The sex steroids have been associated with brain development and global functioning directly or indirectly for thousands of years. Writings from the ancient Greeks and Egyptians linked emotional instability or any unacceptable behavior from a female with the uterus.⁺¹ Modern medicine continues to be perplexed by the influence of reproductive steroids on behavior. For example, a PubMed search of "estrogen and brain" gives 10,846 references since 1963. As 21st-century neuropsychiatrists try to understand the brain circuits and neurotransmitters that underlie emotion and behavior, the influence of sex steroids begins to establish its importance. Estrogen and mood is a much-studied area. Recent evidence indicates that at least a subset of depressive disorders are influenced by estrogen's effects at many points, particularly the premenstrual and perimenopausal times. A recent issue of a neuropsychiatric journal was dedicated to summarizing the current literature on the effects of estrogen on mood throughout the female life cycle.⁺² But what about estrogen's effects on other cortical functions? Cognition is an area of particular interest to the neuropsychiatrist. The current database and established truths for a relationship between cognition, neurodegeneration, and the sex steroids are much more limited. As our population ages and dementias become more prominent in our medical practices, it is vital that we more clearly understand any relationship between these steroids and neurodegeneration. The following discussion summarizes some of the most recent evidence linking one sex steroid, estrogen, with cognition and neurodegeneration.

There have been numerous reports of an association between estrogen and performance on specific measures of higher cognitive function. In particular, higher estrogen levels have been associated with improvements in verbal performance and decrements in visual-spatial performance.^+3—+5 However, a recent summary of the literature concluded that although there is observational evidence for estrogen enhancement of certain aspects of cognitive functioning, quantitative comparison across studies cannot be performed because of heterogeneity among subjects and variability in the cognitive measures used.⁺⁶ A comparison of two recent studies illustrates this problem. Both were prospective studies in which performance on the Forward and Backward Digit Span test was compared in groups of postmenopausal females over time. In one, females receiving estrogen replacement therapy (ERT) performed better than females not receiving ERT at both study initiation and 18 months later.⁺⁷ In the other, there were no differences in performance on this task, either at initiation or 2 years later.⁺⁸ However, the second study matched participants for educational level. In general, females on ERT are better educated, have higher socioeconomic status, and are healthier than females who are not.⁺⁹ Mood is also a critical variable that is sometimes overlooked. In most studies circulating levels of hormones have not been measured, making comparisons difficult. The situation is further complicated by the report that cerebrospinal fluid levels of estradiol (presumably representing hormone levels in the brain) are very different from circulating levels and do not decrease as much following menopause.⁺¹⁰

In addition to the well-known localization of estrogen receptors within hypothalamic nuclei, which are important for regulation of sexual and reproductive behaviors (not illustrated), estrogen receptors have now been found in other areas of the brain (see Cover and +Figure 1). There are at least two forms of the estrogen receptor (alpha and beta), and their distributions within the brain are different.^+11—+13 Receptor mapping studies do not all agree, perhaps because several different techniques (autoradiography, in situ hybridization, immunocytochemistry) and multiple species (mouse, rat, guinea pig, monkey, human) have been used. There are clear species differences, so the following summary is based only on studies of human and nonhuman primates (see Cover and +Figure 1). Several studies have found estrogen receptors in the hippocampal formation (hippocampus proper, dentate gyrus, subiculum, and entorhinal cortex), basal forebrain (septal nucleus, diagonal band of Broca, and nucleus basalis of Meynert), and mamillary bodies (Cover and +Figure 1, blue areas).^{+11,+13—+19} These support an influence on declarative (autobiographical, explicit) memory. Presence in the amygdala and dorsal raphe nucleus may underlie some of the effects of estrogen on mood and emotion (Cover and +Figure 1, pink area).^{+11,+14—+19} Estrogen receptors are present in movement-related areas including the substantia nigra and the subthalamic nucleus (Cover and +Figure 1, green areas).^{+13,+17—+19} They may also be present in the cerebellum (not illustrated), but exact localization is still unclear.^+11,+13 Most studies have not found estrogen receptors in the basal ganglia (caudate, putamen, or globus pallidus).^{+11,+13—+16,+18,+19} Several studies have reported estrogen receptors in various areas of the cerebral cortex (not illustrated), but this is still controversial.^+11,+13,+18 No obvious differences have been reported between females and males in regional distribution of the alpha estrogen receptor.^+15,+16 There may be gender differences in the distribution of the beta form of the receptor.⁺¹¹ One study has reported similar distributions of estrogen receptors in intact and ovariectomized females.⁺¹⁵

Estrogen has multiple modes of action within the central nervous system.^{+12,+20—+24} It binds to a nuclear receptor, acting intracellularly to alter gene expression. Estrogen actions via this genomic mechanism, which are necessarily slow, may include inhibition of apoptosis, suppression of inflammatory reactions, and modulation of neurotrophins and growth factors as well as neuronal structure and synapse formation. Estrogen also has rapid actions, occurring far too quickly for genomic mechanisms. These may include its antioxidant effects as well as enhancement of cerebral blood flow and cerebral glucose utilization. These nongenomic effects probably occur via both plasma membrane receptors and non—receptor-mediated pathways. Some actions, such as modulation of neurotransmitters, may occur by both genomic and nongenomic mechanisms.

