Despite almost two decades of research, it is not yet clear how hormone replacement therapy affects a woman’s risk of getting Alzheimer’s disease. The curators of the AlzRisk database, hosted by Alzforum, undertook a review of what the literature on this controversial topic amounts to. Released on 27 October 2011, their meta-analysis of seven observational studies reveals a modest protective effect from estrogen or estrogen plus progesterone therapy. Women who reported using any estrogen had about 20 percent less risk of developing AD than women who used none, curator Jennifer Weuve at Rush University, Chicago, Illinois, told ARF. However, the authors, who also include Deborah Blacker at Massachusetts General Hospital, Boston, and Jacqueline O’Brien at the Harvard School of Public Health, hasten to add significant caveats. No randomized controlled trials met AlzRisk’s inclusion criteria, because the trials considered only risk for total dementia and did not look specifically at AD risk. Randomized controlled trials have shown that estrogen therapy begun late in life increases the risk for dementia, breast cancer, stroke, and heart disease (see ARF related news story). The AlzRisk results, which are comprehensively discussed here, highlight the stark differences between positive epidemiologic and negative trial data, and the need for better answers.

A possible explanation for this difference, favored by many scientists, is that hormone replacement therapy must be initiated at menopause to be protective (see, e.g., ARF Live Discussion). Two large trials, the Kronos Early Estrogen Prevention Study (KEEPS) and the Early Versus Late Intervention Trial With Estradiol (ELITE), are currently testing this “critical window” hypothesis, with data expected to come out over the next two years.

Other factors may account for the divergent results obtained from randomized, placebo-controlled clinical trials and analysis of epidemiological data, Weuve told ARF. In the past, women who took hormone replacement therapy were more likely to be well educated and health conscious than those who did not. This “healthy user bias” might have led to better cognitive outcomes among the group using hormones. Moreover, there are many different hormone formulations and ways of taking them: estrogen alone, or estrogen plus progesterone, higher versus lower doses, continuous versus sequential dosing, oral versus transdermal administration. Many epidemiological studies do not include these data, or if others do, sample sizes are too small to get meaningful data from comparing subgroups, Weuve said. This lack of data makes it difficult to make specific public health recommendations.

Francine Grodstein at Brigham and Women’s Hospital, Boston, wrote to ARF: “The meta-analysis was conducted across studies of ‘any’ hormone use versus no use…. However, ‘any’ hormone use is a construct used in research studies, and has no clinical relevance. One cannot tell a woman to take hormone therapy ‘any’ time; she must know when to take it, what to take, and when to stop taking it. Since such information is largely lacking from the literature, we cannot make any meaningful conclusions regarding the relation of hormone therapy to AD risk” (see full comment below).

Answering these questions will require large studies with well-characterized populations, Weuve agreed. “Right now it’s pretty clear that a 65-year-old woman who has never used estrogen should not start it,” Weuve noted, but other issues remain murky. The North American Menopause Society currently recommends against hormone replacement therapy to prevent dementia, but allows that short-term hormone use may be appropriate to treat menopausal symptoms.

ARF invites the scientific community to comment on this latest addition to the AlzRisk database.—Madolyn Bowman Rogers.


  1. The recent AlzRisk meta-analysis reported a statistically significant decrease in AD risk with “any” use of postmenopausal hormone therapy (HT) compared to no use across seven epidemiologic studies. Unfortunately, this result has many caveats. The primary issues are outlined below.

    1. The meta-analysis was conducted across studies of “any” hormone use versus no use, because there have been too few studies with additional detail to yield meaningful results. However, “any” hormone use is a construct used in research studies, and has no clinical relevance. One cannot tell a woman to take HT “any” time; she must know when to take it, what to take (e.g., estrogen alone or combined with progestin, oral, or transdermal, higher or lower doses, etc.), how long to take it, and when to stop taking it. Since such information is largely lacking from the literature, we cannot make any meaningful conclusions currently regarding the relation of HT to AD risk.

    2. The finding was borderline statistically significant, with a wide confidence interval (95 percent confidence interval 0.523-0.995); the upper bound of the confidence interval was only just below 1.0. This makes the results relatively difficult to interpret.

    3. Confounding is a very important problem in observational studies of hormone therapy. It has been demonstrated in studies of general populations that hormone users are healthier than women who do not take hormone therapy. Thus, careful control for confounding is critical in research on hormone use and risk of chronic diseases. Without extensive control for confounding, studies risk overestimating benefits of hormone therapy. However, many of the Alzheimer's disease studies in the meta-analysis did not control for an extensive array of potential confounding factors, and thus may be overestimating observed relations. Since the meta-analysis was borderline statistically significant, it is likely that more careful control for confounding in the studies included would have led to an attenuated and non-significant summary of relative risk.

    4. Finally, and perhaps most importantly, definitive data establish that hormone therapy causes strokes in women. These findings are consistent in the Women’s Health Initiative (1) and in well-conducted observational investigations (2) as well. There is no effect of timing of hormone initiation on stroke risk. In the Women’s Health Initiative (1), and in the Nurses’ Health Study, the largest observational study of HT and stroke (3), increased risks of stroke for women currently taking HT were found regardless of whether women used HT near menopause or many years later. How can we consider an agent that causes strokes as potentially neuroprotective?


    . Postmenopausal hormone therapy and risk of cardiovascular disease by age and years since menopause. JAMA. 2007 Apr 4;297(13):1465-77. PubMed.

    . A prospective, observational study of postmenopausal hormone therapy and primary prevention of cardiovascular disease. Ann Intern Med. 2000 Dec 19;133(12):933-41. PubMed.

