Summary of the National Institute on Aging-Sponsored Conference on Depressive Symptoms and Cognitive Complaints in the Menopausal Transition

The National Institutes of Health-sponsored presymposium, entitled “Depressive Symptoms and Cognitive Complaints in the Menopausal Transition,” was held on September 29, 2009, at the Manchester Grand Hyatt in San Diego, CA. This presymposium was held in conjunction with and on the day before the annual meeting of The North American Menopause Society. The presymposium was supported by award no. R13AG033944 from the National Institute on Aging and also received support from the Office of Research on Women's Health and The North American Menopause Society.

The presymposium brought together leading scientific experts, menopausal healthcare providers, and researchers in an interactive forum with the following goals: imparting state-of-the-science knowledge about the impact of menopause on cognition and mood, understanding the clinical relevance of recent findings, identifying scientific gaps, and reviewing the diagnosis and treatment of cognitive and mood disorders in perimenopausal and postmenopausal women. The presymposium was divided into a morning session focusing on depressive symptoms in the menopausal transition and an afternoon session on cognitive function in the menopausal transition. There were four speakers per session, and each session was structured to cover four methodological approaches including longitudinal cohort studies, randomized intervention trials, pharmacological challenge studies, and clinical diagnosis. Interactive panel discussions focused on translating research findings to clinical practice. Relevant questions included who is at risk for depressive symptoms and cognitive decline, to what extent are endocrine markers effective in identifying at-risk individuals or guiding treatment decisions, and how clinicians should approach diagnosis. Below, we summarize the proceedings of the workshop for each topic.

Part 1. mood and the menopausal transition

Risk of depression during the menopausal transition

Women and clinicians generally identify the approach of menopause by changes in menstrual bleeding. Hot flashes, which are widely considered to be the cardinal symptom of menopause, also signify the menopausal transition for many women. Depressive symptoms, poor sleep, aches, and joint pains are other frequent complaints in this transition period, but whether these symptoms are associated with hormone changes of the menopausal transition is controversial.

Identifying the risk of depression in the menopausal transition is particularly important because of its significant disability and its strong associations with other diseases in midlife women, such as metabolic syndrome, osteoporosis, and cardiovascular disease (1–3). Although some epidemiological studies found no significant association between depression and menopause status, several recent longitudinal studies identified an increased risk of depressed mood in the transition to menopause (4–9).

Three large prospective cohort studies indicate that the likelihood of depressed mood is approximately 30% to three times greater in the menopausal transition compared with the premenopausal stage (6–8). Women with a history of depression are nearly five times more likely to have a diagnosis of major depression in the menopausal transition (5). Most importantly, women with no history of major depression are two to four times more likely to report depressed mood in the menopausal transition compared with the premenopausal stage (6, 7). The changing hormonal milieu contributes to depression in the transition period as evidenced by the associations of depressed mood with menopausal stage, variability in estradiol, and increasing follicle-stimulating hormone levels (5, 6, 10). Other risk factors for depression in these longitudinal studies include poor sleep, hot flashes, stressful or negative life events, lack of employment, age, and race.

These observations of an increased risk of depression in perimenopausal women support the concept that the perimenopausal stage, which is framed by the changing hormonal milieu of ovarian aging, is a “window of vulnerability” for some women.

Clinical trials of estrogen and depressive symptoms

Estrogen alters monoaminergic systems (eg, serotonin and noradrenalin) that are intimately involved in mood and behavior regulation. For example, estrogen increases serotonin receptor density in select brain regions such as the hypothalamus, preoptic area, and amygdala. Estrogen also increases noradrenalin availability and synthesis, while reducing its turnover.

Findings from a double-blind, placebo-controlled, randomized trial demonstrated that 17β-estradiol 100 μg given transdermally for 12 weeks is efficacious in the treatment of depression in perimenopausal women, with remission rates as high as 60% to 80% (11). Conversely, in at least one randomized trial (12), transdermal estradiol has failed to show antidepressant properties when given to postmenopausal women. Together, these studies suggest that the menopausal transition may constitute not only a window of vulnerability for the occurrence of depression in some women but also a “window of opportunity” for the use of estrogen-based strategies for the management of depression.

