Endocrinology of the Perimenopausal Woman

Natural menopause is traditionally defined as the time period that occurs after 12 consecutive months of amenorrhea. The average age of menopause is about 51 years. The menopausal transition refers to the time period before the onset of menopause, when changes in a woman's hormonal milieu are associated with irregular menstrual cycles and increased amenorrhea. In 2001, STRAW proposed a menopause nomenclature that divides the menopausal transition into two phases: the "early menopausal transition" in which cycle lengths vary more than 7 days from the typical length for that individual (or a woman skips a cycle completely but has at least one menstrual period within the past 3 months); and the "late menopausal transition", characterized by at least two skipped cycles and at least one period of amenorrhea exceeding 60 days.¹ The 'typical' transition lasts about 4 years and has its onset between ages 45 and 55 in 95% of cases.^{2, 3}

Smoking is the greatest independent risk factor for earlier menstrual irregularity and earlier menopause. Smoking causes an earlier perimenopause and menopause by about 1–2 years.^{4, 5} Another strong indicator for an earlier age at menopause is a maternal history of early menopause.⁶

Once a woman older than 45 has had 1 year of amenorrhea, she has less than a 10% likelihood of ever menstruating again.⁷ However, there is no clear-cut transition period from the pre- to the postmenopausal state. Cessation of ovulation occurs spontaneously at some point during the transition.

Regarding the hormonal milieu of women traversing the menopause, Metcalf and associates⁸ could not distinguish any hormonal changes between those found in the irregular cycles of the perimenopausal woman from those of the immediately postmenopausal woman, except that no progesterone was ever made after the final menstrual period. Thus, it was suggested that the point at which menstruation ceases during the menopause transition involves not only hormonal changes, but possibly also endometrial factors. In other words, aging may affect the ability of the endometrium to respond to estrogens. However, the preponderance of evidence suggests that changes are occurring in a woman's hormonal environment during the perimenopause.

During the perimenopause, there is an increase in the proportion of cycles that are anovulatory. However, the mechanisms responsible for perimenopausal anovulation remain unclear. The anovulatory cycles occurring during the perimenopause appear similar to those occurring in adolescents and may reflect an inability to produce a preovulatory surge of gonadotropins after exposure to estrogen, at least in some women.⁹ There may be central changes in the hypothalamic–pituitary axis that affect gonadotropin secretion and that may be caused by a relative hypothalamic–pituitary insensitivity to estrogen in association with reproductive aging.¹⁰ This has been suggested by the lack of response to an estradiol challenge with a luteinizing hormone surge in perimenopausal women with dysfunctional uterine bleeding.^{9, 11, 12} However, abnormalities in ovarian steroid or peptide secretion may also play a role.

During the perimenopause, ovarian function appears to be highly variable. Length of menses and the intermenstrual interval varies, and anovulatory cycles become more common. Hormone levels may fluctuate widely during this time, and as estrogen levels decrease, some of the inherent protective effects of estrogen on bone health and endothelial function may also decrease. Thus, the hormonal changes associated with aging may have both short and long term detrimental effects that must be recognized, addressed, and ameliorated when possible.

Many of the physiologic changes associated with menopause occur or begin before the last menstrual period¹³ and may be associated with somatic aging. Somatic aging is reflected by decreases in somatotropic axis function, adrenal androgen production, and continuous loss of bone mineral density after peak bone mass has been attained. Several hormonal systems have age-related changes that are concurrent with reproductive aging and there is evidence that the two processes may interact.⁹

Ovarian reserve is the most important factor mediating the pace of the menopausal transition. The ultimate causative factor for perimenopause and menopause is depletion of ovarian follicles. However, aging changes in non-ovarian tissues, such as the hypothalamus, pituitary, uterus, and ovary, may also contribute to the perimenopause. As of yet, however, none of these systems has been fully examined in detail in the human. Therefore, their contribution to the perimenopause and menopause remains unclear.

Changes in the Hormonal Milieu

During the perimenopause, midcycle estrogen concentrations have been observed to be normal or increased,^{14, 15, 16, 17} and levels of progesterone and androgens have been observed to be normal or decreased, independent of major changes in sex hormone-binding globulin.^{18, 19}

Changes in Gonadotropins

Sherman and associates^{15, 20} followed six women for several years, including the time of their last menstrual period, and noted: (1) a monotropic rise in follicle-stimulating hormone (FSH) that occurred even with normal menstrual cycles; (2) occasional anovulatory cycles; and (3) in one woman, a final menstrual period that occurred immediately after an ovulatory cycle.