    . Postmenopausal hormone therapy and stroke: role of time since menopause and age at initiation of hormone therapy. Arch Intern Med. 2008 Apr 28;168(8):861-6. PubMed.

  2. While hormone therapy is the most effective treatment for vasomotor symptoms associated with menopause, its effect on the risk for Alzheimer’s disease (AD) remains controversial. O’Brien and coworkers provide an excellent overview of the methodological context for the observational prospective findings associated with the use of postmenopausal hormone therapy and its risk on the development of AD. The authors clearly and succinctly identify the caveats that have led to conflicting interpretations between results from observational studies and randomized clinical trials. Discrete variables, such as hormone formulation, route of administration, type of menopause (natural vs. surgical), and duration of treatment need to be considered when evaluating the findings from these studies. Moreover, a consensus is building that it is crucial to consider the timing or “critical window,” when hormone therapy is initiated in women relative to the start of menopause. The “critical window” hypothesis proposes that initiating hormone therapy closer to the start of menopause will be most beneficial in reducing the risk for AD by exploiting the “healthy cell bias,” in which healthy cells respond positively to estrogen. Currently, there are two clinical trials, the Kronos Early Estrogen Prevention Study (KEEPS) and the Early Versus Late Intervention Trial with Estradiol, which will provide greater insight into the timing of hormone therapy and its effects on cognition.

    Although the use of exogenous hormones for the treatment of symptoms associated with menopause and its potential effects on AD is an area of active research, the role that aromatase inhibitors (AIs) play in this equation is unknown. Aromatase inhibitors prevent the conversion of testosterone to 17β-estradiol and androsteinedione to estrone by the aromatase enzyme, which is found in various tissues including the hippocampus, a component of a limbic memory circuit, which is affected in AD. Although understudied in the AD field, AIs are routinely given to postmenopausal women to prevent the recurrence of breast cancer. Clinical trials evaluating the effects of AIs on cognition are extremely limited and the data are inconclusive. For example, two small clinical trials suggest that AIs have an adverse effect on hippocampal-dependent learning (Bender et al., 2007; Castellon et al., 2004), while initial results from a larger study found no change in cognitive ability two years into a five-year trial (Jenkins et al., 2008). Although estrogen has been shown to affect cognition in aged animals and to alter the deposition of amyloid pathology in transgenic animal models of AD, there are virtually no data on the effects of AIs upon either cognition or amyloid pathology in a mouse model of AD. Our group recently investigated whether the AI, anastrozole, crossed the blood-brain barrier and altered amyloid pathology in a triple-transgenic mouse model of AD (Overk et al., 2011). Interestingly, we found in male and female triple-transgenic AD mice that anastrozole entered the brain, and that it reduced amyloid-β (Aβ) plaque areas in the female hippocampal complex. At the same time, anastrozole treatment increased intraneuronal Aβ only in females. Although these data lend support to the hypothesis that hormone brain levels have an effect on Aβ plaque pathology, the cognitive consequences of altering brain hormone levels using AIs remain to be determined. Evaluation of the long-term cognitive consequences of manipulating endogenous (AIs) and exogenous (HT) hormone levels will aid in our understanding of the role that hormones play in the risk for AD.


    . Memory impairments with adjuvant anastrozole versus tamoxifen in women with early-stage breast cancer. Menopause. 2007 Nov-Dec;14(6):995-8. PubMed.

    . Neurocognitive performance in breast cancer survivors exposed to adjuvant chemotherapy and tamoxifen. J Clin Exp Neuropsychol. 2004 Oct;26(7):955-69. PubMed.

    . Effects of anastrozole on cognitive performance in postmenopausal women: a randomised, double-blind chemoprevention trial (IBIS II). Lancet Oncol. 2008 Oct;9(10):953-61. PubMed.

    . Effects of aromatase inhibition versus gonadectomy on hippocampal complex amyloid pathology in triple transgenic mice. Neurobiol Dis. 2012 Jan;45(1):479-87. PubMed.

  3. Response to Alzrisk Meta-Analysis on ERT and AD
    Results from the meta-analysis suggest a moderate protective effect of hormone therapy (HT) on risk of AD. One important limitation of this study was that it focused on the protective effect of HT on AD risk alone, rather than global cognitive decline and dementia. As discussed by O’Brien et al., the current set of criteria used to make a definitive AD diagnosis is much more restrictive than the criteria used a decade ago. It is highly probable that some patients with an AD diagnosis in earlier studies (Brenner et al., 1994; Tang et al., 1996) would not meet the necessary criteria for inclusion if those studies were carried out within the last five years.

    Consideration of the human genetic background and responsiveness to hormone therapy are also important factors. One important predictor of AD risk and outcome is the presence of one or more copies of the ApoE4 allele. To date, ApoE4 is the most significant risk factor known that is associated with increased age of onset and disease severity in late-onset AD. While the human genome produces three different ApoE protein isoforms from a single gene, rodent genomes encode a single ApoE isoform whose functionality is a hybrid between human ApoE3 and human ApoE4. The Cache County (Zandi et al., 2002) and WHICAP (Tang et al., 1996) studies in the meta-analysis included ApoE genotype as a variable, but did not detect any significant effects. Conversely, data from our laboratories and others suggest that individuals or mice with one or more ApoE4 alleles may be less responsive to estrogen therapy than those with ApoE3 or ApoE2 alleles (Brown et al, 2008; Wang et al., 2006; Kang and Grodstein, 2010; Burkhardt et al., 2004; Yaffe et al., 2000). Recently released data on the ApoE genotypes of participants in the Women’s Health Initiative (WHI) will allow investigators to assess whether significant associations exist among cognitive decline, ApoE genotype, and assignment to estrogen replacement therapy in the form of Premarin (conjugated equine estrogens). Several analyses of these associations are currently underway by several investigators.

