Thinking with your gonads: testosterone and cognition
Jeri S. JanowskyE-mail The Corresponding Author
Department of Behavioral Neuroscience, Oregon Health & Science University CR131, 3181 SW Sam Jackson Park Road, Portland, OR 97239, USA
Available online 4 January 2006.
Sex hormones play a crucial role during brain development, but do they modulate or maintain cognition throughout life? Despite several million prescriptions annually for testosterone supplementation, we do not really know the answer. Here I review recent evidence that testosterone alters neural activity essential for learning and memory, and plays an important role as a neuroprotective agent in aging. In particular, testosterone deprivation is associated with poor memory in men and replacement can enhance memory and spatial cognition. However, there is little evidence that testosterone selectively affects only those cognitive domains where sex differences in performance have been found. There are also gaps in our knowledge surrounding the individual cognitive processes altered by testosterone, their neural basis, and the degree to which testosterone affects cognitive performance in women.
Article Outline
Introduction
Biological plausibility
Mechanisms of action
Locations of action
Cognitive effects of testosterone
Memory effects: animal models
Translating between animals and humans
Testosterone and cognitive aging
Testosterone and neurodegenerative disease
Testosterone in women
Brain activity
A human ‘gonadectomy’ model
Why results differ?
Do sex differences signal testosterone effects?
Summary and future directions
References
Introduction
The sex hormone testosterone plays a crucial role in the brain organization necessary for sexual development and sexual behavior. However, testosterone is present throughout life in cortical regions important for cognition. This review comes at a crossroad in cognitive neuroscience. Designer pharmaceuticals are in development that are capable of modifying specific cognitive processes and brain systems. New formulations of testosterone make testosterone replacement in men and women a reality, but without a full understanding of its effects on the brain and almost no data on its effects on cognition in women. Here I review neurobiological and cognitive studies that show that testosterone’s role as a cognitive modulator is complex but that it could be particularly important for maintaining cognition in aging. There is very little known about cognitive effects of testosterone in women, so much of this review will focus on findings in men. Not reviewed here are studies of the organizational effects of testosterone during early development, the effects on cognition of anabolic steroid abuse, effects in people with congenital endocrine abnormalities [1] or effects of endocrine manipulations in people seeking gender reassignment [2]. Findings in those groups are of great scientific interest, but not likely to be generalizable to testosterone’s effects in young and elderly healthy normal adults. In sum, researchers are beginning to gather some knowledge but there is much room for focused, hypothesis-driven studies of this topic in the future.
Biological plausibility
Testosterone is readily available in pill, injection and gel forms that will raise levels in a dose-dependent fashion. Testosterone has activities in multiple organ systems, and despite its historical lore as a cognitive enhancer (see Box 1) its effects are not completely understood, and potential negative consequences of supplementation have also received attention [3]. Our limited understanding is particularly true of its effects on the brain and cognition.
Box 1
Real versus placebo effects of testosterone
The idea that testosterone had powers beyond that of sexual virilization, including cognitive effects, predates the discovery of the hormone. A testosterone extract was first obtained from bull's testicles in about1927, purified in 1935 and synthesized in 1936 [71]. Before testosterone's discovery, testicular transplants became a cottage industry in both Europe and the United States for the express purpose of reversing the negative affects of aging on both the body and the mind [72]. It was later apparent that sufficient physiological levels of testosterone could not have been achieved from either injections of extracts from testes or the transplant technique, and thus the early reports were placebo effects or falsehoods [73].
Mechanisms of action
Production of testosterone by the testes is regulated through a hormonal feedback loop that requires signals from the hypothalamus and pituitary [4]. Testosterone is lipid soluble, thus it can enter cells throughout the body and nervous system regardless of whether there are androgen receptors in the cell. Only in cells where it encounters a receptor does it form a complex that will influence protein synthesis in the nucleus of a neuron and modify cell function. This modification can be in the form of specific changes in neurotransmitter production and release, synapse conformation that modifies efficacy, as well as many other possibilities (see Figure 1). For instance, gonadectomy selectively reduces acetylcholine in hippocampus and anterior cingulate cortex but not other regions [5] and transiently decreases then increases dopamine innervation in cingulate cortex of rats [6]. Testosterone induces increased firing of serotonin neurons in male rats, although gonadectomy has no effect [7].
