Minireview: Translational Animal Models of Human Menopause: Challenges and Emerging Opportunities

Abstract

Increasing importance is placed on the translational validity of animal models of human menopause to discern risk vs. benefit for prediction of outcomes after therapeutic interventions and to develop new therapeutic strategies to promote health. Basic discovery research conducted over many decades has built an extensive body of knowledge regarding reproductive senescence across mammalian species upon which to advance animal models of human menopause. Modifications to existing animal models could rapidly address translational gaps relevant to clinical issues in human menopausal health, which include the impact of 1) chronic ovarian hormone deprivation and hormone therapy, 2) clinically relevant hormone therapy regimens (cyclic vs. continuous combined), 3) clinically relevant hormone therapy formulations, and 4) windows of opportunity and optimal duration of interventions. Modifications in existing animal models to more accurately represent human menopause and clinical interventions could rapidly provide preclinical translational data to predict outcomes regarding unresolved clinical issues relevant to women's menopausal health. Development of the next generation of animal models of human menopause could leverage advances in identifying genotypic variations in estrogen and progesterone receptors to develop personalized menopausal care and to predict outcomes of interventions for protection against or vulnerability to disease. Key to the success of these models is the close coupling between the translational target and the range of predictive validity. Preclinical translational animal models of human menopause need to keep pace with changes in clinical practice. With focus on predictive validity and strategic use of advances in genetic and epigenetic science, new animal models of human menopause have the opportunity to set new directions for menopausal clinical care for women worldwide.

Reproductive senescence across mammalian species is complex and particularly so in the human (1–5). Animal models of human menopause serve as windows into the multitiered biology of reproductive senescence and are particularly informative regarding events that occur in organs and systems, cellular, molecular, and genomic, that are not readily accessible to investigation in the human (1, 6–10) The power of animal model systems to elucidate underlying biological processes underlies their increasing importance on the translational validity of these animal models to discern mechanisms of human disease, discover new therapeutic targets, and predict outcomes of therapeutic interventions (11–14). The disparity between predicted outcomes based on discovery and population epidemiological science and the clinical outcomes of the intervention trials embedded within the Women's Health Initiative (15, 16) [see Maki (27), this issue] has brought into sharp relief the need to address the validity, strengths, and weaknesses of translational animal models of human menopause and their use to predict outcomes of therapeutic interventions that target this transition.

Decades of rigorous basic discovery research has created a strong foundation in the fundamental biology of mammalian, nonprimate and nonhuman primate, reproductive senescence and the extent to which these animal models share or diverge from features of the human menopause (6–8, 10, 17–23). The focus on translational animal models of human menopause is within the context of a global aging population and within that context that women can expect to live a third or more of their lifetime in the menopausal state. By the year 2030, more than 1.2 billion women will be 50 yr or older worldwide (24). In the United States, an estimated 6000 women reach menopause every day, which converts to over 2 million women per year with a projected 45 million women over the age of 55 by 2020. Each year approximately 1.5 million American women between the ages of 45 and 55 enter into the perimenopause, and within this age range, an estimated 5% have already reached natural menopause (25). Of the menopausal women in the United States, an estimated 2 million have undergone surgically induced menopause from hysterectomy with removal of the ovaries (26). An estimated 6 million menopausal women in the United States currently use some form of prescription-based hormone therapy (www.menopause.org/hormonetherapystats.aspx), whereas another 4 million use over-the-counter products for menopausal symptoms (28).

Given the aging of national populations globally, the increasing reliance of economies on women in the workforce, along with the advancing age of retirement, basic science discoveries that elucidate the biology of the menopausal female and the translation of those discoveries to promote menopausal female health is of utmost importance. This review builds on the solid foundation of basic science in reproductive senescence to address how discovery models can advance to reliably predictive translational models of human menopause to address issues of critical importance to women's health.

