A number of genetic and non-genetic factors may influence the development of AD. Mutations in genes on chromosomes 14, 1, and 21, which increase ß- amyloid deposition and lead to dementia symptoms before the age of 60, are inherited in an autosomal dominant manner. 2 These relatively rare mutations do not contribute to late-onset dementia, where dementia risk is likely influenced by susceptibility genes, the best recognized of which encodes apolipoprotein E,2 a lipid transport protein involved in neuronal repair processes. Other factors linked to AD risk include head injury, depression, lower education, absence of physical exercise, and conditions associated with cardiovascular disease (CVD). Estrogen, as will now be described, has several effects on brain function that could influence AD; exposures to endogenous estrogens and to estrogencontaining hormone therapy (HT) are relevant to AD pathogenesis and treatment.
Menopause and Estrogen
Menopause represents the permanent cessation of menses due to the loss of ovarian follicular function,3 including the near-complete loss of estrogen production by the ovaries. Natural menopause, usually heralded by several years of menstrual cycle irregularity, occurs at an average age of 51 years. Approximately 40% of a woman’s adult life occurs after her final menstrual period.Two forms of estrogen receptors, which function as ligand-activated transcription factors, are expressed by neurons in the human brain.
Estrogen has been shown to promote neurite growth, modulate synaptic plasticity, and enhance long-term potentiation (a physiological process thought to underlie the formation of episodic memories).4 Estrogen protects neurons from oxidative stress, excitatory neurotoxicity, and ß-amyloid toxicity,5 reduces neuronal death from apoptosis,6 and diminishes ß-amyloid deposition in the brain.7
Cholinergic deficits are prominent in AD, and cholinergic neurons express receptors for estrogen.8 The effects of estrogen on noradrenalin and serotonin,9,10 two other neurotransmitters affected in AD, may be important to cognition and mood. The competing effects of estrogen on inflammation and coagulation have the potential for harm, and estrogen can increase the risk of ischemic stroke.11
Estrogen, Memory, and Cognition
Although complaints of poor memory are common around the time of menopause, only a few studies have evaluated cognitive change during midlife. Findings from short-term clinical trials in younger women indicate that estrogen may benefit episodic memory for verbal (but not non-verbal) information, particularly where ovarian estrogen production has been abruptly curtailed by oophorectomy.12,13 However, observational evidence suggests that women undergoing the natural menopause do not experience discernible impairment on tasks of verbal episodic memory,14,15 and serum estrogen levels at midlife are unrelated to memory.14 Most other cognitive domains also appear generally unaffected by the natural menopausal transition.15,16
Among older post-menopausal women, observational studies provide contradictory evidence regarding a potential role of HT in enhancing or protecting cognition. In the Nurses Health Study (NHS) of women older than 69, cognitive change over a twoyear follow-up period did not differ between women using HT and those who had never used HT.17 However, in the Cache County cohort, women over the age of 64 who had used HT experienced less decline over a two-year period than women who had never used HT.18 Findings from large randomized controlled trials are more consistent, indicating that HT initiated by older women does not improve memory or other cognitive domains. Clinical populations in these studies included women with established coronary heart disease (CHD)19 and cerebrovascular disease,20 as well as relatively healthy women in the Women s Health Initiative Memory Study (WHIMS).21,22 The latter was designed as a primary prevention trial and included over 7,400 women between the ages of 65 and 79 who were assessed annually.Active treatment in the WHIMS was with conjugated estrogens given alone to women without a uterus and was combined with a progestin (medroxyprogesterone acetate) for women with a uterus. Over a mean follow-up period of approximately five years, women randomized to HT performed slightly less well on a measure of global cognitive ability than women randomized to placebo.22
Hormone Therapy and AD
Using both case-control and cohort methodology, a number of observational studies have investigated the association between the use of HT and a woman s subsequent risk of developing AD. Many of these studies (for example, the Leisure World retirement community,23 a New York City community-based cohort,24 the Baltimore Longitudinal Study of Aging,25 and the Cache County cohort26) report reductions in risk. Meta-analyses suggest overall risk reductions of approximately one-third.27
These protective associations for AD are challenged by findings on dementia from the WHIMS.28 For women with a uterus, the rate of dementia from any cause was doubled for women assigned to HT. For women without a uterus, the risk was elevated by approximately one-half, although this increase was not statistically significant. When women with and without a uterus were considered together, the pooled risk of dementia was 76% higher among women randomized to receive HT.28 AD was the most common dementia diagnosis in the WHIMS, but the number of AD cases was considered too small for separate analysis.
