- Split View
-
Views
-
Cite
Cite
H.G. Burger, G.E. Hale, D.M. Robertson, L. Dennerstein, A review of hormonal changes during the menopausal transition: focus on findings from the Melbourne Women's Midlife Health Project, Human Reproduction Update, Volume 13, Issue 6, November/December 2007, Pages 559–565, https://doi.org/10.1093/humupd/dmm020
- Share Icon Share
Abstract
The menopause, defined as the permanent cessation of menstruation resulting from the loss of ovarian follicular activity, marks the end of natural female reproductive life. It is preceded by a period of menstrual cycle irregularity, the menopausal transition, which usually begins in the mid-40s and is conventionally divided into early and late phases. The endocrine changes, which underlie the transition, are predominantly the consequence of a marked decline in ovarian follicle numbers. The most significant changes include a decrease in early cycle inhibin B and in anti-Mullerian hormone (AMH) levels. The decline in inhibin B results in an increase in FSH, which appears to be an important factor in the maintenance of estradiol (E2) concentrations until late in reproductive life. In the post-menopause, FSH levels are markedly raised, E2 levels are low, whereas inhibin B and AMH are undetectable. The menopausal transition is a time of marked hormonal instability. The Melbourne Women's Midlife Health Project has been an extremely productive study in which it has been possible to describe longitudinal changes in hormone levels throughout the menopause transition and to separate the effects of hormone change from the effects of ageing on a number of endpoints. This review provides the background for an accompanying manuscript in which a novel approach to modelling the hormonal changes during the transition is described.
Introduction: endocrine regulation of ovarian function
With the exception of the mid-cycle ovulatory gonadotrophin surge, the hypothalamo-pituitary-ovarian axis is a classic endocrine closed loop feedback system, in which the gonadotrophins stimulate ovarian hormone production, which in turn exerts a negative feedback effect on the gonadotrophins, to maintain a regulated system. During mid-reproductive age, circulating pituitary FSH concentrations begin to increase about 4 days premenstrually, reach a mid-follicular phase peak, gradually fall prior to the mid-cycle surge and then decline to low levels during the luteal phase. LH levels, in contrast remain relatively constant throughout the cycle, except for a mid-cycle ovulatory surge. The concentration of LH falls significantly during the luteal phase under the influence of progesterone and estradiol (E2). The ovary secretes steroid (E2, progesterone and testosterone) and peptide hormones (the inhibins) under gonadotrophin control, and anti-Mullerian hormone (AMH), also called Mullerian inhibiting substance, independently of the gonadotrophins. During the follicular phase of the cycle, both E2 and inhibin B are stimulated by FSH and in turn regulate its secretion, while in the luteal phase E2 and progesterone are components of the negative feedback loop, and regulate FSH and LH. The role in gonadotrophin regulation of inhibin A, also produced by the corpus luteum, is not clear. The secretion of testosterone by the ovary is under the control of LH. Its levels are lowest during the perimenstrual phase, increase during the follicular phase, peak at mid cycle, then decrease during the luteal phase, when they are higher than during the follicular phase. There is little evidence for a direct negative regulatory feedback loop. Major consideration of female androgen secretion from the adrenals as well as the ovary is beyond the scope of this review.
The dual follicular phase negative feedback by E2 and inhibin B provides an elegant system, which operates during reproductive ageing and allows the preservation of E2 secretion into late reproductive life. Inhibin B is a product of follicular granulosa cells, as is E2. As discussed in more detail below, a selective fall in inhibin B, as follicle numbers in the ovary decline with increasing age, allows a selective rise in FSH, which is hypothesized to drive the granulosa cells within the developing dominant follicle or possibly an enlarged cohort of developing follicles to produce increasing amounts of E2. This review focuses on changes in FSH, E2, inhibins and briefly discusses testosterone, sex hormone binding globulin and AMH.
The Melbourne Women's Midlife Health Project
This was the first major longitudinal study of the experiences of women transitioning from late reproductive age, with continuing regular menses, through the final menstrual period (FMP) and beyond. It began with a cross-sectional survey of a randomly selected population of 2001 Australian-born Melbourne women, aged 45–55 years at the time of the initial interview (Dennerstein et al., 1993). Of these, 438 women who had menstruated within the preceding 3 months and were not using hormonal contraception or hormonal therapy, entered the longitudinal phase of the study, in which annual interviews were conducted in their homes, where early follicular phase blood samples were collected for subsequent hormonal and other analyses. Measurements of height and weight were recorded. Some aspects of the endocrine findings in this study are discussed in detail below. A summary of the major findings was published in 2004 (Guthrie et al., 2004). The accompanying paper (Dennerstein et al., 2007) presents a novel approach to modelling the longitudinal changes in FSH and E2.