Role of gonadal steroids in perimenopause-related depression: evidence from pharmacological challenge studies

Schmidt and colleagues (9) have examined the effects of estrogen withdrawal and the recent onset of hypogonadism on mood symptoms using two strategies. First, Harsh et al (13) administered a gonadotropin hormone-releasing hormone (GnRH) agonist for 2 to 3 months to 53 regular cycling, premenopausal women. In contrast to previous reports from gynecology clinic-based samples (14, 15), all women had the absence of current or past psychiatric illness confirmed by a structured psychiatric diagnostic interview and completed daily symptom ratings for 2 months before study entry to confirm the absence of significant mood or behavioral symptoms associated with their menstrual cycle. In addition, all women had normal gynecological and medical examination results. Mood and behavioral symptoms during GnRH agonist administration were measured by the Beck Depression Inventory (BDI), and a self-report symptom rating form was completed on a daily basis. Plasma hormone measures confirmed that the GnRH agonist suppressed the secretion of both ovarian steroids and gonadotropins. Only three women (5.7% of the sample) reported BDI scores greater than seven (suggestive of clinically significant symptoms of depression) and in only one of these women did the increased BDI scores persist beyond 2 weeks' duration. In contrast to the relative absence of depressive symptoms in these women, we did observe the significant appearance of several symptoms including both daytime and nocturnal hot flashes, disturbed sleep, and diminished libido. Thus, in otherwise healthy women, the induction of neither hypogonadism nor hot flashes (with an accompanying sleep disturbance) uniformly precipitated depressive symptoms.

In a second ongoing study, we are evaluating the effects of the acute withdrawal of estradiol therapy in women with and without a history of perimenopausal depression. Preliminary results suggest that estradiol withdrawal induces depressive symptoms in women with a history of perimenopausal depression, but not in those without such a history. No significant depressive symptoms emerged in the women lacking a history of a past perimenopausal depression who were either withdrawn from or maintained on estradiol therapy. Thus, in contrast to our findings with GnRH agonist-induced hypogonadism in premenopausal women with no past psychiatric history, estradiol withdrawal in women with a history of perimenopausal depression triggers mood symptoms. These data are consistent with findings from epidemiological studies showing that, for a subgroup of women, the endocrine events of the late menopausal transition may represent important triggers for mood destabilization and the onset of depression. Both the markers of this risk and the mechanisms underlying estradiol withdrawal-induced depressive symptoms remain to be identified.

In summary, endocrine studies of depression during the menopausal transition suggest the following:

Recognizing perimenopausal depression

Depression is a common and disabling illness. The sex difference in depression rates is most pronounced during the reproductive years, and, at times, women may have twice the rates of depression as those of men (16). Sadness, the hallmark symptom of depression, is a normal emotion and is common during times of change. Treatment of depression is warranted when it is persistent, causes functional impairment, or is accompanied by other physical and cognitive symptoms. Even “minor” depression—states that do meet diagnostic criteria for a major depressive episode—can cause significant impairment. It may be challenging to distinguish between the physical symptoms of menopause and those of depression. Nevertheless, when mindful of Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria and available screening tools, it is possible to improve the detection of depression.

DSM-IV criteria for depression include the presence of five or more symptoms of the symptoms listed below during the same 2-week period (17). At least one of these symptoms must include depressed mood or loss of interest or pleasure in most activities.

One challenge in primary care is the low detection rate in diagnosing depression, with the best detection rates estimated to be 36% to 56% (18). Patient and clinician factors also play a role. Patients infrequently report stressors and frequently report somatic symptoms. Appointment times are short, making a lengthy assessment of depression challenging. One screening instrument that is particularly helpful in quickly screening for depression is the Patient Health Questionnaire (PHQ-9) (19, 20). The PHQ-9 is a nine-item, self-report questionnaire that is useful in clinical settings because it is easy to score and items are linked to DSM-IV depression criteria. The questionnaire can be used to assess and track treatment response (21). Both English and Spanish versions are available free of charge online.

A final note, when screening for perimenopausal depression, clinicians should consider key additional risk factors, including previous episodes of depression, psychosocial stressors, and severe climacteric symptoms (eg, sleep disturbance and vasomotor symptoms) (5, 22).