The increase in FSH that occurs during the menopause transition has been attributed to a loss of ovarian inhibin B with aging.²¹ This relationship is supported by available immunoassay data;^{21, 22} however, a cause-and-effect relationship between the loss of inhibins (A or B) and a rise in FSH remains to be established definitively in the human.

Although FSH levels increase progressively with age, there is a great deal of overlap between phases of the menopausal transition and even in the postmenopause. Thus, although it may be useful to measure FSH for the clinical purpose of providing a fertility prognosis, it is a relatively poor predictor of the time of menopause for an individual woman.¹⁴ The age at which the FSH rise first appears may not necessarily correlate with menopause. Longitudinal studies imply that FSH increases progressively throughout life and a more distinct rise occurs as early as the early 40s in normal women.²³ Along with the elevation in FSH, there is a lesser, but still significant, rise in perimenstrual levels of luteinizing hormone (LH).^{24, 25} Thus, in comparison to midreproductive age women, women in the menopausal transition have higher levels of FSH and LH but not lower estrogen levels.^{14, 25} Therefore, elevated FSH concentrations in the early menopausal transition are most likely not due to low estrogen associated with follicle failure, but are likely to be caused by other factors, such as inhibin B. 

To date, a number of serum biomarkers of ovarian follicle reserve exist, but none are sufficiently sensitive to provide a prediction of menopause that exceeds what can be achieved by taking a careful clinical history. The marked variability of FSH and LH across a typical ovulatory menstrual cycle indicate that they are unlikely to provide a reliable estimate of the effect of age on ovarian function. It is difficult if not impossible to recognize early ovarian failure in the clinical setting using currently available markers – FSH, LH, inhibin A or B or Mullerian inhibiting substance (MIS) or antimullerian hormone (AMH).²⁶

Changes in the Estrogenic Environment

Hyperestrogenemia may be a feature of the early perimenopause, but immediately premenopausal cycles may have reduced estrogen.^{9, 14} An important but often clinically frustrating aspect of the perimenopause is that estradiol levels do not gradually decrease; instead, they fluctuate greatly around the normal range until menopause, when no more responsive follicles are available.²⁶ The anovulatory cycles often seen may be associated with elevated levels of estradiol.^{14, 17} Thus, as a woman ages, there is not a progressive downward spiral in the estrogenic milieu, but instead a "roller-coaster" in estrogen production.⁹ The perimenopausal fluctuations in estradiol may result from a combination of a reduced follicle pool and the aging ovary's reduced responsiveness to FSH. Thus, greater circulating amounts of FSH are needed to initiate folliculogenesis. Once commenced, the FSH rise may cause an overstimulation of the ovary and lead to relative hyperestrogenemia.^{9, 14}

After a woman's final menstrual period, progesterone is no longer produced, but for a brief time, small amounts of estrogen may still be produced. Metcalf and co-workers⁸ observed that although elevations in FSH and LH were common before the final menstrual cycle, episodes of significant estrogen production were not uncommon in the first year after the final menstruation.

Changes in the Progesterone Environment

Both normal^{15, 20} and decreased^{14, 16} levels of progesterone secreted by the corpus luteum have been observed in the perimenopause. Further clarification regarding perimenopausal progesterone levels would be clinically very useful, because if decreased levels of progesterone are associated with increased levels of estradiol, this may also predispose women to dysfunctional uterine bleeding and endometrial hyperplasia.