Nomenclature of the stages of reproductive ageing
A WHO Scientific Group on Research on the Menopause proposed a series of relevant definitions in 1980 (Research on the Menopause, 1981). The term ‘natural menopause’ was defined as the permanent cessation of menstruation resulting from the loss of ovarian follicular activity, and natural menopause was recognised to have occurred after 12 consecutive months of amenorrhoea, for which no other obvious pathological or physiological cause could be determined. The term ‘menopausal transition’ was recommended to be reserved for that period of time before the FMP when variability in the menstrual cycle is increased. How such variability should be determined was not explicitly defined. These definitions were re-emphasized by the second WHO Scientific Group on Research on the Menopause (WHO Scientific Group, 1996). Guidelines for the classification of the stages of reproductive ageing were proposed in 2001 at the Stages of Reproductive Ageing Workshop (STRAW, Soules et al., 2001). Reproductive life was divided into three phases (early, peak and late) with late reproductive age being characterized by an elevated early-cycle FSH concentration in the setting of regular menstrual cycles.
The menopausal transition was divided into early and late stages with some ambiguity as to the definition of the onset of the early menopausal transition. In one section of the STRAW report, it is stated that ‘a woman's menstrual cycle remains regular in stage −2 (early menopause transition), but the duration changes by 7 days or more’. However, in the figure depicting the STRAW stageing system, stage −2 is characterized by ‘variable cycle length, ≥7 days different from normal. Many investigators have assumed that menstrual cycle variability is the hallmark of the early transition. For example, Metcalf et al. (1981a) in their classic studies, defined the onset of the transition or perimenopause on the basis of the first break in a pattern of regular menstrual cycles, presumably based on subject history. In the MWMHP, self reported cycle irregularity was also used as a marker of its onset (Dennerstein et al., 1996). Other more specific definitions included those of Mitchell and colleagues (2000) who proposed the occurrence of >6 days difference in cycle length. Gracia et al. (2005) used a difference of at least 7 days in cycle length but proposed a late premenopause stage where only one cycle showed this change in length. The early transition was then defined as two or more cycles with a difference in length of ≥7 days. Subsequently, on the basis of the work by Taffe and colleagues (2002), Burger and colleagues (2005) suggested that the early menopause transition began on 'the date of the first of more than two cycles in any consecutive series of 10, the length of which lay outside the normal reproductive range of 23–35 days'.
The late menopausal transition was defined in STRAW as the occurrence of two skipped cycles and an interval of amenorrhoea of 60 days or more. In the MWMHP, a slightly different nomenclature was employed (Dennerstein et al., 1996), the late menopausal transition being defined by the occurrence of 3–11 months amenorrhoea at the time a subject was interviewed. The STRAW definition for the late menopause transition was recently reviewed in a reanalysis called ReSTAGE (Harlow et al., 2006). Data from the MWMHP was included in this reanalysis, along with data from several large studies including the Tremin (Treloar et al., 1967), Study of Women's Health across the Nation (SWHN, Randolph et al., 2003) and Seattle MWHS (SMWHS, Mitchell et al., 2000) studies as well as the MWMHP. It was concluded by the collaborators (Harlow et al., 2006) that the most appropriate definition for the onset of the late menopause transition was 60 or more days (rather than 90 days) of amenorrhoea. In the present review, we have retained the definitions originally adopted in MWMHP, as these are the basis of the publications arising from that study.