Animal Models of Menopause

Natural reproductive senescence

Strengths and weaknesses

The stages of human reproductive senescence, from regular cycling to irregular cycling to acyclicity, are features common to preclinical discovery rodent and nonhuman primate animal model systems. Furthermore in humans, nonhuman primates, and rodents, follicular atresia initiates reproductive senescence. Of the animal models of human menopause, nonhuman primates have multiple strengths as they closely recapitulate many features of human menopause transitions including a relatively long lifespan and endocrine changes consistent with pathways comparable to the human (1, 8, 23). The long lifespan, high expense, and the substantial barriers to access make use of the nonhuman primate as a translational model feasible for relatively few laboratories. However, the nonhuman primate models of human menopause are a critical link across the translational divide between data obtained in rodent models and ultimate translational predictive validity to humans (29–31).

Remarkably, rodents, rats and mice, share multiple features and endocrine changes found in the human. Although the transition in the rodent is of shorter duration, it is characterized by multiple features found in the human, which include 1) decline in follicles, 2) irregular cycling and steroid hormone fluctuations, and 3) irregular fertility (9, 19). Irregularity of cycle length is a reliable indicator of irregular fertility and impending reproductive senescence and is characteristic of laboratory rodents of most genotypes with initiation of irregularity occurring approximately 8 months of age (19). The reliability of changes in the reproductive cycle of laboratory rodents provides a well-defined and tractable model of the irregular cycles of human perimenopause.

The weakness of the rodent model is that approximately 60–70% of the aging rodents spontaneously transition into a polyfollicular anovulatory state of constant estrus characterized by sustained levels of plasma 17β-estradiol and low levels of progesterone that can last 10–100 d (19). The remaining 30–40% of aging rodents transition from irregular cycling to anestrus with consistent low levels of ovarian steroids (19), which is a better model of the human perimenopause to menopausal transition. Thus, in an aging colony of rodents, 25–40% (in our experience, it is 25% of the animals and has not ever approached 40%), of the animals will naturally model the human menopausal transition. To identify animals falling into the two endocrine states, vaginal cytology is conducted to detect keratinized epithelial cells indicative of vaginal cornification (19). Although eventually all rodents transition to the anestrus state, substantial differences in brain gene expression occur between the constant estrus and anestrus animals (32, 33). One potential strategy to enrich the number of anestrus animals is to ovariectomize the animals after a standardized number of irregular cycles, which would prevent constant estrus-induced changes but would also introduce the variable of a loss of ovarian secreted factors like testosterone (34, 35).

Translational opportunities

Two obvious translational gaps exist between the preclinical and clinical realms: 1) stage of reproductive senescence intervention and 2) hormone interventions in the ovary-intact female. The majority of women experience natural menopause and thus undergo the transitions at a cellular and molecular level that ultimately lead to reproductive senescence. Recent data suggest that intervention with hormone therapy during the perimenopause in ovary-intact women improved cognitive function later in life (36). To date, preclinical models of human menopause use ovariectomized animals to determine the impact of hormone interventions, whereas the human experiment/clinical practice is to treat women with hormone interventions during their perimenopausal transition or at menopause when ovaries are secreting steroidal factors such as dehydroepiandrosterone (35) and testosterone (34).

Ovariectomy

Strengths and weaknesses

The use of ovariectomy, removal of both ovaries, as a model of menopause is widespread throughout discovery and preclinical translational research (9, 37). Typically, in this surgical model of menopause, bilateral removal of the ovaries occurs in young reproductively competent healthy animals (37). Experimental interventions occur either at the time of ovariectomy or commence once 17β-estradiol has reached a low to nondetectable level in plasma, which typically occurs within 1–2 wk. This model system has been and remains a widely used model of human menopause. As a discovery strategy, data derived from the ovariectomy animal model has created much of our fundamental understanding of ovarian hormone action in every organ system in the body (38–43).

Ovariectomy in reproductively capable animals, both rodents (37–42) and nonhuman primates (23, 44), serves as a remarkably predictive model of oophorectomy in premenopausal women (45). Rocca and colleagues (46) have explored the impact of oophorectomy before and after natural menopause on risk of mortality and neurodegenerative and cardiovascular disease. Outcomes of these analyses in humans indicate that loss of ovaries before natural menopause increases the risk of death, Parkinson's, Alzheimer's, depression, anxiety, and cardiovascular disease (46–53).

