation consists of a rapid follicular enlargement with subsequent protrusion of the follicle from the ovarian surface. About 24–36 hours after the initiation of the gonadotropin surge or 18 hours after the gonadotropin peak, follicle rupture results in the expulsion of an oocyte-cumulus complex. Rupture does not seem to be caused by an increase in intrafollicular pressure; rather the LH surge induces an increase in follicular volume, which is related to an increase in follicular blood flow, a decrease in vascular permeability, and subsequent changes in the properties of the follicular wall.⁷⁶ These changes in vascular function most probably reflect the large local concentrations of angiogenic factors in the mature follicle and a functional VEGF system.⁷⁷ The LH surge stimulates a cascade of ovarian gene expression in the preovulatory follicle⁷⁸ resulting in an acute inflammatory-like process which generates an increase in proteolytic enzymes, such as plasminogen activator, plasmin, and matrix metalloproteinases, which bring about the degradation of the perifollicular matrix.⁷⁹ Pharmacologic blockage of any of these enzymes results in a reduction in the ovulation rate. The periovulatory follicle produces prostaglandins in response to the LH surge, and these also seem to be crucial to follicle rupture, although their direct role in primates remains to be shown.⁸⁰ A paracrine role of progesterone in the mediation of LH effects on follicular rupture has been suggested in lower species.⁸¹

The luteal phase

Shortly after the ovulatory gonadotropin surge, the granulosa cells of the ovulated dominant follicle fold, the basal lamina, which separates the granulosa and theca layers, is disrupted, and capillaries from the theca interna invade the granulosa layer (which until now had been avascular) to form such an extensive capillary network that each steroidogenic cell is now close to microvascular elements. Thus, after ovulation, a new ovarian structure emerges, the corpus luteum. New key steroidogenic enzymes are activated so that the hallmark of the human corpus luteum is its secretion of both progesterone and estradiol (see Fig. 1). Significantly, the passage of granulosa and theca cells of the now defunct dominant follicle into progesterone secreting cells is a terminal cell differentiation in that it is associated with an arrest in cell proliferation. The corpus luteum also secretes significant amounts of inhibin A (see Fig. 8).

The mammalian corpus luteum is an exceptionally dynamic organ in which growth and development occur rapidly. As for the follicle, vascular growth plays a central role in this process. Angiogenic factors, such as VEGF, are present in high quantity in the forming and developing corpus luteum.^{82, 83} In monkeys, experimental treatments that interfere with normal VEGF activity in the early and mid luteal phase suppress the intense luteal endothelial proliferation and vascular development; as a result, luteal function is compromised, as indicated by a marked fall in plasma progesterone levels.^{84, 85} It is well documented that corpus luteum function depends primarily on pituitary LH secretion throughout the luteal phase.^{86, 87, 88} Studies in hypophysectomized women, in whom ovulation was induced by LH treatment, indicate that continuous administration of small amounts of LH is essential to maintain the viability of the corpus luteum.³⁰ GnRH antagonist treatment readily disrupts luteal cell morphology and suppresses plasma progesterone.⁸⁹ The above experimental results indicate that LH and its support of local VEGF are as crucial to corpus luteum function as they are to follicular function.

Progesterone dominance in the luteal phase results in a significant decrease in GnRH/LH pulse frequency throughout this stage of the cycle (see Fig. 7).⁹⁰ There is good experimental evidence to indicate that this is an effect by progesterone which is mediated by central opioid peptides, because administration of competitive antagonists to endogenous opiates, such as naloxone, is particularly effective in increasing LH pulse frequency (Fig. 12B).^{91, 92, 93, 94, 95} A relevant endogenous opiate known to inhibit pulsatile LH secretion is β-endorphin. Its neuronal cell bodies are preferentially concentrated in the arcuate nucleus, an area known to be involved in the control of gonadotropin secretion. β-Endorphin release from the hypothalamus reflects the ovarian endocrine milieu. In the absence of significant concentrations of ovarian steroids, such as after ovariectomy or at menstruation, β-endorphin release is lowest, whereas it is highest in the presence of estradiol and progesterone, such as during the luteal phase (Fig. 12A).^{96, 97} The naloxone test is an indirect indicator of central opiate activity.