Display Full Size version of this image (47K)
Figure 1. Testosterone acts by combining with a receptor and affecting protein synthesis in the cell nucleus. Thus, testosterone can promote changes in the conformation or number of synapses, synaptic efficacy, and neurotransmitter or receptor production. Any of these effects could have a downstream effect of modifying cognition and behavior. However, sex hormone binding globulin (SHBG) prevents testosterone from being biologically active (lower left). Thus, changes in ‘total’ testosterone levels are not an accurate reflection of testosterone's capability to modify behavior.
Testosterone can be made biologically inactive by sex hormone binding globulin (SHBG), which does not permit it to bind to receptors (Figure 1). This is important, as SHBG increases with aging, with changes in hormone levels, and in various medical disorders. Thus, ‘total testosterone’ might not reflect the testosterone that is biologically active.
Testosterone activity in the brain occurs through similar mechanisms to those in the musculoskeletal and reproductive systems. That is, testosterone is a steroid hormone that is synthesized from cholesterol (see Figure 2) in the testes and adrenal gland, but can be made de novo in the brain [8]. Testosterone can be converted to estradiol in the brain by the aromatase enzyme. Aromatase is found in the hippocampus and amygdala, structures vital to learning and memory. Alternatively, it can be metabolized to dihydrotestosterone (DHT) and bind to androgen receptors (see Figure 2). As both men and women have estradiol and androgen receptors, testosterone can exert its effects either as an androgen or an estrogen. Thus, when evaluating studies of testosterone’s effects on cognition, it is important to note that manipulations of testosterone will modify other hormones in the brain, and that effects of testosterone can be modified by several factors, including disease and age.
Display Full Size version of this image (19K)
Figure 2. Testosterone is a hormone that is metabolized from cholesterol. Thus, its production can be influenced at several metabolic steps. In addition, it can be further metabolized to form estradiol or other androgen molecules. When testosterone affects cognition, it is not known in most instances whether this is as an androgen or via its conversion to estradiol.
Locations of action
Androgen receptors are found in brain regions that are crucial for learning and memory including the hippocampus 9 and 10, prefrontal cortex [11], and amygdala, [12] but are not found in other cortical regions of the brain [13] (Figure 3). In primates, androgen receptors are found in these regions but throughout the cortical layers, in multiple types of neurons as well as in glia 13 and 14. Studies using animal models show that testosterone modifies the physiology and function of the hippocampus. For instance, androgen deprivation causes a 40% decrease in synaptic density in the hippocampus of both rats [15] and monkeys [16] and testosterone replacement in male animals normalizes synaptic density [15]. This is likely to be particularly important in aging when the loss of synapses in hippocampus and prefrontal cortex [17] is accompanied by functional loss of memory
Display Full Size version of this image (85K)
Figure 3. Summary of the brain regions with known androgen activity (androgen receptors or other evidence of activity) in humans and associated cognitive effects. Image created and reproduced here by kind permission of Arthur W. Toga.
Cognitive effects of testosterone
Memory effects: animal models
Androgen deprivation by gonadectomy in rodents impairs performance on memory measures that require the hippocampus such as maze learning and fear conditioning 18 and 19. Testosterone replacement normalizes performance. Hippocampal-independent measures such as the ability to balance on a rotating rod are unaffected [18]. The accumulation of beta amyloid, a risk factor for Alzheimer’s disease, increases with testosterone deprivation [20]. Blockade of androgen receptors in transgenic mice expressing the apolipoprotein E4 allele, which is a risk gene for Alzheimer’s disease in humans, impairs memory on a maze learning task [21]
Translating between animals and humans
Parallel studies in humans and animal models are not simple. The great majority of the animal studies compare models with complete androgen deprivation (e.g. gonadectomy) to replacement of testosterone or its metabolites. However, in humans the main interest has been in whether native testosterone levels are related to cognitive abilities and whether lower testosterone (but not no testosterone) has cognitive effects. A focus of interest has been whether testosterone can protect people from aging-related cognitive loss, with much of the research focusing on elderly men. By constrast, most studies of the neurobiology and neurobehavioral effects of testosterone use adult but not old rodent or non-human primate models. Thus, the devil is in the details when evaluating what cognitive functions are modified by testosterone, under what circumstances and at what time of life.