Depression as a risk factor for cardiovascular disease

Depressive disorders and coronary heart disease (CHD) are leading causes of the global burden of disease. Both diseases are inversely related to socioeconomic status. Depression and CHD are comorbid; although the nature of the relationship between these two diseases is unclear, when depression is present, CHD prognosis is poor. CHD is the leading cause of death of women. Depression is twice as likely in women as in men. Thus, the comorbidity of CHD and depression may be especially important for women's health. In women, depression is associated with life phases that are characterized by changes in reproductive system function, including puberty, the late luteal phase of the menstrual cycle, the postpartum period, and the perimenopause. Nonetheless, most animal models of depression are male.

We have developed an adult female nonhuman primate model of depression in cynomolgus monkeys, which have been used effectively for decades to model CHD risk in women (23). Like women, female cynomolgus monkeys develop atherosclerosis in their coronary arteries. Coronary artery atherosclerosis and its sequelae cause CHD. Like women, socially stressed monkeys are more likely to become depressed. Also, like depressed women, depressed monkeys are insensitive to negative feedback in dexamethasone suppression tests and have smaller anterior hippocampi and lower neural serotonin 1a receptor binding throughout brain areas associated with depression. Depressed monkeys also have higher heart rates, dyslipidemia, poor ovarian cyclicity characterized by lower ovarian steroid concentrations, and lower levels of physical activity than do their nondepressed counterparts; all of these are risk factors for coronary artery atherosclerosis and CHD. Depressed monkeys develop more than four times the coronary artery atherosclerosis as their nondepressed counterparts.

Female cynomolgus monkeys are a unique model in which to study the comorbidity of depression and CHD. Using this model, we are addressing the question of whether depression causes coronary artery atherosclerosis.

Summary depressive symptoms and the menopausal transition

Ovarian steroids regulate many of the signaling pathways, neurocircuits, and behaviors that are hypothesized to be abnormal in depression. Recent evidence from prospective studies suggests that for a subgroup of women, endocrine events during the menopausal transition play a role in the onset of depression. In addition, although perimenopausal depression is not caused by abnormalities of basal ovarian hormone secretion, this disorder, nonetheless, may be effectively treated with estradiol. The specificity of the relationship between the endocrine events of the menopausal transition and depression in these women is further suggested by reports of the lack of antidepressant action of estradiol therapy in postmenopausal depressed women. Nonetheless, studies in which menopause is induced pharmacologically demonstrate that estradiol withdrawal and hypogonadism are sufficient to trigger depression in only a subgroup of women. Nonhuman primate studies are exploring depressive symptoms as a risk factor for the development of cardiovascular disease. Future studies need to identify the biochemical factors and markers of risk underlying the differences between those women who remain asymptomatic during the menopausal transition and those who develop depression. The biological underpinning of this differential behavioral phenotype also may serve to reconcile and/or predict differences in response to hormone therapy (HT).

Part 2. cognition and the menopausal transition

Cognitive change during the menopausal transition

A majority of women report memory problems during the menopausal transition, but studies of measured cognitive performance during the menopausal transition are rare (24). The first study to examine the relationship between menopausal stage and memory was a cross-sectional investigation of women enrolled in the Melbourne Women's Midlife Health Project (25). Women in the early perimenopause, late perimenopause, and postmenopausal stages did not differ in their memory performance. In a secondary analysis, women who initiated HT before the final menstrual period had higher memory scores compared with women who initiated HT after menopause.

The relation between reproductive stage and measured cognitive performance has been the subject of only three published longitudinal studies (26–28). These longitudinal studies included estimates of memory performance during the pre- menopausal stage. A cohort study involved 573 Chinese women aged ≈46 years at baseline and having an average education of ≈6.5 years (26). Comparison of cognitive performance among those who remained premenopausal with those who became perimenopausal during a follow-up interval of 18 months showed that, as expected in this age range, performance on all tests improved. However, those who transitioned to perimenopause had statistically significantly less improve- ment on a test of verbal fluency than did those who remained premenopausal. The Chicago site of the Study of Women's Health Across the Nation (SWAN) initiated a site-specific study of measured cognitive performance (verbal and working memory) at SWAN baseline (the other six SWAN sites did not assess cognition at the outset of SWAN) (27). After 2 years of follow-up in ≈800 women aged 42 to 52 years at baseline, with adjustment for sociodemographics, there were small time-related improvements in both cognitive domains, but no significant effects related to menopausal stage were identified (27).