Estrogen interacts with multiple neurotransmitter systems at multiple sites. For instance, it has been shown to modulate the levels of dopamine (upregulation), serotonin (downregulation), norepinephrine (downregulation), and acetylcholine (upregulation) in prefrontal cortex.⁺²⁵ This may occur via direct actions within cortex, or indirectly via estrogen receptor—mediated changes in brainstem or basal forebrain areas. Estrogen co-localizes with some neurotransmitters. Estrogen receptors have been found in serotonergic neurons within the dorsal raphe nucleus, where estrogen appears to facilitate serotonergic transmission by several mechanisms.⁺²⁶ Estrogen alpha receptor—containing neurons in the amygdala are heavily invested with cholinergic terminals projecting from the basal forebrain (+Figure 2).⁺¹⁵ Estrogen modulates aspects of neuronal plasticity, including dendritic spine formation. In the hippocampus, for instance, estrogen receptors are localized in GABAergic interneurons. Estradiol exposure decreases the activity of these inhibitory interneurons (via an interaction with a neurotrophin), resulting in an increase in pyramidal cell excitability, which in turn promotes formation of new dendritic spines and synapses (+Figure 3).^+27—+29

Estrogen has neuroprotective and neurotrophic actions that may be mediated by a variety of routes. Estrogen receptors (both alpha and beta forms) have been found in microglia and within reactive astrocytes.^+30,+31 In addition, there is evidence that estrogen suppresses activation of microglia and astrocytes and thereby the inflammatory cascade.^+31,+32 Circulating estrogen is critical to the health of some types of neurons. This has been clearly demonstrated for a subpopulation of dopaminergic neurons in the substantia nigra, suggesting the importance of estrogen in Parkinson's disease.⁺³³ It is not yet clear whether this effect on dopaminergic neurons is mediated via the intracellular estrogen receptor or by a plasma membrane receptor for estradiol, although the antioxidant actions of estrogen have been implicated.⁺²¹

Estrogen may also be important in other neurodegenerative diseases. Epidemiological studies generally support the view that estrogen replacement therapy (ERT) reduces the risk of developing Alzheimer's disease (AD),^{+22—+24,+34,+35} which occurs more frequently and progresses more quickly in females.⁺²³ It has been suggested that estrogen's anti-inflammatory action may be one important protective mechanism.⁺³¹ The process is not yet clear, since there is not a decrease in cortical estradiol or testosterone in AD.⁺³⁶ The influence of ERT on the progression of neurodegeneration after onset is more controversial, as are its effects on cognition. There was an absence of therapeutic effect on measures of cognition, mood, and cerebral blood flow in a recent double-blind, placebo-controlled study of ERT in AD,⁺³⁷ suggesting that its therapeutic use needs further study.

The incidence of stroke is lower in premenopausal females than in other groups (males, postmenopausal females), perhaps a result of estrogen's influence on cerebral vasculature, its effect on circulating levels of cholesterol, and/or its antioxidant actions.^+20,+23,+38 Although some reports indicate ERT lowers the risk of stroke after menopause, not all studies agree.⁺³⁸ A recent large study actually found more clinically significant brain atrophy in postmenopausal females who were receiving ERT than in those who were not.⁺³⁹ In that study, the prevalence of infarcts as demonstrated by magnetic resonance imaging was not different between ERT and non-ERT groups, and measures of cognitive functioning did not correlate with duration of estrogen treatment. Similarly, although some studies have reported that females fare better than males following traumatic brain injury, a recent meta-analysis of the few studies available reporting outcome by gender found that females fared worse than males on virtually every outcome measure, with the interesting exception of "return to work."^+40,+41 Thus, although the neuroprotective actions of estrogen have been convincingly demonstrated in animal models of ischemia, contusion, hypoxia, and drug-induced toxicity, its protective effects in humans are less clear.^+20,+23,+24 There is a need for careful clinical studies in which potentially confounding factors such as premorbid conditions, severity of injury, and treatment differences can be assessed.

The influence of estrogen on such a wide range of brain functions has important implications for neuropsychiatry. It may be one reason why there are gender differences in both vulnerability to some mental illnesses and disease course. It suggests, as well, the potential for differences in response to therapeutics both across gender and as a function of life stage. The neuroprotective effects of estrogen raise the hope that it can be used to treat degenerative diseases of the CNS, such as Alzheimer's disease. It may also be helpful in salvaging tissue after stroke and traumatic brain injury. The existence of at least two forms of the estrogen receptor, and the differences in their distribution among brain areas, raises the possibility of designing pharmaceutical interventions that are targeted to specific aspects of estrogen's wide range of functions. It is clear, however, that estrogen has such widespread effects that it will be very difficult to predict what the effect of estrogen agonists and antagonists will be for particular aspects of cognitive function. Thus, new therapeutics will have to be very carefully evaluated.