    The continuing debate on whether animal models, in particular transgenic animal models, are appropriate models to study AD is also emphasized in the meta-analysis. Comparisons between the animal models that show protection and clinical studies that show moderate or no protective benefit clearly demonstrate that numerous factors contribute to differences among animal models, observational studies, and clinical trials. An appropriate interpretation of results in animal models must consider several variables, including sex, chronological age, and reproductive age. Estrogens have been shown to be protective in some animal models of AD where the mice are typically young or middle-aged adults subjected to a surgical menopause and estrogen replacement (Carroll et al., 2007). Conversely, other studies have shown that estrogens lose their neuroprotective function in reproductively senescent, aged rodent models (Sohrabji and Bake, 2006). The importance of both chronological age and reproductive age underscores the importance of using aged animal models for preclinical studies before extrapolating to clinical practice. Continued funding as well as additional funding for studies that incorporate young, middle-aged, and aged mice matched for chronological and reproductive age are imperative for basic science researchers to uncover meaningful, clinically relevant cellular and molecular mechanisms underlying AD and other aging-related disorders.

    See also:

    O'Brien J, Weuve J, Blacker D. Hormone therapy. The AlzRisk Database. Alzheimer Research Forum. Accessed 10/28/11.


    . Postmenopausal estrogen replacement therapy and the risk of Alzheimer's disease: a population-based case-control study. Am J Epidemiol. 1994 Aug 1;140(3):262-7. PubMed.

    . The APOE4 genotype alters the response of microglia and macrophages to 17beta-estradiol. Neurobiol Aging. 2008 Dec;29(12):1783-94. PubMed.

    . Oestrogen replacement therapy may improve memory functioning in the absence of APOE epsilon4. J Alzheimers Dis. 2004 Jun;6(3):221-8. PubMed.

    . Progesterone and estrogen regulate Alzheimer-like neuropathology in female 3xTg-AD mice. J Neurosci. 2007 Nov 28;27(48):13357-65. PubMed.

    . Postmenopausal hormone therapy, timing of initiation, APOE and cognitive decline. Neurobiol Aging. 2012 Jul;33(7):1129-37. PubMed.

    . Age-related changes in neuroprotection: is estrogen pro-inflammatory for the reproductive senescent brain?. Endocrine. 2006 Apr;29(2):191-7. PubMed.

    . Effect of oestrogen during menopause on risk and age at onset of Alzheimer's disease. Lancet. 1996 Aug 17;348(9025):429-32. PubMed.

    . Activation of estrogen receptor alpha increases and estrogen receptor beta decreases apolipoprotein E expression in hippocampus in vitro and in vivo. Proc Natl Acad Sci U S A. 2006 Nov 7;103(45):16983-8. PubMed.

    . Estrogen use, APOE, and cognitive decline: evidence of gene-environment interaction. Neurology. 2000 May 23;54(10):1949-54. PubMed.

    . Hormone replacement therapy and incidence of Alzheimer disease in older women: the Cache County Study. JAMA. 2002 Nov 6;288(17):2123-9. PubMed.

  4. This is an excellent review on hormone replacement therapy/estrogen replacement therapy (HRT/ERT) from clinical data to basic reports, with, in particular, extensive, detailed information about the formulation, dosage, duration, route of administration, and age at initial usage of hormone treatments. While multiple factors might be implicated in estrogen deficiency-related risk for AD in females, our previous report suggested that brain estrogen deficiency causes an early and more severe AD-like pathology in female transgenic mice (Yue et al., 2005). In addition, our recent unpublished data also suggest that endogenous brain levels of estrogen might determine the response to ERT in AD female transgenic mice (manuscript in submission). We used a gene targeting approach to delete aromatase, an estrogen synthase, in Alzheimer’s model mice (APP23) to create estrogen-deficient animals (APP/Ar+/-). We discovered that early treatments with 17β-estradiol (E2) and the phytoestrogen genistein (Gen) in these APP/Ar+/- mice significantly reduced brain amyloid plaque formation and inhibited β-secretase (BACE1) mRNA and protein expression. No effect of any ERT on AD pathology was found in ovariectomized female APP23 mice (APP/OVX), which have normal brain estrogen levels, or sham-operated APP23 mice. Our studies indicate that early and long-term usage of E2 and Gen treatments prevent AD pathologies in a manner dependent on endogenous estrogen in the brain in females. Our findings support the “critical therapy window” theory, which suggests that the initiation of hormone therapy at or soon after menopause is beneficial or neuroprotective, whereas starting treatment at a time remote from menopause may delete the beneficial effect or even be harmful. While there is no clear understanding of why there is a “window” and what are the underlying mechanisms of the “window,” it will be important to investigate specific genes or protein-protein interactions involved in responding to HRT/ERT under various conditions such as age, formula, dosage, and duration of the treatment.


    . Brain estrogen deficiency accelerates Abeta plaque formation in an Alzheimer's disease animal model. Proc Natl Acad Sci U S A. 2005 Dec 27;102(52):19198-203. PubMed.

  5. This analysis of observational studies suggesting a decreased risk of Alzheimer’s disease with hormone treatment is well carried out, and follows and confirms most of the earlier reviews investigating this topic. In sharp contrast to these promising findings from observational data, and also from animal and cell culture studies and several small treatment studies (1), results from large, well-controlled trials published after the turn of the twenty-first century suggested that treatment did not reverse cognitive decline in women with dementia (2-4), and that dementia symptoms worsened after 12 months of treatment (2). The much-quoted Women’s Health Initiative Memory Study (WHIMS) also showed that estrogen treatment was not indicated to maintain cognitive function in older women without dementia (5,6), and could increase the risk for dementia (7,8).