Several assumptions that limit the ovariectomy model of menopause are considered here: 1) plasma levels of ovarian hormones are indicative of ovarian hormone concentrations in organs; i.e. the ovaries are the only organs capable of producing ovarian hormones; 2) outcomes derived from analyses of short-term ovariectomy generalize to long-term ovarian hormone deprivation; and 3) removal of the ovaries is a model that broadly generalizes to human menopause. The assumption that plasma levels are indicative of organ levels of ovarian hormones was challenged by the data from Melcangi and colleagues (54), who found that changes in plasma levels of neuroactive steroid levels after gonadectomy do not necessarily reflect the steroid levels in either the peripheral or central nervous system. Furthermore, generation of neurosteroid compensatory responses were region specific and time dependent (54).

The assumption that responses to short-term ovariectomy are indicative of outcomes of long-term ovariectomy belies the remarkable adaptive responses that organs undergo to compensate for the loss of ovarian hormones. As mentioned above, the peripheral and central nervous systems are capable of steroidogenesis that changes with duration of ovariectomy. Our own research has shown that short-term and long-term ovarian hormone deprivation are not comparable, and in the case of the brain, multiple bioenergetic adaptations occur within the estrogen-regulated components of this system (55). Rapid responses to ovariectomy are well documented and can occur within days to 2 wk (37). However, adaptive responses to ovariectomy continue throughout multiple systems over a long period of time. In the normal rodent brain, generation of detectable β-amyloid occurred after 8 wk of ovariectomy (56), whereas changes in temperature regulation occur after 3–5 wk of ovariectomy (57). Furthermore, responsiveness to estrogen changes with time is an early indicator of the window of opportunity (58).

Lastly, the assumption that ovariectomy in young animals is a general model of menopause is not well supported. One illustrative example is the adaptations that occur in the estrogen-regulated metabolic pathways during natural reproductive senescence in the mouse brain that are abrogated by ovariectomy. In this system, natural reproductive senescence is associated with adaptive compensatory responses in brain metabolism pathways (59, 60), whereas in the ovariectomized rodent brain, the adaptive metabolic responses did not occur. In the hippocampus of ovariectomized mice, there was a direct transition to use of fatty acids and the generation of β-amyloid, which is indicative of accelerated metabolic aging of the brain (59–61). Furthermore, we found that in a mouse model of Alzheimer's disease, ovariectomy increased mortality 3-fold over a period of months (57).

Translational opportunities

A major translational gap that exists between the preclinical and clinical realms for this model is between the surgical interventions experienced by women and those used in preclinical animal models. Approximately 600,000 hysterectomies, removal of the uterus, are performed each year in the United States and is the second most frequent major surgical procedure among reproductive-aged women. From 2000 through 2004, an estimated 3.1 million U.S. women had a hysterectomy with the highest number among women aged 40–44 yr just before natural menopause (26). Hysterectomy is frequently accompanied by bilateral or unilateral removal of the ovaries (62) for either prevention of ovarian cancer or postsurgical complications. Conservative estimates report approximately 300,000 prophylactic bilateral oophorectomies per year at the time of hysterectomy in US women (52). Surgical procedures conducted in preclinical translational animal models that mirror those that women experience and at a comparable age (6–10 months of age in the mouse, before reproductive senescence to correspond to approximately 38–44 yr of age, or perimenopause transition in the human) remain to be fully explored. Translationally, preclinical models of surgical interventions that women experience worldwide have the potential to significantly advance understanding of the near- and short-term implications of these procedures and the efficacy of hormone strategies to prevent outcomes of ovarian deprivation. Given the strong predictive validity of the preclinical ovariectomy animal model to adverse outcomes in women, this model, if modified to reflect current surgical practice, has the potential to significantly advance understanding of the impact of clinically relevant surgical interventions that women experience worldwide.