Changes in Growth Hormone and Insulin-like Growth Factor 1 Levels

Growth hormone (GH) is a pulsatile hormone released from the anterior pituitary under hypothalamic regulation by growth hormone-releasing hormone (GHRH) and inhibitory regulation by somatostatin. With aging, there is a decrease in GH secretion. It remains to be elucidated whether the decrease in GH is secondary to decreased GHRH, increased release of somatostatin, decreased pituitary sensitivity to GHRH, or a combination of these factors.⁹

Estrogen appears to play an important role in modulating GH secretion in women. There is a positive association between estrogen and GH concentrations. Thus, in a decreased estrogenic environment, such as in menopause, there is decreased GH secretion.²⁷

Age itself may be a more important factor affecting GH concentration than estrogen alone. Studies have shown that a decrease in somatotropic axis activity is detectable before any changes occur in menstrual cyclicity or decreased production of estradiol. Older, regularly cycling women (age 42–46) have lower daytime GH concentrations than younger, regularly cycling controls (age 19–34). This occurred in the older women despite higher estradiol levels on the day of sampling (compared with the younger controls) and overall normal parameters of gonadal hormones. The older reproductive-age women had twice the early follicular phase concentration of estradiol compared with the younger controls (mean ± standard error of the mean, 368 ± 51 vs. 167 ± 20 pmol/L).²⁸

Older reproductive-age women with elevated estradiol and decreased GH levels have been observed to demonstrate a trend toward lower levels of insulin-like growth factor 1 (IGF-1).²⁸ How these changes in GH and IGF-1 affect the physiology of a perimenopausal woman is not fully understood. It remains to be shown whether the changes in function of the somatotropic axis and hormonal environment affect sensitivity to insulin. This is an important association to be determined, because during the perimenopause, insulin sensitivity decreases, especially when there is weight gain.^{29, 30, 31} Wing and colleagues³⁰ noted a direct correlation with perimenopausal weight gain and insulin resistance. Thus, aging is associated with decreased GH and IGF-1 levels, decreased insulin sensitivity, increased insulin resistance,³² and weight gain.

A prospective study of 485 middle-aged women (42–50 years) revealed that after 3 years, the women gained an average of 2.25 ± 4.19 kg. This change is not entirely dependent on menopausal status, as there were no significant differences between the amount of weight gained by women who remained premenopausal versus those who became menopausal (2.07 vs. 1.35 kg).³⁰

The possible sequelae of these changes in weight and IGF-1 may have great clinical impact because they are predictive of cardiovascular disease.^{30, 31} Thus, the perimenopause is an important time for a woman to mitigate her risk factors for cardiovascular disease (weight control, diet, and exercise).

Insulin sensitivity and diabetes

Decreased pancreatic beta-cell function and decreased insulin sensitivity are two major risk factors for the development of type 2 diabetes. There are ethnic differences in insulin sensitivity and beta-cell function among women in the menopausal transition. African-American, Chinese-American and Japanese-American women have decreased insulin sensitivity when compared with non-Hispanic white women.³³ However, African-American women have a compensatory increase in beta-cell function, whereas Chinese-American and Japanese-American women do not.³³ Type 2 diabetes can be prevented or delayed with lifestyle modifications or pharmacologic therapy. Therefore, these differences in insulin sensitivity and beta-cell function risks should guide physicians to targeted prevention strategies for different ethnic groups.

Changes in the Androgenic Environment

During a woman's reproductive years, the three major sources for circulating androgens are the ovary, the adrenal cortex, and peripheral conversion of circulating androstenedione and dehydroepiandrosterone (DHEA) to testosterone. The ovary produces 25% of circulating testosterone, 60% of circulating androstenedione, and about 20% of circulating DHEA. The adrenal cortex produces 40% of circulating androstenedione, 25% of testosterone, and almost all of the circulating DHEA and DHEA sulfate (DHEAS). After menopause, peripheral conversion of androstenedione accounts for about 50% of circulating testosterone levels.³⁴ About 50% of postmenopausal women still have androgens produced by the ovaries. The quantities are minimal compared to men, and their physiologic significance, if any, is unknown.³⁵

The perimenopausal ovary may actually be producing increased estrogen secondary to increased pituitary secretion of FSH, but total androgen (from both adrenal and ovarian sources) levels decrease.¹⁹ During the perimenopause, androgens continue the decline that commenced after age 20. By age 40, serum androgen levels are about half those found at age 20.³⁶

In normally menstruating women, there is a preovulatory increase in intrafollicular and peripheral androgens. At midcycle, peripheral androstenedione and testosterone increase by 15–20%.³⁷ There are several speculations regarding the role of the midcycle rise in androgens. It may help accelerate follicular atresia so that at ovulation, there is a single dominant follicle.³⁸ It may be involved in the stimulation of libido: it has been shown that female-initiated sexual activity occurs most often at midcycle.³⁹

The increase in midcycle free testosterone and androstenedione seen in younger women has been found to be reduced-to-absent in older women (43–47 years) who were cycling regularly and had normal levels of prolactin and thyroid-stimulating hormone. The decreased concentrations of free testosterone and androstenedione, without significant changes in sex hormone-binding globulin, suggest that in older women these hormones are produced in less quantity.