    On the other hand, pooled analyses of 11 studies from eight European countries (EURODEM) showed that AD, the most common form of dementia, is more common in older women than it is in men after the age of 75 years (9). There could be several reasons for this gender difference. In midlife, around the sixth decade of life, women experience a significant drop in sex steroids after menopause, the final menses. Estradiol (the most potent estrogen) levels can fall by as much as a factor of 10 or more (10,11). In the 1990s, estrogens were thought to protect against dementia and cardiovascular disease. As such, cardiovascular disease was the primary outcome measure in most of the large treatment trials, such as the Heart and Estrogen/progestin Replacement Study (HERS) (13) and the Women's Health Initiative (WHI) (14). Both showed no positive effects of combined hormone treatment on cognition (which had been included as a secondary outcome). It had been hypothesized that at least some of the positive effects of estrogen on the brain would be mediated through their positive effects on lipid profiles, vasodilation, and improved cerebral perfusion. However, the large trials, including HERS and WHI, with older women showed that the risk for several vascular outcomes, such as coronary heart disease and stroke, were also increased with conjugated equine estrogen (CEE) use, in particular when this was combined with a medroxyprogesterone acetate (MPA) (7). In the late 1980s, MPA was already known to counteract estrogen’s positive effect on vascular systems (15). Other progestogens did not seem to have such negative effects, and the use of MPA in these studies was much criticized in some academic and clinical circles.

    However, in the U.S., most women with an intact uterus would have probably been prescribed CEE and MPA in the form of Premarin and Prempro. Some argued that if CEE, with or without MPA, had not been an appropriate choice, the observational U.S.-based studies should also not have shown positive associations of hormone use with prevention of dementia and cardiovascular disease, as is also reflected in the current review.

    Age and the "Window of Opportunity/Critical Period" Hypothesis
    The negative association of estrogens with dementia symptoms and increase in dementia risk (2,7,8) mentioned above were mainly reported in women who were over the age of 65 years, for whom hormones for menopausal complaints normally would not have been indicated. This age group was chosen in those studies because the incidence of vascular disease and dementia is too low in the younger middle-aged women (e.g., those undergoing menopause and for whom, normally, treatment might be indicated) to obtain sufficient power to detect the suspected lowering of morbidity risk with hormone treatment. Cardiovascular disease and AD have substantial overlap in risk factors, but it has become increasingly clear that it is important to target these risk factors in midlife to prevent both cardiovascular disease and dementia in later life (12). Whether use of estradiol or CEE in midlife will act to protect against both cardiovascular disease and AD in later life is still under investigation. Several large trials, such as the Kronos Early Estrogen Prevention Study (KEEPS), for women who are closer to the natural age of menopause, are now ongoing to establish whether this is indeed the case.

    The "Window of Opportunity" or "Critical Period of Treatment" hypothesis (16) has substantial support from animal research (17) and smaller treatment studies (18), which suggested that estradiol should be given close to the time of menopause to exert positive effects on the brain. If the interval is too long (as it would have been in WHIMS [7,8], which included women who would have been, on average, at least 15 years past the age of menopause), estrogens were seen to no longer promote positive effects. This effect was seen in mouse models where the time interval past ovariectomy determined sensitivity to estradiol treatment on memory tasks (17). Indeed, data from the WHIMS showed that women who had taken hormones in midlife had less risk for dementia in later life (19). Similarly, other cohorts, such as the Cache County cohort (20), showed that former, but not current, older hormone users had less risk of dementia, suggesting that there may be a critical period of time (i.e., around the age of menopause) during which hormone use should be initiated to be protective against dementia later in life.

    Explaining Negative Effects of Estrogens on the Older Brain: The Healthy Cell Bias Hypothesis
    However, the "Window of Opportunity" hypothesis does not explain some of the negative effects of estrogens in doubling dementia risk later in life, as occurred in WHIMS. A possible theoretical framework for this has been posed by Roberta Brinton-Diaz and is called the "healthy cell bias" hypothesis (21). This model explains the negative effects of estrogens on brain cells in older women, while supporting initial positive effects in younger, more recently menopausal women. Her cell culture research shows that healthy neurons in the hippocampus and cortex function optimally in an estrogenic environment. Estrogen plays an important role in calcium signaling and mitochondrial function to enhance glucose metabolism, which is shown to be affected in AD. However, neurons which have been exposed to neurodegenerative insults, and which are subsequently undergoing the type of pathological changes described above, are actually detrimentally affected by estrogen-induced signaling onto mitochondria, which ultimately exacerbates and accelerates neuronal degeneration (21).

    As this type of pathogenic pathway is perhaps more likely to occur with older age, given fewer defense and regenerative mechanisms, this might explain the initial positive effects of estrogens in midlife and the negative effects of elevated estrogen levels on brain cells later in life. This theory of age and estrogen levels interacting with pathology is attractive, but it is also, by itself, insufficient to explain the findings of all studies combined. For instance, several observational studies found that current hormone users who were over the age of 60 years had less risk of dementia (1). On the other hand, one of the largest studies to investigate this association in 13,807 nurses over the age of 70 years (the Nurses’ Health Study) found that long-term users of estrogen, with or without a progestogen, had substantially increased risk for cognitive decline, particularly when use was initiated at an older age (22). These data would substantiate the "healthy cell bias" hypothesis with an increased risk of treatment for older women. So how can we explain protective associations of estrogen use with decreased dementia risk and better cognitive function in the other cohorts (1)?