The WHIMS findings of increased dementia risk were surprising, given prior observational research that had generally associated HT use with lowered risk.27 In trying to understand the discrepancies, there are two important considerations. Firstly, it is possible that much of the observational research on HT and AD was biased by the healthy-user effect.Women who take HT tend to be healthier and to lead healthier lifestyles than women who do not take HT.29 The question of whether some of these factors explain the association between HT use and reduced AD risk, rather than HT use per se, now arises. Secondly, many of the women included in observational trials were relatively young, beginning HT for menopausal symptoms around the time of menopause. In contrast,women in the WHIMS were at least 65 years of age. The question of whether HT effects differ when used at a younger age now arises.30 In the Multi-Institutional Research in Alzheimer s Genetic Epidemiology (MIRAGE) study, the protective association of HT was modified by age, being present among younger post-menopausal women but not older women;31 in Cache County, a protective association was found for women who had used HT in the past, but not for women who were using HT at the time of cohort enrollment.26 Indeed, other biological effects of estrogen may vary according to age or time since menopause, such as hormone effects on bone fracture32 or the progression of atherosclerosis.33
Even before HT was evaluated for its effect on dementia risk, HT was considered a therapy for women with AD. Small, open-label clinical trials suggested possible benefit, but results from subsequent randomized clinical trials were less salutary. No significant improvement was reported in a 16-week trial involving 42 women34 or in a 52-week trial involving 120 women without a uterus.35 In these studies, active treatment was with conjugated estrogens. Similarly, French investigators reported no benefit of transdermal estradiol in a 38-week trial of 117 women, all of whom also received a cholinesterase inhibitor.36 In contrast, women receiving transdermal estradiol performed somewhat better on several neuropsychological tasks in an eight-week trial of 20 women with AD.37
Selective Estrogen Receptor Modulators and AD
One consideration for adverse cognitive outcomes in the WHIMS is that HT sometimes increases the risk of vascular disease,11 and vascular factors might contribute to dementia from AD or from other causes. Speculatively, the adverse effects of estrogen on one tissue (the vasculature) might therefore have overshadowed the beneficial effects on another (the brain). Interestingly, there are compounds whose estrogenic effects are limited to specific tissues. The so-called selective estrogen receptor modulators (SERMs) induce unique conformational changes in the estrogen receptor,38 leading to tissue-specific effects. Well-known examples of SERMs are tamoxifen, used to treat breast cancer, and raloxifene, used to prevent osteoporosis. Not all SERMs act equivalently, and within the brain agonist or antagonist the actions of tamoxifen and raloxifene differ from each other.39
Cognitive outcomes of raloxifene were evaluated in the Multiple Outcomes of Raloxifene Evaluation (MORE) study, a randomized clinical trial of over 7,400 post-menopausal women with osteoporosis. The mean age of women in MORE was 66 years.After three years of treatment, raloxifene had no strong effect on memory or other aspects of cognition, although there was a trend for less decline on a test of verbal memory for women receiving raloxifene compared with placebo.40 In separate analyses involving over 5,000 women, MORE participants receiving a higher than standard dose of raloxifene were less likely to develop mild cognitive impairment (MCI) (33% risk reduction) or AD (48% risk reduction), although only the former reduction was statistically significant.41 Tamoxifen has not been evaluated as carefully, but one observational study implied that this SERM might impair cognitive skills.42
Laboratory research provides a rationale for the use of estrogen-containing HT for the prevention or treatment of AD; however, clinical research findings to date are generally disappointing or inconclusive. For treatment, it appears that HT initiation is unlikely to improve symptoms of dementia.34 36 For the older postmenopausal woman without dementia, HT does not provide cognitive benefit or reduce the risks of dementia.19,20,22,28 However, long-term cognitive effects of short-term HT use in midlife are unknown. This issue is important because millions of women continue to use HT,43 primarily in middle age. Further research is needed before it can be known whether potential cognitive risks and benefits of midlife HT will differ from recognized risks later in life, or whether the tissueselective effects of a SERM might benefit AD even when estrogen does not.