Because the changes were dependent on menstrual cycle stage, it was concluded that an ovarian, not an adrenal, process is involved.³⁸

Aromatization of androgens, which occurs mostly in extrasplanchnic tissue, is strongly affected by age. The specific activity of aromatase increases with age,⁴⁰ and it has been suggested that age may be a stronger predictor of aromatization than weight or body mass index. It has been observed that as women traverse the menopause, the interconversions of androstenedione to testosterone, estrone, and estradiol change. Thus, in the circulation, there is a greater decrease in the concentration of the products (androstenedione and estradiol) than the precursors (testosterone and estrone).¹⁸

Adrenal androgens are the most abundant hormones in the body. Their production peaks in the early 20s, and with increasing age their secretion decreases greatly. However, a transient rise in the mean level of DHEAS during the late stage of the menopausal transition (when women are skipping periods) has been observed in one longitudinal study (SWAN, the Study of Women's Health Across the Nation).⁴¹ This DHEAS rise is followed by an age-related decrease in women between the ages of 45 and 50 years⁴¹ and the decrease accelerates after menopause. Concentrations of DHEAS in the elderly (>80 years old) are only about 10% of those in people in their 20s.⁴² The age-associated decrease in DHEAS is independent of cortisol.⁴³

DHEAS is not biologically active unless it is converted to testosterone or estradiol. The decrease in adrenal androgen secretion that occurs during the perimenopause appears to be independent of reproductive aging and instead represents somatic aging. However, studies to substantiate this notion still need to be performed.⁹ Because the adrenal androgens, DHEA and DHEAS, have such no cognate receptor, they do not have intrinsic biologic activity unless converted to more active androgens. They have been considered to be potentially important in immunocompetence and general well-being.⁴⁴ Their role in the perimenopause has yet to be fully established.

Vasomotor Symptoms 

The prevalence of vasomotor symptoms in premenopause and during the menopausal transition varies by ethnicity. In premenopause, Hispanic women report the highest prevalence of vasomotor symptoms (42%) compared to African-American women (38%), and the lowest prevalence of symptoms were observed among Caucasian (28%), Japanese (28%) and Chinese (23%) women.⁴⁵ The transition to late perimenopause is associated with the largest increase in prevalence of vasomotor symptoms among all ethnic groups with 79% of African-American, 70% of Caucasian, 65% of Hispanic, 62% of Chinese, and 55% of Japanese women reporting these symptoms.⁴⁵ This ethnic variation was also observed in another study, with African-American (21.8% for moderate to severe hot flashes) and Hispanic women (19%) 2–3 times more likely to report moderate to severe vasomotor symptoms than Caucasian (7.9%) and Asian women (6.4%).

FSH was higher in women who had at least one hot flash per day and estradiol levels were higher in women who had one or no hot flashes per week.⁴⁶ In addition, women with increased cortisol (>10 ng/mg creatinine) during the late menopausal transition have more severe vasomotor symptoms than those with normal cortisol, but did not differ in terms of age, body mass index, or perceived stress level.⁴⁷

Changes in Cycle Length and Menstrual Bleeding

Women aged 18–24 have an average follicular phase length of 15 ± 2 days, but those aged 40–44 have an average length of 10 ± 4 days, which tends to shorten the menstrual cycle.⁴⁸ Thus, menstrual cycle length may shorten before it lengthens, as women progress through the transition and eventually skip menstrual cycles. One hallmark of the menopause transition is a change in the bleeding pattern. Van Voorhis and colleagues studied reproductive hormone production and menstrual bleeding pattern in a large subcohort (the Daily Hormone Study) of the SWAN participants aged 42–52.⁴⁹ They found that twenty percent of all cycles were anovulatory. They also noted that short cycle lengths (<21 days) were common early in the menopause transition whereas long cycle intervals (36+ days) were associated with the late menopause transition. Both short and long cycles as well as short and long duration of menstrual bleeding were frequently associated with anovulation.⁴⁹