The endocrinology of late reproductive ageing
As noted above, the STRAW definition of late reproductive ageing is the presence of an early cycle FSH concentration variably higher than levels observed in young reproductive-aged women, while menstrual cycles remain regular. Elevated early follicular phase serum or urinary FSH levels have been recognized as indicative of late reproductive ageing by many investigators (Sherman et al., 1976; Reyes et al., 1977; Metcalf, 1988; Lee et al., 1988; MacNaughton et al., 1992; Fitzgerald et al., 1994; Klein et al., 1996). Elevated FSH may precede the onset of cycle irregularity and hence the menopausal transition by 3–10 years. The rise has been associated with a decrease in follicular phase and menstrual cycle length. Raised FSH but with regular cycles is impossible to detect without serial detailed endocrine measurements which are impractical clinically on a wide scale. Therefore, the only clinical sign may be cycle length. In women of late reproductive age, the interval between day 1 of the cycle and the follicular phase rise in E2 (termed the ‘lag phase’), has been shown to be inversely correlated to the early follicular phase rise in FSH levels, i.e. the higher the early follicular phase rise in FSH, the shorter the ‘lag phase’ (Miro et al., 2004a). The elevated early cycle FSH levels have been reported to occur as early as 30 years of age in infertility clinic patients (Ahmed Ebbiary et al., 1994) and by 40 years of age in randomly selected healthy women (Lenton et al., 1988; Burger et al., 2000a). They do not, however, occur in all women of advanced chronological age who continue to cycle regularly (Burger et al., 2000a) and are not consistent between cycles in the same woman (Burger, 1994). Single early cycle FSH measurements thus have a limited capacity to predict reproductive status reliably, and there is a major degree of overlap between levels in regularly cycling women of the same age over 40 years.
While the FSH elevations in some women of advancing reproductive age were once thought to be secondary to a fall in E2 levels, many studies have now shown increased FSH levels are associated with normal or even higher than normal E2 or urinary estrogen levels (Reyes et al., 1977; Metcalf et al., 1981b; Lee et al., 1988; Lenton et al., 1988; Shideler et al., 1989; Burger, 1994; Fitzgerald et al., 1994; Klein et al., 1996; Burger et al., 2002). Significantly elevated mean cycle levels of E2 have been observed in regularly cycling women even up to age 55 (Klein et al., 2004). It is plausible that elevated FSH levels play a role in maintaining or even increasing E2 levels. As noted above the elevated levels of follicular phase FSH are associated with decreased levels of inhibin B with several studies demonstrating raised FSH and lowered inhibin B levels in women of advanced reproductive age compared with young women (Klein et al., 1996; Danforth et al., 1998; Reame et al., 1998; Soules et al., 1998; Santoro et al., 1999; Welt et al., 1999; Burger et al., 2000a; Muttukrishna et al., 2000). Inhibin B levels are correlated with the number of developing antral follicles seen on ultrasonography during the early follicular phase (Tinkanen et al., 2001) and the levels fall in parallel with the number of ovarian antral follicles (Danforth et al., 1998). No specific concentration of inhibin B has been shown to be diagnostically discriminatory.
In contrast to inhibin B, inhibin A is a product of the dominant follicle and levels do not change appreciably until dominant follicles are no longer being generated in the ovary (Burger et al., 1998).
Recently, there has been growing interest in the usefulness of serum AMH as a marker of reproductive ageing and decline in reproductive capacity (Van Rooij et al., 2004, Van Rooij et al., 2005; LaMarca and Volpe, 2006). Van Rooij and colleagues have provided evidence that AMH is a powerful predictor of the early transition with the combination of AMH and inhibin B measurements improving that prediction. So far, there are no longitudinal data for AMH on the predictability of the late menopause transition or the final menses. Another aspect of reproductive ageing endocrinology is the question of luteal function. Some studies have suggested that advanced reproductive age is associated with lowered luteal phase levels of progesterone (Reyes et al., 1977; Ballinger et al., 1987), or urinary pregnanediol glucuronide excretion (Santoro et al., 1996), but this is not a consistent finding (Sherman et al., 1976; Lee et al., 1988; Fitzgerald et al., 1994; Landgren et al., 2004) and unfortunately, the MWMHP has not provided any data on this.
Endocrine characteristics of the menopausal transition and menopause
Although some neuroendocrine changes may contribute to the onset of the menopausal transition (Wise et al., 2002), the major factor is generally thought to be the decline in the numbers of ovarian follicles to a critically low level. Richardson et al., (1987) described follicle numbers in three groups of older women, those who were continuing to cycle regularly, those who were in the menopausal transition and those who were post-menopausal. In the menopause transition group, the mean number of primordial follicles per ovary was ∼100, whereas in the post-menopausal group the ovaries were virtually devoid of follicles.
The first major description of the circulating hormonal concentrations during the menopause transition was by Sherman and Korenman (1975), involving six women followed in detail up to and including final menses. The features included a monotropic rise in FSH levels, evidence of continued folliculogenesis and ovulation up to the FMP and periods of hypoestrogenemia concomitant with large FSH rises. These investigators first postulated that the FSH increase resulted from loss of inhibin restraint, though inhibin had not been characterized at that time.