    Bias in Observational Studies Explaining the Association Between Hormone Use and AD Risk
    It has been suggested that observational data on estrogen use and dementia were contaminated by two important forms of bias: 1) a recall bias and 2) a "healthy-user" bias. By the very nature of their disorder, women with dementia would be less likely to recall their hormone use. Memory loss, an inability to correctly recall recent events, is the most common cognitive deficiency in most dementias. Indeed, one study (23) showed that women with dementia were half as likely as controls to accurately report hormone use when this was compared to pharmacy records. Most earlier observational studies (1) that reported reduced risk of dementia with hormone use had either relied on self-report or carers' report (which is, in all likelihood, equally unreliable). One early study using pharmacy records (24) had already reported no significant association between hormone use and AD. Later observational studies using medical records (23), including a study using the U.K.-based General Practice Research database (25), also reported no significant associations between dementia risk and hormone use. These findings combined would give credibility to the argument that faulty recall or a recall bias may be related to the reduced risk in the reviewed cross-sectional studies.

    However, this would have presumably played little to no role in longitudinal studies which followed women who did not have dementia at baseline and only later developed this at follow-up (unless they already displayed some degree of deficiency in recall at that time, e.g., because they had MCI and were in the early stages of the disease, but had not yet been diagnosed with dementia). For instance, one follow-up study reported a reduced risk for incident AD with hormone treatment at follow-up in women without dementia at baseline, but, as mentioned earlier, only in past and not current hormone users (20). However, the numbers of AD cases were small in this study for such sub-analyses. Most importantly, despite control for potential confounds, these longitudinal studies can never fully take into account the many ways in which women who choose to take hormones are different before they use hormones when compared to women who choose not to use hormones for menopausal complaints. While some factors might be considered as "noise," this is not the case when differences in, for instance, lifestyles also affect cognitive function and decline, and may be closely tied to the choice of hormone use during the menopausal transition.

    There are at least two studies that followed women from before the onset of menopause and through subsequent hormone use, which allow investigation of this "healthy user" bias. One Danish study found that women who would choose to use hormones already had better cognitive function before the use of these treatments (26). Another study found that women who would choose to use estrogens for menopausal complaints already had healthier lifestyles and overall better health in the years preceding the choice of hormone use for menopausal complaints (27). Their lower body weight, fasting glucose, and cholesterol levels and lower blood pressure may have been related to the fact that they were also less likely to smoke and were more active. These factors are traditionally all associated with lower risk for cardiovascular disease. Several studies which followed participants from midlife to older age found that protection against cardiovascular disease (by reducing risk factors such as obesity, smoking, high cholesterol, and blood pressure, leading a healthier lifestyle) also led to a lower risk of dementia in later life (for a review see [12]). The observational findings thus seem to suggest that taking hormones during the menopausal transition was part of a chosen healthy lifestyle. It may be the case that hormone use, which was common in these cohorts, was individually related to these lifestyle variables within groups of individuals in the cohort, in which each in turn contributed to cognitive health. Particular patterns of combinations, each leading to the reduced risk of dementia outcome, could all have in common that many women at the time would have thought that hormone treatment was part of a healthy lifestyle. Hormone use was actively promoted as such in many popular articles, as a panacea for many ailments. Hormone use then acts statistically as an "umbrella variable" explaining the variance in the analyses because of its associations with all other variables and their patterns, rather than having a strong association with reduction of risk for cognitive decline because of its presumed biological activity on the brain. These available data do not allow us to disentangle the possible additive effects of estrogen use on top of other healthy lifestyles. For this we would need to exactly match users to non-users who had similar healthy lifestyles and fewer risk factors.

    However, there may be factors that cannot be easily controlled for, which are independently associated with both risk for dementia and the choice of taking hormone replacement therapy. For instance, many cohorts found that hormone users had often received better education (1). Perhaps this allowed them to make informed choices about the effect of a healthy lifestyle and screening for risk factors on disease prevention. Importantly, one study showed that children with higher IQs also had less risk of dementia after the age of 72 years (28), and lower midlife blood pressure (29). It could be hypothesized that this high IQ, which is strongly related to educational level obtained, reflects a healthy childhood with many intellectual and biological resources, which later in life may delay the onset of clinical dementia symptoms. For instance, if one has a larger vocabulary, word-finding problems are less apparent, one can choose to use different strategies, etc., and/or this effect may be related to the actual physical presence of more abundant connections between neurons, and/or a system that is more robust to insults and/or has had fewer exposures to health risks, or a combination of all of the above (30). Indeed, those with high levels of education seem to have a later onset of dementia symptoms, but a more rapid decline (31), perhaps reflecting the presence of these resources.

    The question remains whether restoration or elevation of estrogen and testosterone levels in midlife could also help protect women against later-life development of dementia and how long this treatment should be given to sustain later beneficial effects. Apart from animal and basic research data supporting this hypothesis, the majority of observational human data indicates that use of estrogens for menopausal complaints protects against the development of dementia and, overall, about halves the risk of AD (1). However, observational data may be tainted by different forms of bias. This could also be the case for observational studies investigating hormone levels and cognition. In general, studies investigating sex steroid levels and cognitive function or dementia have found contrasting associations (11), which may, at least partly, be related to a lack of assay sensitivity and/or adequate control for confounds and mediators (32). If sex steroids would have major effects on the brain, it would also be expected that the menopausal transition (which, as said, is characterized by a substantial drop in sex steroids) would show a sharp and accelerated deterioration in cognitive function. However, several observational studies have not been able to objectify this phenomenon using standard neuropsychological tests (see Thilers, 2009, for a review). That sex steroid levels probably do not play a determining role in cognitive decline may also be reflected by the fact that in several cohorts, older men without dementia had worse cognitive function than age- and education-matched women. In these cohorts, despite having higher levels of sex steroids, men did not seem to be protected against cognitive decline (see [33] for a review). This contrasts with the data on AD sex differences, but signifies again the possibility of healthy user bias and survivor bias in those data.