Ovotoxins to induce accelerated ovarian failure

Strengths and weaknesses

Of the two classes of ovotoxins to induce accelerated ovarian failure, the diepoxide form of 4-vinylcyclohexene (VCD) and polycyclic aromatic hydrocarbons found in cigarette smoke, VCD is the best studied. This ovotoxicant is an industrial chemical intermediate generated in the production of synthetic rubber, flame retardants, and polyolefins and present in the off gases during these synthetic processes, which can lead to exposure of workers to varying levels of this toxicant over varying durations (63). VCD accumulates in lipid compartments and is not environmentally biodegradable. VCD-induced follicular loss induced by chronic exposure to low-dose VCD is by apoptosis, consistent with the mechanism of ovarian follicular loss in the mammalian ovary (9, 63). The strengths of the VCD ovotoxin menopausal animal model are that it allows the dissociation between effects of age and effects of ovarian hormone loss and that it induces an apoptotic mechanism, of follicular atresia that mimics that of the human female. Furthermore, the VCD model is quite effective at inducing follicular atresia in multiple mouse strains allowing use of mechanistically targeted transgenic mouse models (9, 63). In rodents, it is possible to inject VCD and achieve ovarian failure, whereas in the nonhuman primate, surgical wrapping of the ovary with a VCD-releasing fiber followed by a subsequent surgery to remove the fiber was required (64). Substantial toxicology data across multiple doses, regimens, and routes of exposure and organ systems currently exist, providing an extensive body of information on which to base use of this model system (9, 63). The VCD ovotoxin animal model of menopause recapitulates many of the system-level outcomes of ovarian hormone depletion including a rise in FSH, increased androgen production, atherosclerotic plaque formation, bone loss, and features of metabolic syndrome (9). Furthermore, an emerging body of evidence indicates the efficacy of the VCD ovotoxin animal model to recapitulate many features of the ovariectomy model in brain (9).

The weakness of the VCD ovotoxin animal model of menopause are the well-documented toxic and or carcinogenic effects of this compound on other organ systems such as lung, liver, and ovaries. These effects are related to both dose and duration of exposure (9) and thus do require that investigators be aware of these parameters for investigative purposes. The chemico-physical properties of this compound, small molecular mass of 108.18 g/mol, cyclohexene structure, and preferential distribution of this compound into lipid suggest the potential of accumulation in brain over the period of exposure to induce follicular atresia.

Translational opportunities

The accelerated ovarian failure model of menopause may be translationally relevant to premature ovarian failure, also referred to as primary ovarian insufficiency or primary ovarian dysfunction. Premature ovarian failure is a syndrome of amenorrhea, low sex steroid levels, and elevated gonadotropin levels among women younger than age 40 yr. Although most frequently idiopathic in origin, premature ovarian failure is also linked to autoimmune disorders, genetic causes, infections or inflammatory conditions, enzyme deficiencies, or metabolic syndromes (65). Premature ovarian failure affects nearly 1% of women under age 40 yr, and spontaneous early menopause is reported to affect approximately 5% of women between ages 40 and 45 yr (65).

A relatively underdeveloped ovitoxicant model of menopause involves the polycyclic aromatic hydrocarbons found in cigarette smoke (66–68). This class includes 9,10-dimethylbenzanthracene, 3-methylcholanthrene, and benzo[a] pyrene (63). Decreased fertility is well documented in women who are exposed to either primary or secondary cigarette smoke (63). The polycyclic aromatic hydrocarbons found in cigarette smoke are known to damage primordial and primary follicles in mice and rats and thus could be a clinically relevant model of ovitoxicant-induced menopause (63).

Translational Opportunities Using Existing Animal Models of Human Menopause

Several key translational gaps exist in the preclinical realm that are easily addressed using current animal models of human menopause and for which there are urgent clinical issues. These include gaps in analysis of 1) acute vs. chronic exposure to ovarian hormone deprivation and hormone therapies,; 2) clinically relevant regimens of hormone therapy (oral vs. transdermal administration and cyclic vs. continuous combined therapies); 3) clinically relevant formulations of hormone therapy with direct comparisons of different hormone therapy formulations, particularly progestogens, on health outcomes; 4) the duration of the critical window (when does it begin, when does it end); and 5) the impact of discontinuing hormone therapy followed by recommencing hormone therapy

In the human, ovarian hormone deprivation is chronic, unless hormone therapy is initiated. A solid foundation of preclinical data does not yet exist because too few preclinical investigations use chronic models of either ovarian hormone deprivation or hormone therapy intervention on which to reliably predict clinical outcomes. Chronic exposure in rodent animal models of human menopause is entirely feasible given the rodent (mouse and rat) to human age conversion. Although the rate of rodent to human age changes across the rodent lifespan, 1 month of mouse life is roughly equivalent to 3.54 yr of human life, whereas 1 month of rat life approximately corresponds to 2.5 yr of human life (69). Mice and rats are sexually mature by 3–6 months of age and approach the endocrine equivalent of human perimenopause by 9 months of age (6, 69, 70). Their reproductive senescence transition occurs between 9 and 12 months of age by which time they are reproductively senescent. Mice in particular have an advantage because their reproductive senescence is driven by follicular atresia as it is in the human (6, 70). Given the accelerated aging of rodent species, conducting chronic exposures is entirely feasible with greater translationally predictive validity (37, 71–73).