Testosterone and cognitive aging
Bioavailable testosterone levels decline with age in men and women. However, unlike menopause when estradiol falls to nearly undetectable levels, testosterone production declines slowly in healthy men such that those in their seventies have approximately 40% lower testosterone than men in their twenties 22 and 23. A precipitous decline in testosterone occurs in women on postmenopausal hormone replacement [24].
Numerous studies suggest a link between testosterone and cognition in men, particularly with aging, but there have been few studies of women. There is a positive relationship between bioavailable testosterone levels and memory in older men 25 and 26 and women [27]. Two recent studies show positive relationships between endogenous testosterone and memory, and processing speed in men [28], and attention in men and women [29], but they suggest that this relationship is particularly true in aging. By contrast, other studies show that age but not testosterone is related to mental rotation or digit span when several common covariates are considered in the analysis [30].
Placebo-controlled and dose-response studies suggest that testosterone supplementation can modify cognition, particularly in older men (Figure 3). Testosterone supplementation improves spatial cognition 31 and 32 and working memory [33] in healthy older men. Visual-spatial cognition improves in older men with testosterone replacement in a dose-dependent manner [34], but similar effects are not found in young men [35]. Spatial memory performance improves in men with testosterone supplementation regardless of whether pharmacological manipulations (aromatase inhibitors) permit its conversion to estradiol [36]. However, verbal memory improves only if conversion to estradiol occurs, suggesting estradiol mediates testosterone’s role in verbal memory in men. There are weak associations between testosterone levels in older men and performance on the Trails-B test which assesses switching between cognitive sets as well as psychomotor speed, but supplementation in men with testosterone levels within the range of younger men does not significantly alter performance [37]. Testosterone replacement has few effects in men with low or no testosterone throughout life (congenital hypogonadism) with the exception that verbal fluency is worse and improves with replacement, in comparison with men with normal testosterone levels [38]. By contrast, in another study word-list learning does not improve in supplemented hypogonadal men [39], nor is verbal and nonverbal memory, or mental rotation affected after a single injection of testosterone in men [40].
Testosterone and neurodegenerative disease
Lower premorbid testosterone levels are associated with a higher risk for Alzheimer’s disease 41, 42 and 43. Although controversial 44 and 45 some studies suggest that low testosterone is associated with increased beta amyloid deposition in men [46], which is considered a major risk factor for the disease. Supplementation with testosterone in men who have Alzheimer’s disease or mild cognitive impairment, a precursor to Alzheimer’s disease, results in improvement in spatial memory or spatial cognition 47 and 48 but results are not consistent across studies [49]. The findings of positive effects on spatial cognition and memory are in contrast to studies showing no effect of estrogen supplementation in women with Alzheimer’s disease (e.g. [50]), despite similar epidemiological evidence in women that long-term estrogen replacement decreases the risk for Alzheimer’s disease, at least if initiated many years before the onset of the disease [51].
Testosterone in women
Few studies of testosterone supplementation are available for women. However, a series of related studies show that mental rotation [52], and object location memory [53] improve, and attentional bias is reduced towards fearful faces but not happy faces [54] within hours of a single injection of testosterone in young women.
Brain activity
There are very few studies that have directly examined testosterone’s effects on brain activity. Testosterone replacement in hypogonadal men increases cerebral perfusion, as measured by single-photon emission-computed tomography (SPECT) [55]. Similarly, in hypogonadal men who improve on mental rotation ability with testosterone replacement, positron emission tomography (PET) shows greater metabolism during task performance [56]. Testosterone treatment normalizes PET activity induced by visually presented sexual stimuli in the right orbital and inferior frontal cortex, claustrum and insula in hypogonadal men [57]. Testosterone levels correlate with overall brain activity during a synonym- (semantic) and letter-string (perceptual) identification task in healthy, normally cycling women, suggesting nonspecific or vascular effects of testosterone in women [58].
A human ‘gonadectomy’ model
Prostate cancer is nearly always androgen responsive and is promoted (but not caused) by testosterone. Thus, a common treatment for men at high risk for prostate cancer progression, or when metastatic progression has occurred is testosterone deprivation. This slows the course of the disease and significantly prolongs life [59]. Hormonal blockade (gonadotropin releasing hormone, GnRH) causes nearly complete androgen deprivation but also blocks several other hormone systems. Drugs that block the androgen receptor are also used. Thus, men with prostate cancer on testosterone deprivation are a model system to ask whether testosterone is important for cognition, particularly in elderly men, but results across studies are mixed.