Progesterone also affects the hypothalamic thermoregulatory center so that an increase in basal body temperature (BBT) accompanies increased progesterone secretion during the luteal phase, giving rise to the typical BBT curve of the ovulatory menstrual cycle. Progesterone dominance during this phase of the menstrual cycle also modifies the genital tract in preparation for possible implantation of a fertilized ovum; there is increased secretory activity by the endometrial glands and changes in the characteristics of the cervical mucus, which is now thick and viscous. At its large luteal concentrations, progesterone also inhibits the estradiol positive feedback loop⁹⁸ and gametogenic follicle growth.⁹⁹

In primates, the life span of the corpus luteum is limited to a period of about 14 days. According to biochemical and histologic criteria, the corpus luteum reaches maturity 8–9 days after ovulation, after which time its secretory capability declines. Midway through the luteal phase, levels of estradiol and progesterone decrease, and menstruation follows ovulation by 13–15 days. Structural luteolysis is a complex process responsible for the elimination of the corpus luteum. Steroidogenic luteal cells undergo characteristic degenerative changes, with intense cytoplasmic vacuolization and invasion by macrophages followed by a perimenstrual apoptotic wave.¹⁰⁰ The factors responsible for luteolysis in primates remain to be identified.¹⁰¹ It is now believed that regression of the corpus luteum may be related to an alteration in age-dependent luteal cell responsiveness and is dictated by various luteotropic and luteolytic agents, the existence and dynamics of which have not been investigated. Only rapidly rising concentrations of chorionic gonadotropin (hCG) can rescue the corpus luteum. Although uterine prostaglandin F_2α seems to be an important luteolytic signal in nonprimate species, the primate uterus is not the source of luteolytic agents because hysterectomy does not result in a prolonged luteal phase.¹⁰¹

After the dramatic decrease in estradiol and progesterone secretion at the end of the luteal phase, there is a characteristic divergence in the ratio of the two gonadotropins, now favoring a specific rise in FSH (see Fig. 2). The precise reason for the increase in the FSH:LH ratio at the end of the cycle remains to be determined, but there are several possibilities, all of which may play a role to a certain degree. Because FSH seems to be slightly more sensitive to the estradiol negative feedback loop than LH,¹⁰² a rise in FSH may be the result of the rapid decline in estradiol at that time. The rise in FSH also may reflect differential effects of GnRH pulse frequency on the synthesis of LH and FSH, as a lower pulse frequency favors FSH β subunit synthesis over that of the LH subunit, and a larger pool of FSH would be available for release at the end of this period (see above). It also has been speculated that the accompanying fall in inhibin A (from a peak in the mid luteal phase)¹⁰³ may play a role in initiating the intercycle FSH rise. The increase in the FSH:LH ratio heralds the new cycle and the recruitment of a new cohort of follicles.

As outlined in the preceding sections, the normal ovulatory menstrual cycle requires a remarkable coordination between all components of the hypothalamic-pituitary-gonadal axis and other target tissues. The ovarian steroids, through their positive and negative feedback loops, play a crucial role by regulating gonadotropin levels. Cyclic dysfunction may be related to abnormalities in estradiol function or may reflect external influences, which, for example, may interfere with the normal function of the GnRH pulse generator. It may include a range of changes from 'qualitative' changes in the cycle, such as in the inadequate luteal phase syndrome, to a cessation of cyclic activity, such as in amenorrhea.

Abnormalities of the hypothalamic-pituitary unit

A properly active GnRH pulse generator is essential for normal gonadotropin release and for a normal ovulatory menstrual cycle to occur. Conditions that prevent or interfere with the function of the pulse generator disrupt the pituitary-ovarian axis and the cycle. Absence of GnRH release, such as observed in Kallmann’s syndrome, in which the migration of the GnRH neuron is not completed and contact with the hypophyseal portal veins is not established (see earlier),⁸ leads to idiopathic hypogonadotropic hypogonadism and delayed puberty. This primary hypothalamic amenorrhea syndrome also is characterized by anosmia, probably reflecting the initial defect in paraolfactory areas.

Frequent causes of cyclic dysfunction are related to lifestyle variables, such as psychogenic stress, and exercise-related or diet-related causes that affect hypothalamic function. The resultant condition frequently is referred to as functional hypothalamic amenorrhea or functional hypothalamic chronic anovulation.¹⁰⁴ The syndrome is characterized by a significant reduction in the activity of the GnRH pulse generator resulting in a decrease in the frequency of pulsatile LH release. Because adherence to a specific regimen of GnRH and LH pulse frequency is crucial for normal menstrual cyclicity and reproductive function (see earlier), such abnormal episodic LH secretion can lead to anovulation.⁵⁸ The degree of inhibition of the GnRH pulse generator determines whether the menstrual cycle is disturbed only slightly or whether the subject is anovulatory. One remarkable aspect of this syndrome is, however, that in most cases, it is readily reversible upon removal of the cause.