    Duration of Treatment Hypothesis
    In seeming contrast with the Window of Opportunity hypothesis, duration-dependent positive effects of estradiol treatment on verbal memory were maintained in women with dementia in the rigorous Cochrane meta-analyses (34). Limited positive effects of both CEE on global cognition and transdermal E2 on verbal memory, however, were only significant for up to two to four months of treatment and not thereafter. On the other hand, the meta-analyses also showed that the data were heterogeneous. Significant effects were not seen on all tests measuring the same functions, nor were these seen in all studies employing the same tests. This could be the case for a variety of reasons (test specificity and sensitivity, slightly different modes of employing the same test, different testers or treatments, etc.).

    Importantly, the women in these studies were usually over the age of 65 years, which would be on average at least 15 years past their age at menopause. A similar pattern was seen in the Cochrane review of women without dementia, who were on average 15 years younger. Only duration-limited positive effects of estradiol were seen for up to two to three months. From these meta-analyses, it would follow that estrogens only have very time-limited positive effects on specific cognitive tests for a maximum of a couple of months. The largest effect sizes were seen on verbal memory in the relatively younger surgical menopausal women who underwent oophorectomy during the study (18,35). This has also been reported in other studies which followed women before and in the weeks after oophorectomy (36,37). These effects of oophorectomy could be reversed completely by a high dose of estradiol (18). One line of reasoning would be that the sudden withdrawal of estrogens leads to a decline in biological parameters sustaining brain health.

    One of the problems in the Cochrane meta-analyses was that not all studies with large effect sizes, which showed positive effects of oral estradiol in symptomatic, recently menopausal women, could be included (e.g., see [38, 39]). In addition, because of recalculation of the mean difference and standard deviation (SD) of that mean difference (often based on graph information which may not have been reliable), when testing calculated data against existing data, the lead author found that this calculation in smaller studies could actually overestimate the SD sometimes even up to a factor of 2. This would lead to a lack of replication of significance in studies which did earlier report significant results. This technique could lead to type 2 error, which is, unfortunately, perhaps insufficiently recognized by the Cochrane system.

    Qualitative analyses of studies showed that oral estradiol (with a progesterone other than MPA) or intramuscular estradiol both could affect verbal memory and some other tests of executive function and attention mainly in symptomatic, relatively young recently menopausal women (18,35,38-40), but again for only up to six months. Most of these studies were not done for more than a couple of months, which makes it difficult to translate this into clinical advice for women on when to stop treatment. Most of the studies also did not find that transdermal estradiol had a positive effect on cognition in younger postmenopausal women without dementia, although limited effects were reported in older women, similar to the dementia studies, again for a limited period of time (41).

    Reversal of Positive Effects After One Year With CEE and Estradiol, Which May Result in Negative Effects on Cognition With Longer Treatment in Older Women
    The negative effects of CEE and MPA on cognition in older women already after one year of treatment in the WHIMS and HERS have been described above. Importantly, evidence from a few studies has also shown a reversal of positive effects of estradiol after one year in both women with (42) and without dementia (39).

    Because of the drop in cognitive function seen after surgical menopause, some scientists have argued that estradiol should continue to be given for longer periods of time to women who undergo oophorectomy. However, two studies reported that women who were still using estrogens approximately 10 years after having undergone oophorectomy actually had worse cognition than those who were past users or non-users at the time of testing (43,44). This is in apparent contrast with data from the Mayo Clinic (45), which showed that women who had undergone oophorectomy and were not treated with hormones had an increased risk of dementia. This study found that the earlier the age of surgery without subsequent hormone treatment, the greater the risk. However, in the Mayo Clinic study, this protective effect of estrogen treatment was measured up to the natural age of menopause, around the age of 50. Similarly, the ARIC study showed that younger (Summarizing the data, while women often complain of cognitive deficiency, most objective tests do not show an abrupt decline after natural menopause. This may be related to insensitive cognitive tests used, but also to possible modification of body mass which may protect against a decline in estrogens, but which is also related to experiencing more menopausal symptoms. More symptoms are also reported after surgical menopause, which has been shown in several studies to be related to an abrupt drop in memory functions, and could be at least temporarily reversed by estradiol. How much of a role symptoms play in mediating effects of estradiol on memory is unclear, as other studies of natural menopausal women which showed positive treatment effects of estradiol and a progestogen had included highly symptomatic women. Our recent meta-analyses in Maturitas (2010) showed that alleviation of symptoms did not explain positive effects of hormone use on cognition. Either way, effects of estradiol in improving cognition seem not to be sustained after two to four months, and may reverse after a year. Anecdotal data from ob-gyn clinics suggest that many women stop taking hormone treatment for menopause before one year of treatment finishes, but this needs further confirmation to investigate whether even such short-term treatment is associated with positive effects in later life. Negative effects of conjugated equine estrogens in the largest trial, the Women’s Health Initiative Memory Study, were seen after one year of treatment. This was the case for both dementia risk, as well as for not showing cognitive improvement in the same manner as was seen in women using placebo. No earlier measurements had been done in this trial to suggest potential earlier (e.g., at two to four months) positive effects. Limited evidence also suggests similar negative effects of estradiol after one year of treatment in women with and without dementia in much smaller trials. If this is the case, we can predict that KEEPS, which only has its first cognitive assessment after one year of treatment, will show no (in relatively recently menopausal women) or negative (in older women) effects of both CEE and estradiol.