In contrast to the 2-year results from Chicago-SWAN, the results of a 4-year longitudinal analysis based on data from the full SWAN cohort showed the effects of the menopausal transition and of hormone use on measured cognitive performance (SWAN added tests of cognitive processing speed, verbal episodic memory, and working memory to annual examination at the fourth follow-up visit) (28). Processing speed improved with repeated testing during premenopause, early perimenopause, and postmenopause, but not during late perimenopause. Similarly, verbal memory scores improved with repeated testing for premenopausal and postmenopausal participants, but not for perimenopausal participants. Former hormone use (used before the final menses) was associated with better processing speed and verbal memory. In contrast, current hormone use during postmenopause predicted poorer processing speed and verbal memory, compared with performance during premenopause. These newer SWAN results probably differ from those of the Chicago site-specific report because the SWAN cohort sample size was three times as large, follow-up was twice as long, and the effects of previous and current hormone use were accounted for. All other published investigations of menopausal transition stage and cognition have been cross-sectional (29).

Clinical trials of estrogen and cognition

The primary outcome of interest in clinical trials of estrogen and cognition is commonly a test of verbal memory, typically a word-list learning task or a paragraph recall task that includes measures of both immediate recall and delayed recall. A focus on verbal memory is justified by its relevance to de- mentia risk. Deficits in verbal memory have been shown to be the earliest neuropsychological predictor of conversion to Alzheimer disease in a number of studies. In addition, women, on average, outperform men on these tasks, women complain about lapses in memory for names and other verbal information during the perimenopause, and new evidence from SWAN reviewed above demonstrates that women show performance deficits on verbal tests during the perimenopausal stage (28).

Given certain patterns in findings across clinical trials, it is helpful to distinguish the effects of estrogen alone versus estrogen plus progesterone and the effects in younger (ie, <65 y) versus older (ie, >65 y) postmenopausal women (30). Trials of estrogen alone in younger postmenopausal women generally have included small sample sizes (ie, <100) but show some benefit to verbal memory. A recent study of transdermal estradiol revealed benefits on a test of verbal memory, but the benefit was evident on outcome measures relating to error monitoring and other executive functions (31). Trials of estro- gen alone in older women consistently show neutral effects on verbal memory. Studies of estrogen plus progesterone treatment in younger postmenopausal women reveal a reliable but small deficit in verbal memory when the treatment is conjugated equine estrogens plus medroxyprogesterone ace- tate (CEE/MPA). Only one trial in younger postmenopausal women used a different estrogen plus progesterone intervention, and that trial revealed a beneficial effect of estradiol valerate and dinogest (32). Studies of estrogen plus progesterone in older women parallel those in younger women and demonstrate a modest deficit in verbal memory with CEE/MPA. Cyclic oral estradiol and norethindrone acetate had neutral effects on verbal memory in older women, although a subanalysis revealed that treatment improved verbal memory in women whose baseline verbal memory was above average (33). The finding that the baseline health or cognitive status of women modulates the effects of estrogen on cognition has been reported in other studies, including the Women's Health Initiative Memory Study and the Women's Estrogen for Stroke Trial (34, 35).

What other factors might modulate the impact of HT on cognition? There is more evidence of an overall benefit in younger versus older women, but CEE/MPA produces modest decrements in verbal memory in both age cohorts. There have been several demonstrations of neutral cognitive effects of other HT formulations in older women, including estradiol. In these studies, it is difficult to determine whether the critical factor associated with benefit is young age and/or other factors confounded with young age, including less time since the final menstrual period or better health. Three studies in older women suggest that the greatest cognitive benefits are evident in women with intact cognitive function. There is preliminary evidence of a negative correlation between memory function and hot flashes, but this relationship is evident only when hot flashes are measured with ambulatory monitors. Future studies are needed to better understand the phenotype of women who might show cognitive benefit with HT.