Psychological or environmental stress has long been reported to produce chronic amenorrhea.^{104, 105, 106} The observation that in this situation cortisol secretion is higher than normal suggests an association between increased hypothalamic-pituitary-adrenal (HPA) axis activity and reduced GnRH drive.¹⁰⁴ Experimental data in nonhuman primates support the concept that stress-induced functional hypothalamic amenorrhea develops in response to alterations in hypothalamic function.^{105, 106} A primary role for corticotropin-releasing hormone (CRH), the principal neurohormone of the HPA axis, in inhibiting the GnRH pulse generator is now well known. For instance, the acute inhibitory effects on pulsatile LH secretion of an immune stress challenge known to activate the HPA axis can be prevented by the administration of a CRH antagonist (Fig. 13).¹⁰⁷ Vasopressin, a peptide frequently co-packaged with CRH within CRH-containing secretory granules and co-released with CRH in stress, especially in chronic stress,^{108, 109} also can inhibit pulsatile LH secretion, and administration of a vasopressin antagonist also reverts the acute inhibitory effects of an immune stress challenge on LH.¹¹⁰ Although CRH or vasopressin administration increases adrenocorticotropic hormone and cortisol secretion, the ensuing acute gonadotropin inhibition is unrelated to these hormones in the primate. LH inhibition after CRH is observed in adrenalectomized monkeys and cannot be mimicked by the infusion of adrenocorticotropic hormone (ACTH) or cortisol, suggesting a central site of action of the HPA neurohormones.^{111, 112} This action is modulated by central endogenous opiates.^{113, 114} A simplified scheme of the central response to stress and the inhibition of the HPA axis is illustrated in Fig. 14.

Energetics and reproduction are linked¹¹⁵ and dietary restriction may suppress pulsatile LH release and affect the menstrual cycle, suggesting that the GnRH pulse generator is modulated by metabolic fuels and energy-related peptides.^{116, 117, 118, 119, 120, 121} The exact metabolic clues that signal the GnRH pulse generator of subtle changes in the metabolic status are presently being investigated. Studies in monkeys indicate that the clue depends on caloric intake but not on changes in body mass or composition and not on intake of a specific nutrient or in plasma glucose or insulin concentrations.¹²² A role for energy-related peptides in these processes is certainly called for. For example, data on the adipose tissue-derived hormone leptin¹²³ indicate that leptin levels decrease with fasting or undernutrition in parallel with decreased gonadotropins.¹²⁴ Observations in animals suggest that leptin may function as a metabolic signal to the neuroendocrine reproductive system and that under conditions of inadequate energy reserves, low leptin levels may act as a metabolic 'gate' to inhibit the neuroendocrine reproductive axis.¹²⁵ In support of this statement is the observation that leptin treatment in fasted animals prevents the decrease in LH pulse frequency that is seen after fasting, indicating that this hormone conveys information about nutrition to mechanisms controlling neuroendocrine function.¹²⁶ Ghrelin levels, another energy-related peptide derived from the stomach,¹²⁷ have been shown to increase during negative energy balance¹²⁸ while an acute ghrelin infusion significantly decreases LH pulse frequency in the monkey.¹²⁹

The extreme prototype of nutrition-related functional hypothalamic amenorrhea is the anorexia nervosa syndrome, which presents as a classic triad of amenorrhea, weight loss (sometimes to emaciation), and behavioral changes. The syndrome is characterized by a hypogonadotropic state (although FSH may be less affected than LH) and, in extreme cases, a regression to a prepubertal pattern of LH secretion, with hypoestrogenism.^{104, 116}  With recovery, LH pulsatility returns, first mimicking the pubertal pattern characterized by a nocturnal increase in pulsatility, then an adult pattern. High ghrelin levels are reported in reported in this syndrome.¹³⁰ Other syndromes associated with weight loss and altered reproductive function include bulimia, weight loss of 10–15% below ideal body weight, and various diets.¹¹⁶

GnRH pulse activity also may be affected by strenuous exercise, including jogging and athletics.^{131, 132} Depending on the intensity of the exercise, cyclic abnormalities may progress to a hypoestrogenic hypogonadotropic pattern. Physical activity alone may not be sufficient to cause amenorrhea, however, and other adjunct factors, such as low body weight, change in diet, weight loss, or the perceived stress of the exercise, may synergize to cause the effect.

Abnormalities of the hormonal feedback loops

Abnormal feedback conditions may result from an inappropriate extraglandular source of estrogen. In women, most of the circulating estrogens derive from estradiol secreted by the ovarian follicle or corpus luteum. Some estrogens also derive from peripheral conversion of androstenedione to estrone in adipose tissue and skin. In several pathologic conditions, such as congenital adrenal hyperplasia, Cushing’s syndrome, and androgen-producing tumors of either adrenal or ovarian origin, estrogen precursor availability is increased, and extraglandular production of estrogen may become excessive.¹³³ The androgen-to-estrogen conversion rate also increases with advancing age and in obese patients. Because this estrogen increase is acyclic and not under the control of gonadotropins, the original estradiol feedback signal may be distorted or masked, and cyclic irregularities may occur. In the polycystic ovary syndrome, which combines metabolic and hormonal disturbances, androgen production also is enhanced; LH pulse frequency, or in some cases LH pulse amplitude, is increased in greater than 60% of the patients.^{134, 135} Accelerated LH pulse frequency in women with polycystic ovary syndrome is not influenced by body mass index¹³⁶ and most probably represents a basic component of hypothalamic dysfunction.^{137, 138} Since an increased GnRH pulse frequency favors LH synthesis (see above), there is a resultant increased LH:FSH ratio. Abnormal steroid production and further disturbances in feedback regulation may occur.