Metcalf (1988) summarized the extensive studies on urinary hormonal excretion patterns observed by their group in 31 perimenopausal women who collected weekly overnight urine samples in 124 cycles. Median cycle length was 29 days, range 18–260. Only 52% met the criteria used to identify an ovulatory cycle. Ovulatory failure occurred in 80% of the 48 cycles longer than 40 days, compared with 20% of shorter cycles. In four of the women, the last cycle prior to FMP was ovulatory and ovulatory cycles were seen at all stages during the transition (as noted also more recently by Burger et al., 2005). There were marked variations both in gonadotrophins and steroid excretion and a classification based on pregnanediol and gonadotropin excretion was proposed (Metcalf et al., 1981a). The most characteristic feature was the sporadic appearance of persistently raised levels of urinary FSH and LH. There was no evidence of a gradual decline in ovarian function—the changes were unpredictable. In eight women studied across the time of FMP, the early post-menopausal hormone changes could not be distinguished from those in the long anovulatory cycles of the transition. The authors concluded than an endometrial rather than a hormonal event may determine the timing of final menses (Metcalf et al., 1982).
The MWMHP has provided a substantial amount of the existing data on changes in follicular phase blood hormone concentrations with the onset of the menopausal transition (Burger et al., 2002). A profound fall in the follicular phase concentrations of inhibin B appears to be the first endocrine marker of the early transition (Burger et al., 1998) with FSH levels being slightly raised, but not statistically significantly higher than those in women continuing to cycle regularly (Fig. 1). At this stage no change in inhibin A or E2 in the early follicular phase was seen. However, once women reached the late transition with >3 months of amenorrhoea, marked falls in E2 and inhibin A, together with significant elevations in FSH were observed. A high degree of both within subject and between subject variation in gonadotrophin and sex steroid levels is seen, when individual women are studied. Gracia et al., (2005) have reported the endocrine differences seen in regularly cycling women who entered the early menopause transition (defined as discussed earlier). Significant falls in inhibin B and elevations in FSH were observed in the subjects, who were aged 35–47 at recruitment, but E2 levels did not change. It should be noted that these data were group observations and no specific change in hormone levels could be used to categorize an individual woman's status. Thus, the hallmark endocrine change associated with early changes in menstrual cycle regularity appears to be a fall in inhibin B and a moderate rise in FSH (Burger et al. 1998; Gracia et al., 2005).
As noted above, as menstrual cycle irregularity occurs, the frequency of anovulatory cycles increases. In a small longitudinal study of late reproductive ageing, Landgren et al., (2004) noted that in the last ten cycles prior to FMP, 62% were anovulatory (though some cycles classified as anovulatory may have been long cycles with late ovulation), and only 38% were ovulatory. Anovulatory cycles were observed to occur mainly within the last 30 cycles prior to FMP and to be uncommon prior to this. While little change in hormone concentrations was observed in ovulatory cycles, anovulatory cycles tended to be associated with low E2 and high FSH concentrations, particularly in the initial phases of the cycle, as noted by Sherman and Korenman (1975) and Metcalf and colleagues (1988).
A number of studies of urinary estrogen excretion during the menopause transition have also observed low levels of estrogen excretion in prolonged ovulatory cycles, during the ‘lag phase’ (Sherman and Korenman, 1975; Metcalf 1988; Shideler et al., 1989; O'Connor et al., 2001; Miro et al., 2004b). The ‘lag phase’ is associated with marked elevations in FSH and moderate elevations in LH excretion. While the correlation between the length of the ‘lag phase’ and early cycle FSH levels is inverse in late reproductive age, it is direct during the menopause transition, i.e. the higher the early cycle FSH level, the longer the ‘lag phase’ (Miro et al., 2004b), suggesting that, with low numbers of follicles the ovary is temporarily refractory to gonadotrophin stimulation. Once the follicular phase began in these elongated ovulatory cycles, estrogen excretion was significantly higher than normal, with a positive correlation occurring between the length of the ‘lag phase’ and the amount of estrogen excretion during the luteal phase. This amount was correlated to follicular LH rather than FSH levels. It is clear that distinctions must be made between ovulatory and anovulatory cycles to fully characterize the endocrine changes occurring during the transition, distinctions that were not possible in the MWMHP.