The use of clinically relevant regimens of exposure is another translational opportunity. Currently, there are two general regimens of hormone therapy, continuous estrogen therapy in women without a uterus and continuous combined estrogen plus a progestin therapy for women with a uterus (73). A very large body of data derived from acute in vitro and in vivo paradigms indicates that simultaneous exposure to an estrogen and a progestin typically results in a decrement or antagonism of estrogen-inducible responses. More recent data derived from a direct comparison of continuous exposure to 17β-estradiol plus cyclic exposure to progesterone vs. a continuous combined regimen of these hormones over several months resulted in very different gene expression profiles in brain (73). As might be expected, continuous exposure to 17β-estradiol plus cyclic exposure to progesterone induced a gene expression profile comparable to ovary intact rats. In contrast, the continuous combined regimen of 17β-estradiol plus progesterone induced a gene expression profile comparable to chronic ovariectomy (73).

Lastly, the issue of the formulation of hormone therapy is clinically highly relevant (74). The use of medroxyprogesterone acetate (MPA) as a progestin in hormone therapy is now a classic case of when preclinical translational research could have had a significant impact on clinical practice. MPA was developed to block the proliferative effect of estrogen in the uterus. However, investigation of MPA effects in brain lagged behind those in peripheral organs. Studies of MPA in neural tissue were in process during and subsequent to conduct of the Women's Health Initiative and were, in many respects predictive of the outcome of the conjugated equine estrogens plus MPA arm of the Women's Heath Initiative Memory Study (75–81) [see also Maki (27), accompanying minireview in this issue]. Overall, MPA was an effective antagonist of estrogen action in brain as it is in the uterus. Furthermore, the mechanism by which MPA antagonizes estrogen-induced proliferation in the uterus, i.e. increasing apoptosis, is recapitulated in brain in neural progenitor cells and effectively antagonizes estrogen-induced neurogenesis (80). The outcomes of the Women's Health Initiative clinical trials serve to highlight the critical need and importance of preclinical translational investigations of clinically relevant regimens and formulations of hormone therapies for women's menopausal health.

In the United States there are currently seven different estrogenic therapies, seven different progestins, and eight different combination therapies given in either oral or transdermal formulations (82). Very few of these therapies have been tested under acute or chronic conditions and their impact on multiple organ systems investigated. Our analysis of acute in vivo exposure to clinically relevant progestins alone or in combination with 17β-estradiol indicated substantial differences between progestin-induced outcomes from increased neuroregeneration to no effect to increased cell death (80).

Development of New Translational Animal Models of Human Menopause

Translational preclinical animal models of genotypes at risk for conditions that develop late in life during the menopausal years has the potential of contributing to personalized interventions for prevention of menopause-associated conditions. For example, polymorphisms in estrogen receptor 1 (α) have been identified in human populations that confer risk of developing cognitive impairment (83, 84) whereas a polymorphism in estrogen receptor 2 (β) was associated with earlier onset of Parkinson's disease (85). In the progesterone receptor, four polymorphisms and five common haplotypes have been identified. One of these polymorphisms (G331A) was associated with increased risk of panic disorder in women (86, 87). Recent advances in elucidating the epigenetic modifications of endocrine systems regulating development, sexual differentiation, and reproductive senescence are paving the way for preclinical translational models to identify both vulnerabilities and strategies to mitigate risk to the health of the menopausal female (10, 88). Key to the translational success of existing and future models of human reproductive senescence will be the close coupling between the clinical realities of perimenopausal and menopausal aging and the limits of predictive validity for each model. With strategic use of advances in genetic and epigenetic science, existing and new animal models of human menopause have the opportunity to set new directions for menopausal clinical care for women worldwide.