Role of gonadal steroids in perimenopause-related cognitive changes: evidence from pharmacological challenge studies

The loss of estradiol production in postmenopausal women has profound negative effects on a number of organ systems including the brain. As reviewed above, loss of estrogen support during the perimenopausal or early postmenopausal period in women can produce measurable and significant impairment of performance on certain cognitive tasks, and estradiol administration seems to prevent these changes and improve cognitive performance in some models.

In an attempt to understand how estrogen effects on brain function are mediated, we have been focusing on the activities of the brain central cholinergic system. Several decades of research support the critical role of brain cholinergic systems in cognition in humans, particularly in learning, memory formation, and attention. Because the cholinergic system is a neurotransmitter system that projects to all regions of the brain and has been shown to decline in aging, it remains the best candidate neurotransmitter system to be related to cognitive changes in aging. The cholinergic changes that occur with aging may be one explanation for the common pattern of alterations in cognitive performance and brain activation seen in older adults on tests of attention, working memory, and episodic memory. The cholinergic hypothesis is general enough to provide an underlying neurobiological framework for these age-related changes. For example, decreases in cholinergic functioning may slow cognitive processing for older adults and result in less efficient cognitive processing.

Estradiol has significant positive effects on brain cholinergic neurons, interacts with trophic factors on neuronal development and plasticity, and improves cholinergic-related cognitive processes in animal models. We have extended the studies to human models to investigate whether estradiol effects on human cognitive functioning are mediated at least in part through cholinergic system activity.

Our studies show that estradiol seems to improve cholinergic function as measured by increased resistance to anticholinergic blockade in normal postmenopausal women (36, 37). This improvement may be dose and domain specific; that is, lower doses improve primarily attentional functioning, and higher doses may influence episodic memory. The effects of estradiol on cholinergic function related to episodic memory may be age specific, with younger women showing benefit but older women showing no benefit or impairment. This result provides direct experimental support for the “critical period hypothesis.” The addition of progesterone to estradiol seems to partially or completely counteract the ability of estrogen to enhance its attention and speed after cholinergic blockade. Intriguingly, the selective estrogen receptor modulator tamoxifen seems to decrease attentional and long-term recall impairment produced by cholinergic antagonists and improves spatial working memory performance. The apolipoprotein E (APOE) genotype, a known risk factor for Alzheimer disease and cholinergic dysfunction, seems to modulate the directionality and magnitude of the effect of the positive estrogen-like effects of tamoxifen on cognition after cholinergic blockade. Taken together, these results provide strong support for our original hypothesis that the ability of estradiol and related compounds to modify cognitive functioning in old age and/or partially prevent cognitive deterioration or neurodegenerative disorders may be related to the effects on the brain cholinergic system.

Recognizing cognitive disorders

In recognizing cognitive disorders, it is useful to distinguish among dementia, mild cognitive impairment (MCI), and cognitive aging. Dementia entails substantial functional decline, and Alzheimer disease and other late-life dementias are an ever-increasing burden in an aging population. This is a burden that affects women disproportionately. Dementia is preceded by a transitional stage of cognitive decline, and the term MCI is increasingly applied to patients meeting research criteria believed to characterize incipient dementia. The inference is that such patients already have pathological changes associated with dementia, but the pathological burden is still modest in extent or has not yet overwhelmed innate compensatory mechanisms. The concept of cognitive aging is closely related to that of MCI. Cognitive skills vary with age. For many aspects of cognition—but not all—cognitive aging represents detrimental change, beginning in middle age and accelerating in old age. The underlying assumption, however, is that cognitive aging reflects processes largely distinct from those that culminate in Alzheimer disease or other defined dementias.

Dementia is rare in midlife, but forgetfulness and other cognitive symptoms are common. The clinician is often called upon to evaluate the seriousness of midlife cognitive symptoms, which may occur in the setting of fluctuating or declining levels of ovarian hormones, and to make recommendations regarding treatment or risk reduction. Fortunately, recent observational and experimental data are largely reassuring regarding the effects of the natural menopausal transition on objective measures of episodic memory and other cognitive skills. From a clinical perspective, identifying a deficit in episodic memory is an important first step to recognizing the cognitive disorders that occur with age. Other worrisome features that might emerge during the patient assessment include a history of functional decline, a family history of dementia before the age of 60 years in a parent or sibling, or a focal abnormality on the neurological examination.