There is evidence that there is a change in hypothalamic-pituitary sensitivity to estrogen feedback in perimenopausal women. Van Look et al. (1977) studied nine perimenopausal women with dysfunctional uterine bleeding and observed a failure of LH response to endogenous and exogenous estrogen stimulation which would normally elicit a mid-cycle LH surge. Weiss et al. (2004) also noted, in ageing perimenopausal women participating in the SWAN study, the frequent occurrence of anovulatory cycles with estrogen peaks equivalent to those which elicit LH surges in younger women, again indicating failure of positive feedback. In some women, follicular phase estrogen levels failed to lower LH concentrations, as also normally occurs in younger women. There was thus evidence of failure both of estrogen positive and negative feedback, indicating a degree of hypothalamic–pituitary insensitivity.
Evidence for the occurrence of diminished luteal function during the transition was initially provided from studies of urinary pregnanediol excretion (Metcalf et al., 1981a). These authors used such changes as one of their criteria to classify cycles on the basis of their hormonal characteristics.
Few studies have measured serum progesterone during irregular ovulatory cycles in the menopause transition (Sherman et al., 1976; Welt et al., 1999). One study showed a marked decrease in luteal phase progesterone in three subjects who were studied several years apart (Welt et al., 1999). Prolonged ovulatory cycles have been observed to be characterized by abnormally low progesterone levels. Diminished luteal progesterone production may reflect diminished dominant follicle quality.
While menstrual blood loss generally remains unchanged during ovulatory cycles, menses associated with anovulatory cycles tend to be abnormal, often with prolonged spotting and/or heavy bleeding (Burger et al., 2005). However, ovulatory dysfunctional uterine bleeding has been reported to occur more commonly than anovulatory dysfunctional bleeding (Livingstone and Fraser, 2002).
When FSH and E2 are measured in relation to prospectively defined final menses as in the MWMHP, E2 levels can be demonstrated to have fallen to ∼50% of their premenopausal concentrations and FSH to have risen to 50% of its finally attained post-menopausal concentrations (Fig. 2; Burger et al., 2002).
It can be noted that no significant change in total serum testosterone concentrations occurs across the time of final menses (Fig. 3; Burger et al., 2000b). In the MWMHP, sex hormone binding globulin concentrations fell significantly across the transition and consequently free testosterone concentrations actually rose (Burger et al., 2000b).
To summarize these observations, a fall in follicle numbers with age reaches a critical level when inhibin B concentrations start to fall and FSH levels rise. Anovulatory cycles increase in frequency in the last 30 cycles prior to FMP with a late fall in E2 and inhibin A as FSH rises. Testosterone levels fall together with dehydroepiandrosterone sulphate (DHEAS) between the ages of 20 and 45 years (Davison et al., 2005), but testosterone shows little change in relation to the menopause transition and DHEAS continues to fall with age, whereas testosterone shows little change with age during the menopausal transition years (Burger et al., 2000b). There are no concentrations of E2 or of FSH which are specific to a particular stage of the transition-as emphasized, the endocrine hallmark of the transition is the extreme variability of such concentrations within and between women.
The mechanisms which lead to menstrual cycle irregularity presumably involve the critical numbers of follicles and temporary ovarian non-responsiveness to FSH stimulation. With the rise in FSH no ovarian response may occur for days or weeks, but ultimately a follicle may start to develop and may in fact hyper-respond with higher than normal E2 concentrations.
Future studies
Future studies of the endocrinology of the transition should aim to elucidate the types of hormonal change which characterize both abnormally short and abnormally long cycles, to describe the relationships between clinical features and these hormonal changes, and to determine which markers most reliably indicate significant decline in ovarian function and hence predict the occurrence of the FMP. Longitudinal studies with detailed data on hormonal changes, AMH concentrations and ultrasound characterization of the underlying follicular dynamics, though difficult to do, would further clarify the physiology of the transition. There is a need to mathematically characterise the hormonal changes of the transition which are clearly not linear, as presented in the accompanying paper (Dennerstein et al., 2007).
Acknowledgements
The assistance of the co-authors of the MWMHP publications is gratefully acknowledged. Jeana Thomas and Sue Elger provided excellent secretarial assistance. The work was supported by grants from the Victorian Health Promotion Foundation, the National Health and Medical Research Council of Australia and an unrestricted grant to the laboratory of HG Burger from Organon Australia Pty Ltd for the hormone assays.