Four studies found that androgen deprivation therapy is associated with worse verbal memory 60, 61, 62 and 63 and attention [62], but no effects on a nonverbal memory measure or tests of ‘executive function’ such as the Stroop test that requires the subject to inhibit a prepotent response to color names [63]. The verbal memory measures used in these studies, word lists or brief stories require medial temporal lobe and hippocampal memory systems [64], suggesting that like the animal models, hippocampal dependent measures are impaired by testosterone deprivation. By contrast, intermittent androgen deprivation improved performance on a verbal memory measure and adversely affected spatial cognition in one study [65], but improved verbal memory while causing significant declines in physical functioning in another study [66]. Severe deprivation, such as induced in men with prostate cancer, causes a loss of estradiol as well (see Figure 2); hence the memory impairment might be due in part to low estradiol levels. Preliminary data supports this possibility as verbal memory improves in men on high-dose estradiol hormonal therapy, an alternative therapy for prostate cancer [60].
Why results differ?
Differences among studies can be because of the age of participants, their cognitive status, the severity of hypogonadism, the degree to which testosterone levels are elevated by the treatment, length (hours to years) and type (gel, injection, patch) of treatment, and the small number of participants, which preclude finding significant effects in some studies. The cognitive measures and processes assessed vary in sensitivity and differ from study to study. The definition of hypogonadal and exclusion of subjects with depression or cognitive impairment varies widely among studies. Any one of these factors might be crucial for finding effects of testosterone, but together they suggest effects are subtle or are specific to memory and spatial cognitive domains.
Do sex differences signal testosterone effects?
Two kinds of information can be brought to bear on this question: (i) Studies that show a relationship between testosterone and performance on measures that also show sex differences but no relationship on measures that do not; and (ii) studies that examine sex steroid levels in normal men and women who perform at the limits of the range for their sex on tasks that show sex differences. The reasoning is that if men show an advantage on a task over women, then testosterone should relate to performance and men with higher testosterone should outperform men with lower testosterone. For example, men on average outperform women on tasks of spatial cognition, particularly, mental rotation. Indeed several studies cited above show effects of testosterone on mental rotation or a relationship between testosterone levels and performance, but some do not. When a single measure of testosterone is related to a single instance of cognitive performance, particularly in high functioning young adults with normal testosterone levels, few consistent results are found [67]. In addition, sex differences in performance are maintained in aging when radical sex steroid changes have occurred, suggesting that circulating sex steroids are not the source of maintained sex differences in cognition in adulthood [68]. Probably more important is that other cognitive domains, that show little or no sex differences are affected by testosterone loss or replacement, notably memory. Thus, sex differences in performance are not reliable signals of cognitive processes that will be influenced by testosterone in adulthood.
Summary and future directions
The effects of testosterone on cognition are not nearly as dramatic as the advertisements for the products suggest. This is true even in the most severe cases of testosterone deprivation or testosterone replacement in hypogonadal older men. However, the findings thus far are by no means piffle. The pattern across studies suggests that testosterone can have effects in adulthood when conditions make cognition fragile, such as when working and long-term memory decline in aging, or when testosterone levels are significantly modified, such as in replacement or hypogonadism. Reliable relationships between testosterone and cognition when both are optimal are less likely. This might be because testosterone has a subtle neuromodulatory role in adulthood but can be neuroprotective with brain aging when both cognition and testosterone levels decline. The specific cognitive and neural processes where testosterone plays a chronic but subtle neuromodulatory role have not been isolated even within domains such as memory and spatial cognition, where testosterone effects have been found. The isolation of testosterone’s relationship to figural encoding, initiation of transformation or decision making but not the rotation process itself is one example of specifying modulated processes [69]. Very few studies have dissociated potential changes in mood, arousal or attention from direct effects on memory and other cognitive processes. Such studies would implicate particular neural systems and improve our understanding of the specificity or breadth of testosterone effects. In addition, there are remarkably few studies of testosterone’s effects on the brain and cognition in women. This is becoming increasingly important, as attention focuses on testosterone’s role in women’s libido [70] and the significant loss of testosterone production caused by postmenopausal hormone replacement [24]. There is plenty of work to be done.