    In addition, the meta-analyses (34,41) did report positive effects of estrogen treatment (including CEE) in older women with dementia for a short duration of time, which negates the "Critical Period" hypothesis. However, treatment effects of estradiol and CEE in women with dementia did not exceed four months, similar to those of estradiol in the younger women without dementia. With treatment over the course of many years in older women, continuous exposure to high estrogen levels may thus confer risks. The healthy cell bias hypothesis (47) offers an explanation for this, stating that neurons which are undergoing pathological change are detrimentally affected by an estrogenic environment, which accelerates their demise. This type of pathological change related to dementia is perhaps more likely to occur in older women, which is why negative effects of long-term hormone treatment may be seen in this group.

    The relatively small effects of sex steroids on the aging brain contrast with the earlier enthusiasm about estrogen’s potential to act as a prophylactic agent to prevent dementia and vascular disease in articles published in the late twentieth century. That other factors may overrule the smaller effects of sex steroids, however, does not mean that they should be ignored all together. Countless articles describing estrogen’s protective effect in cell culture and animal studies do still indicate that estrogens may yet have a role to play in prevention of dementia.

    This comment was based on the following earlier published articles: Hogervorst and Bandelow, 2009 and Hogervorst and Bandelow, 2010.

    The author has no vested interest and currently does not act as a consultant for the pharmaceutical industry involving estrogens, although consultancy was received from Memryte to investigate the role of gonadotropin lowering in reducing AD symptoms.

    See also:

    Gibbs, R., Animal studies that support estrogen effects on cognitive performance and the cholinergic basis of the critical period hypothesis, in Hormones, Cognition and Dementia, E. Hogervorst, et al., Editors. 2009, Cambridge University Press: Cambridge. p. 45-54.

    Hogervorst, E., The short-lived effect of HRT on cognition function, in The Effects of Estrogen on Brain Function., N. Rasgon, Editor. 2006, Johns Hopkins Press: Baltimore. p. 46-78.


    . The nature of the effect of female gonadal hormone replacement therapy on cognitive function in post-menopausal women: a meta-analysis. Neuroscience. 2000;101(3):485-512. PubMed.

    . Estrogen replacement therapy for treatment of mild to moderate Alzheimer disease: a randomized controlled trial. Alzheimer's Disease Cooperative Study. JAMA. 2000 Feb 23;283(8):1007-15. PubMed.

    . Effects of estrogen on cognition, mood, and cerebral blood flow in AD: a controlled study. Neurology. 2000 Jun 13;54(11):2061-6. PubMed.

    . Estrogen for Alzheimer's disease in women: randomized, double-blind, placebo-controlled trial. Neurology. 2000 Jan 25;54(2):295-301. PubMed.

    . Conjugated equine estrogens and global cognitive function in postmenopausal women: Women's Health Initiative Memory Study. JAMA. 2004 Jun 23;291(24):2959-68. PubMed.

    . Effect of estrogen plus progestin on global cognitive function in postmenopausal women: the Women's Health Initiative Memory Study: a randomized controlled trial. JAMA. 2003 May 28;289(20):2663-72. PubMed.

    . Estrogen plus progestin and the incidence of dementia and mild cognitive impairment in postmenopausal women: the Women's Health Initiative Memory Study: a randomized controlled trial. JAMA. 2003 May 28;289(20):2651-62. PubMed.

    . Conjugated equine estrogens and incidence of probable dementia and mild cognitive impairment in postmenopausal women: Women's Health Initiative Memory Study. JAMA. 2004 Jun 23;291(24):2947-58. PubMed.

    . Rates and risk factors for dementia and Alzheimer's disease: results from EURODEM pooled analyses. EURODEM Incidence Research Group and Work Groups. European Studies of Dementia. Neurology. 1999 Jan 1;52(1):78-84. PubMed.

    . Menopause transition: Annual changes in serum hormonal patterns over the menstrual cycle in women during a nine-year period prior to menopause. J Clin Endocrinol Metab. 2004 Jun;89(6):2763-9. PubMed.

    . Serum levels of estradiol and testosterone and performance in different cognitive domains in healthy elderly men and women. Psychoneuroendocrinology. 2004 Apr;29(3):405-21. PubMed.

    . Maintaining cognitive health in elderly women: an invited review. Women's Health, 2009. in press.

    . Effect of postmenopausal hormone therapy on cognitive function: the Heart and Estrogen/progestin Replacement Study. Am J Med. 2002 Nov;113(7):543-8. PubMed.

    . The Women's Health Initiative Memory Study (WHIMS): a trial of the effect of estrogen therapy in preventing and slowing the progression of dementia. Control Clin Trials. 1998 Dec;19(6):604-21. PubMed.

    . Ovarian hormones and the circulation. Maturitas. 1990 Sep;12(3):287-98. PubMed.

    . Hormone therapy and Alzheimer's disease: benefit or harm?. Expert Opin Pharmacother. 2004 Feb;5(2):389-406. PubMed.

    . Estrogen and/or androgen replacement therapy and cognitive functioning in surgically menopausal women. Psychoneuroendocrinology. 1988;13(4):345-57. PubMed.

    . Prior use of hormone therapy and incident AD in the WHIMS. Neurology, 2007. 68(S1): p. A205.

    . Hormone replacement therapy and incidence of Alzheimer disease in older women: the Cache County Study. JAMA. 2002 Nov 6;288(17):2123-9. PubMed.

    . The healthy cell bias of estrogen action: mitochondrial bioenergetics and neurological implications. Trends Neurosci. 2008 Oct;31(10):529-37. PubMed.