It is also important to identify the factors that could affect subjective memory, even if there is no objective decline beyond that expected with age. These include cognitive aging per se, depression, stress effects, fatigue, hot flashes, sleep disorders, medication effects, systemic illness, and, rarely, early stages of a dementing disorder. As in other areas of medicine, prevention is the key to reducing the long-term burden of more severe forms of age-related cognitive loss. There is mounting evidence that cognitive aging can be mitigated or remediated by reducing cardiovascular risk factors, engaging in mental and physical activity, maintaining an active social network, and eating a balanced, nutritious diet that includes whole fruits and vegetables, nuts, and fish.

Summary cognitive function and the menopausal transition

There is a wealth of evidence from basic science studies that ovarian steroids regulate critical neurobiological determinants of memory and other cognitive functions. Recent evidence from a prospective study suggests that menopausal stage and hormone use impact cognitive performance. Perimenopausal women did not show the improvements in processing speed and verbal memory with repeated testing that were evident in premenopausal and postmenopausal women, and HT use before the final menstrual period was associated with better processing speed and verbal memory. Clinical trials show some benefit for estrogen-alone therapy on verbal memory in younger postmenopausal women (<65 y) but modest decreases in verbal memory for both younger and older women taking CEE/MPA. Age, baseline cognitive performance, and formulation of HT may modulate the effects of HT on memory function. Pharmacological challenge studies show that estradiol seems to improve cholinergic function. This effect seems to be age specific, with younger women showing a benefit but older women showing no benefit or impairment. Progesterone counteracts the beneficial effects of estrogen on attention and speed after cholinergic blockade. Clinical studies demonstrate that serious cognitive disorders such as Alzheimer disease are rare during the perimenopausal and early postmenopausal stages. Regardless of age, however, a clinical evaluation should be conducted if cognitive declines are associated with a significant decline in daily function.

Clinical and research priorities

Routine evaluation of depressive symptoms in perimenopausal women is warranted by the finding that the perimenopausal stage represents a window of vulnerability to mood disorders. Routine evaluation is made easy in the clinical setting by the availability of quick and validated screening tools such as the PHQ-9, which is accessible online and is free of charge. By implementing routine screening for depression into their clinical practice, menopausal health practitioners can play a primary role in the prevention and treatment of depressive disorders in perimenopausal women. A similar screen for cognitive disorders is not warranted by the available data, as the risk of cognitive dysfunction is rare and the magnitude of cognitive change across the transition is small.

An important priority for future investigations is identifying which women may be at risk for cognitive and mood disorders across the menopausal transition. Clinical trials could then target those phenotypes. History of a depressive disorder, severe climacteric symptoms, and psychosocial stressors increase risk for depression at this life stage, but the risk is evident even when those factors are controlled. Estradiol levels do not predict who is at risk for depression, but estradiol has been shown to be effective in treating perimenopausal depression. Genetic studies may prove helpful in identifying vulnerable women and guiding future intervention trials. Nonhuman primate studies suggest that depression may be a risk factor for CHD, so early intervention may prevent CHD. Women with a large number of physiological hot flashes seem to be vulnerable to memory dysfunction, but future studies are needed to further characterize this relationship.

Issues related to the measurement of cognitive change in midlife complicate our understanding of the magnitude of cognitive changes in the menopausal transition. Prospective studies require repeated assessments with standardized neuropsychological tests, and midlife women show significant practice effects on such tests. Therefore, studies require large sample sizes and the use of memory tests on which it is difficult to reach maximum performance. Neuroimaging studies, including resting positron emission tomography and functional magnetic resonance imaging during cognitive tasks, may be helpful in identifying biomarkers of cognitive vulnerability in midlife women. To date, there are no prospective neuroimaging studies of women as they transition through menopause. Pharmacological challenge studies have been helpful in identifying which women may be susceptible to mood disorders during the menopausal transition, and this approach may also help to identify subsets of women who are especially susceptible to cognitive dysfunction.

Through continued research, we can better identify who is at risk for depressive symptoms and cognitive decline in the menopausal transition and guide treatment decisions.