In a few patients, abnormally low gonadotropin secretion has been attributed to an increase in the sensitivity of the hypothalamic GnRH pulse generator to the estradiol negative feedback loop.¹³⁹ Shifts in feedback sensitivity to estradiol have been suggested to form the basis for seasonality in the reproductive process in sheep: seasonal photic information is relayed to the pineal gland, which transduces the message into a hormonal signal in the form of melatonin. The circadian rhythm of melatonin, variable with the season, determines the capacity of the GnRH pulse generator to respond to the negative feedback action of estradiol, and LH pulse patterns inductive or suppressive of normal cyclicity are set.¹⁴⁰

During the normal menstrual cycle, the estradiol positive feedback loop ensures the coupling of follicular maturity with the pituitary gonadotropin surge (the stimulus to ovulation). Delayed estradiol increments or insensitivity of the hypothalamic-pituitary axis to the positive estrogen feedback loop (as may occur during the first pubertal cycles) may delay the LH surge beyond the time that ovulation is possible. Premature estradiol increments or increased hypothalamic-pituitary sensitivity to the estrogen stimulus may induce a premature LH surge, which may interfere with or arrest follicular maturation. Experimental evidence in monkeys and humans shows that certain stress stimuli, when occurring in the presence of midfollicular phase levels of estradiol, can stimulate rather than inhibit LH release.^{106, 141, 142} Thus, it may be that a 'stress' episode occurring during the follicular phase could affect the menstrual cycle by inducing a gonadotropin surge at an inappropriate time of the cycle, interfering with proper ovarian-pituitary signals.

Abnormalities of the pituitary

Defects in menstrual cyclicity may be caused by pituitary lesions, which may result in trophic hormone deficiency, or by pituitary tumors. A typical pituitary lesion is the one that occurs in conjunction with peripartum or postpartum hemorrhage and ischemic shock (Sheehan’s syndrome). Depending on the extent of trophic hormone deficiency, symptoms of gonadotropin, thyroid, or adrenal hormone secretory defects may coexist. The most common type of pituitary tumor disrupting the menstrual cycle is the prolactin-secreting adenoma causing the amenorrhea-galactorrhea syndrome.¹⁴³ Through central mechanisms that are not yet known, hyperprolactinemia suppresses the GnRH pulse generator and pulsatile LH and FSH release. Hypothalamic control of prolactin is predominantly inhibitory, and the hypothalamic factor responsible for controlling this hormone is the neurotransmitter dopamine. In this case, dopamine is released from high-density dopamine nerve endings in the median eminence into the hypothalamic-hypophyseal vessels. The functional aberrations in prolactin and in pulsatile LH and FSH secretion in this syndrome can be reversed readily by the administration of dopamine agonists.

Abnormalities of the corpus luteum

Another example of a frequent cyclic abnormality is the inadequate luteal phase syndrome, which is characterized by blunted progesterone secretion throughout the luteal phase.¹⁴⁴ In turn, the lack of progesterone interferes with the proper luteal endometrial sequence. The resultant abnormal secretory activity of the endometrium and other uterine deficiencies related to insufficient progesterone levels may prevent the normal implantation process and may result in infertility. Because the corpus luteum derives entirely from the follicle that has matured during the follicular phase of that cycle, abnormal events that occur in the early follicular phase may influence luteal function. Inappropriate patterns of FSH secretion early in the follicular phase may lead to deficiencies in the development of the graafian follicle and in estradiol secretion, which may result in the inappropriate luteinization of the granulosa cells and a smaller sized corpus luteum with less secretory capability.^{145, 146}

Relevant observations in experimental stress models in nonhuman primates suggest that the first clinical manifestation of cyclic dysfunction in an otherwise normal ovulatory primate subjected to stress usually occurs in the form of a significant reduction in integrated progesterone concentrations during the luteal phase, for example in non-human primates subjected to short inflammatory-like or psychogenic stress challenges.^{147, 148, 149} A similar prevalence of luteal defect in cycling women has been reported as a result of stress, exercise, or eating disorders.^{150, 151, 152, 153}