Changes in Bone Mineral Density

Decreasing ovarian function in the perimenopause is associated with reduced trabecular bone mass and altered calcium metabolism.^{50, 51} Although premenopausal bone loss is especially significant in cortical bone such as the hip, peri- and postmenopausal bone loss occurs in all skeletal sites, especially trabecular bone. In premenopausal women, bone loss has been significantly associated with lower concentrations of androgens; however, in peri- and postmenopausal women, lower levels of androgens and estrogens have been noted. Thus, sex steroids are important before menopause to maintain integrity of the skeleton and also are important during the peri- and postmenopausal years.⁵²

Bone mass measurements may be predictive of perimenopausal traumatic fractures in addition to postmenopausal fractures secondary to osteoporosis. Fractures in perimenopausal women can be weakly but significantly predicted by bone mass quantification (especially of the lumbar spine) using dual-energy x-ray absorptiometry (DEXA) of the hip and spine. One study of screening DEXA involving 1000 perimenopausal women noted a 2% incidence of stress fractures in women in the 2 years before screening.⁵³

SWAN found that over a 4-year period lumbar spine bone mineral density decreased 3.2% among pre- and perimenopausal women who progressed from the early to the late menopause transition, 3.9% in surgically menopausal women and 5.6% in naturally menopausal women.⁵⁴ Higher baseline and subsequent high FSH concentrations over a 4-year period were associated with lower bone mineral density at the spine and total hip. Although a low absolute level of estradiol (<35 pg/ml) was associated with lower bone mineral density, measures of baseline estradiol and its 4-year variation were poor predictors of bone mineral density.

One study comparing bone mineral density of pre-, peri-, and postmenopausal women revealed that compared with premenopausal women, bone mineral density was lower only in postmenopausal women not currently using HRT.⁵⁵ However, in the SWAN Study, longitudinal follow-up of bone density combined with careful menopausal staging revealed clearly that bone mineral density is not appreciably lost until a woman reaches the late menopausal transition.⁵⁴ These findings clearly imply that prolonged hypoestrogenemia is required before reductions in bone density are seen at menopause.⁵⁴ However, bone mineral density decreased and bone resorption increased with age in the perimenopausal group. Compared with premenopausal women, perimenopausal women had twice the gonadotropin levels and 20% greater urine N-telopeptide excretion. However, their serum estradiol levels and bone formation markers were no different. In postmenopausal women, bone resorption markers were decreased in women using HRT. The researchers concluded that the overall major independent predictors of bone mineral density were age and levels of urine N-telopeptide, serum bone alkaline phosphate, and serum FSH. Urine free deoxypyridoxine was positively associated with bone mineral density in pre- and perimenopausal women. Further studies are necessary to establish the role of urinary bone markers in the clinical care of the perimenopausal woman.

Changes in Mood

Several epidemiological studies have demonstrated that transition to menopause is strongly associated with depressed mood and depressive symptoms, even in women with no lifetime history of depression.⁵⁶ Whether other risk factors, such as the presence of vasomotor symptoms, use of hormone therapy and the occurrence of adverse life events independently modify this risk is less well studied. However, health care providers need to be aware of and sensitive to the increased vulnerability of women to depressive symptoms or even a major depressive episode during the menopausal transition.  

Changes in Sleep

Sleep disturbances have been identified as an important symptom associated with the menopause transition and early menopause.⁵⁷ The SWAN study found that 38% of women in the menopause transition and early menopause endorsed difficulty sleeping.⁵⁸ In a subcohort of about 900 women from SWAN (the Daily Hormone Study cohort), 19% of premenopausal and perimenopausal women reported trouble sleeping severe enough to meet criteria for insomnia.⁵⁸ This study also found that perimenopausal women were more likely to suffer from sleep disruption than premenopausal women. The most trouble sleeping was observed in both groups at the beginning and end of the menstrual cycle.⁵⁹ This association between trouble sleeping and the menopause transition has not been confirmed by all studies. Freeman and colleagues found no link between self-reported poor sleep and menopausal stages.⁶⁰