    . Postmenopausal hormone therapy and risk of cognitive decline in community-dwelling aging women. Neurology. 2004 Jul 13;63(1):101-7. PubMed.

    . Prevalence of dementia in users of hormone replacement therapy as defined by prescription data. J Gerontol A Biol Sci Med Sci. 2002 Aug;57(8):M532-8. PubMed.

    . Postmenopausal estrogen replacement therapy and the risk of Alzheimer's disease: a population-based case-control study. Am J Epidemiol. 1994 Aug 1;140(3):262-7. PubMed.

    . Postmenopausal estrogen replacement therapy and the risk of Alzheimer disease. Arch Neurol. 2001 Mar;58(3):435-40. PubMed.

    . The influence of hormone replacement therapy on the aging-related change in cognitive performance. Analysis based on a Danish cohort study. Maturitas. 2002 Jul 25;42(3):209-18. PubMed.

    . Prior to use of estrogen replacement therapy, are users healthier than nonusers?. Am J Epidemiol. 1996 May 15;143(10):971-8. PubMed.

    . Childhood mental ability and dementia. Neurology. 2000 Nov 28;55(10):1455-9. PubMed.

    . Childhood mental ability and blood pressure at midlife: linking the Scottish Mental Survey 1932 and the Midspan studies. J Hypertens. 2004 May;22(5):893-7. PubMed.

    . Brain ageing and dementia: what makes the difference?. Br J Psychiatry. 2002 Nov;181:369-71. PubMed.

    . Predictors of progression of cognitive decline in Alzheimer's disease: the role of vascular and sociodemographic factors. J Neurol. 2009 Aug;256(8):1288-95. PubMed.

    . Measuring serum oestradiol in women with Alzheimer's disease: the importance of the sensitivity of the assay method. Eur J Endocrinol. 2003 Jan;148(1):67-72. PubMed.

    . Hormone replacement therapy to maintain cognitive function in women with dementia. Cochrane Database Syst Rev. 2009;(1):CD003799. PubMed.

    . Effects of estrogen on memory function in surgically menopausal women. Psychoneuroendocrinology. 1992 Oct;17(5):485-95. PubMed.

    . Effect of surgical menopause on cognitive functions. Dement Geriatr Cogn Disord. 2002;13(3):193-8. PubMed.

    . Memory functioning at menopause: impact of age in ovariectomized women. Gynecol Obstet Invest. 1999;47(1):29-36. PubMed.

    . The influence of oestrogens on the wellbeing and mental performance in climacteric and postmenopausal women. Acta Obstet Gynecol Scand Suppl. 1977;64:1-91. PubMed.

    . 1998 Curt P. Richter Award. The effect of hormone replacement therapy on cognitive function in elderly women. Psychoneuroendocrinology. 1999 Jan;24(1):43-68. PubMed.

    . Double-blind, placebo-controlled psychometric studies on the effects of a combined estrogen-progestin regimen versus estrogen alone on performance, mood and personality of menopausal syndrome patients. Arzneimittelforschung. 2001;51(3):238-45. PubMed.

    . Meta-analyses of the effect of hormone treatment on cognitive function in postmenopausal women. Womens Health (Lond Engl). 2007 Mar;3(2):173-94. PubMed.

    . An evaluation of psychological effects of sex hormone administration in aged women. II. Results of therapy after eighteen months. J Gerontol. 1954 Apr;9(2):168-74. PubMed.

    . Trough oestradiol levels associated with cognitive impairment in post-menopausal women after 10 years of oestradiol implants. Psychopharmacology (Berl). 2002 Apr;161(1):107-12. PubMed.

    . Hysterectomy, oophorectomy, and cognitive function in older women. J Am Geriatr Soc. 2002 Jan;50(1):55-61. PubMed.

    . Increased risk of cognitive impairment or dementia in women who underwent oophorectomy before menopause. Neurology. 2007 Sep 11;69(11):1074-83. PubMed.

    . Estrogen replacement therapy and cognitive functioning in the Atherosclerosis Risk in Communities (ARIC) Study. Am J Epidemiol. 1996 Dec 1;144(11):1048-57. PubMed.

    . The healthy cell bias of estrogen action: mitochondrial bioenergetics and neurological implications. Trends Neurosci. 2008 Oct;31(10):529-37. PubMed.

    . Brain and cognition. Is there any case for improving cognitive function in menopausal women using estrogen treatment?. Minerva Ginecol. 2009 Dec;61(6):499-515. PubMed.

    . Sex steroids to maintain cognitive function in women after the menopause: a meta-analyses of treatment trials. Maturitas. 2010 May;66(1):56-71. PubMed.

  6. The paper does not contradict current research, in that the WHIMS/WHI study looked at effects on much older women (most >70), was underpowered to comment on AD and instead reported an increased risk in all-cause dementia in women prescribed combined CEE/MPA, and also failed to look at younger women during the "critical window."

    In short, this analysis sums up what we already know. It simply emphasizes the need for randomized, controlled trials to look at the longer-term effects of estrogen therapy when prescribed during the critical period on longer-term risk of AD. RCTs to analyze this are currently underway in the U.S. (e.g., KEEPS).

Make a Comment

To make a comment you must login or register.


News Citations

  1. Dementia Risk Increases, at Least in Those Who Start Hormone Therapy Late

Webinar Citations

  1. Not Dead Yet: Estrogen Deserves Another Chance

External Citations

  1. AlzRisk database
  2. comprehensively discussed here
  3. Kronos Early Estrogen Prevention Study
  4. Early Versus Late Intervention Trial With Estradiol
  5. AlzRisk database

Further Reading