Cardiovascular Changes in the Perimenopause

After age 40, cardiovascular disease is the leading cause of death in women in the United States.⁶¹ Also, use of combined oral contraceptives increases the risk for myocardial infarction after age 40. Although there is a substantial increase in absolute risk of cardiovascular disease associated with the transition into menopause, there is no abrupt increase in cardiovascular disease incidence during this transition, and therefore the effects of menopause on cardiovascular risk may be delayed or may be largely confounded by aging. Risk factors for cardiovascular disease include high cholesterol and other alterations in the lipid profile, abnormal glucose tolerance, insulin resistance, hypertension, smoking, and obesity. After the menopause, higher triglycerides, cholesterol, total/high-density lipoprotein cholesterol ratio, very low-density and low-density lipoprotein cholesterol, insulin levels, and body weight are present.^{62, 63} Lipoproteins are modified by the menopausal transition. Increases in LDL cholesterol and triglycerides and declines in HDL cholesterol are greater during the menopausal transition than in postmenopause.⁶⁴ Systolic blood pressure, arterial stiffening and risk for a clinical thrombotic event show a linear increase with advancing age, and appear to be less related to the menopause transition per se. Estrogen may have cardioprotective effects independent of its effects on lipids, including changes in endothelial function, vasodilation, improved pulsatility index, improved blood flow, and inhibition of atheromatous plaque formation,^{65, 66, 67, 68} although its beneficial effects decrease with age and advancing coronary disease.

Cardiovascular risk factors can be modified by lifestyle changes, including exercise, weight loss, careful diet, blood-pressure monitoring, and stress reduction. The perimenopause presents an ideal time for a woman to modify her risk factors to maximize not only her perimenopausal years but also her menopausal years.

Hormonal therapy (HT) given during the perimenopause may be used to alleviate symptoms such as hot flashes, difficulty sleeping and vaginal dryness. If a patient has complaints associated with the perimenopause such as hot flashes and difficulty sleeping, it is reasonable to consider starting HT if there are no contraindications. HT can be administered in different formulations. Typically estrogen is administered orally or transdermally with cyclic or continuous oral progestin. New low and ultra-low-dose estrogen oral preparations (containing 0.5 mg estradiol and 0.3 mg conjugated equine estrogens, respectively) appear to maintain benefits for symptom relief and bone sparing while minimizing side-effects and risks. Estrogen-based hormone therapy, especially locally released estrogen therapy such as vaginal tablets or cream, is effective in treating vaginal dryness. 

Standard HT regimens are inadequate for contraception. A sexually active woman who has had less than 1 year of amenorrhea before beginning HT is at a low but present risk for an unwanted pregnancy. She should be advised and encouraged to consider other forms of contraception, such as male sterilization or placement of a Mirena intrauterine system. Alternatively, very low-dose (20 μg) ethinyl estradiol oral contraceptives have a high rate of acceptance and an overall excellent safety profile in the nonsmoking older reproductive-aged woman and may be safely used up to menopause. The contraceptive benefits of oral contraceptives should be weighed against their age-related risks. 

The transition from oral contraceptives to hormone replacement therapy is a common clinical dilemma. There is no biochemical test that definitively predicts the menopausal state. Without conclusive clinical data, it is our policy to establish prospectively a date at which oral contraceptives will be discontinued and hormone replacement therapy will be commenced. This can be done in partnership with the perimenopausal patient and must take into account the fact that hormone replacement therapy is not an adequate contraceptive. For most women, a comfortable age at which to make this transition is about 51 years, or the average age at natural menopause.

The transition to menopause is a poorly defined period in a woman's life. Clinically, it is important to appreciate that the entire reproductive system, not just the ovary, is undergoing change across the transition. One of the hallmarks of the menopause transition is a change in menstrual cycle length and bleeding pattern. Other symptoms include hot flashes, difficult sleeping, changes in bone mineral density (osteopenia/osteoporosis) and depressed mood. Reproductive hormonal fluctuations and changes in androgen and GH/IGF-1 levels may underlie some of the common symptomatology of the menopause transition. Although the treatment of women in the menopause transition is sometimes clinically challenging due to the lack of a neatly organized transition period and the variation that exists between women and within each woman, it is important that we become familiar with the symptoms experienced by these women and their needs. The transition into menopause may lead to specific risk factor loading for later life cardiovascular disease and metabolic syndrome, and therefore represents an ideal time for amelioration of risk factors that may affect quality of life, not just during the menopause transition but also during menopause.