Although men do not experience an abrupt cessation of gonadal hormone production at midlife as do women, levels of testosterone (T) decrease gradually with aging. Because estradiol (E) arises mainly from the conversion of T in men, the availability of E-2 also decreases with increasing age. In randomized clinical trials, E-2-replacement therapy has been shown to maintain aspects of cognition in postmenopausal Women, specifically on verbal memory. The present prospective, randomized, cross-over trial is being undertaken in order to determine whether E-2 will enhance verbal memory in men with mild cognitive impairment (MCI). Men with MCI will randomly receive E-2 or placebo for the first 3 months of treatment and will then be crossed over to the other treatment for an additional 3 months. A battery of neuropsychological tests will be administered at pretreatment and, again, following each 3-month treatment phase. It is hypothesized that elderly men with MCI will perform better on tests of explicit memory when they are being treated with E-2. compared with their performance under placebo conditions. Evidence from epidemiological studies (Resnick, Metter, & Zonderman, 1997; Jacobs, Tang, Stein, et al., 1998; Halbreich, Lumley, Palter, et al., 1995) and prospective controlled studies of healthy elderly women (Sherwin,, 1988; Sherwin Phillips, 1990; Phillips & Sherwin,1992) has demonstrated that exogenous estrogen helps to maintain aspects of memory in aging women. There is also accumulating epidemiological evidence that postmenopausal estrogen replacement therapy might lower the risk of Alzheimer's disease (AD) in women (Paganini-Hill & Henderson, 1994, 1996; Kawas, Resnick, Morrison, et al., 1997; Waring, Rocca, Peterson, & Kokmen, 1997) and/or might delay the onset of the disease (Tang, Jacobs, Stern, et al., 1996). Unlike in women, there is not a drastic decrease in testosterone (T) production in middle-aged men. However, the vast majority of studies have shown an age-associated decrease in T. It has been estimated that 50% of healthy men over the age of 50 years will have a BT level below the lowest level seen in 20- to 40-year-old healthy men (Morley, Kaiser, Perry, et al, 1997). Because some proportion of T is aromatized to estradiol (E-2), the decrease in the production of T also means that E-2 levels would decrease in men as they age. Despite our knowledge of these endocrine changes with normal aging, very little information is available to examine the possible influence of the diminishing levels of sex hormones on cognitive functioning in older men. There is a paucity of information on the possible effects of T on cognition in males. In male rats (Alexander, Packard, & Hines,1994), testosterone had rewarding affective properties and facilitated learning of associations using a place-preference task. However, intrahippocampal injections of E-2 enhanced spatial memory in intact male rats (Alexander, Packard, Nores, & Olson, 1994). This is consistent with the hypothesis that EZ aromatized from T determines sex differences in spatial functions in males. Correlational studies on sex hormones and cognition in young men have yielded inconsistent results. A positive, linear relationship between plasma levels of T and visuospatial ability and a negative linear relationship between T and verbal ability were found in European men (Christiansen & Knussman, 1987) and in African bushmen (Christiansen, 1993). Similarly, a positive relationship between T and visuospatial ability was also found in healthy but not in alcoholic men (Errico, Parson, Kling, & King, 1992). On the other hand, Kampen and Sherwin (1996) reported a positive correlation between plasma levels of E (2), but not T, and scores on tests of visual memory in healthy young men. Others have argued that a curvilinear relationship between T and spatial memory exists, such that both high and low plasma T levels dampen spatial abilities, while mid-level concentrations of T optimize them (Gouchle & Kimura, 1991). A second paradigm used to investigate effects of T on cognition involves testing hypogonadal men before and after treatment with T. Scores on tests of spatial ability were significantly lower in men with idiopathic hypogonadotropic hypogonadism, compared with scores of healthy control men, but scores of those with acquired hypogonadism did not differ from those of the healthy controls (Hier & Crowley, 1982). When six of the hypogonadotropic hypogonadal men were treated with T for 3 months, their scores on tests of spatial ability failed to improve. These data, on a small number of Ss, suggest that the influence of androgens on spatial ability is exerted at or before puberty and does not wane with falling levels of androgens. In contradistinction, the administration of T to elderly hypogonadal men caused an increase in scores on the Block Design subtest of the WAIS-R, a test of visuospatial ability (Janowsky, Oviatt, & Orwoll, 1994), suggesting that spatial ability wanes with falling levels of T and can be restored by T administration. In a recent study of hypogonadal and eugonadal men who received exogenous T, no between-group differences on scores of visuospatial ability occurred (Alexander et al., 1998). However, in that study, plasma testosterone levels were in the hypogonadal range at the time of cognitive testing. Estradiol levels were not measured in that or in any other hormone replacement study of hypogonadal men. Neither were spatial abilities or visual memory tested. These methodological problems make these findings difficult to interpret. In a recent prospective study of hypogonadal men (mean age = 65 years), 15 participants randomly received placebo and 17 were given testosterone cypionate, 200 mg IM, every 14 to 17 days for 12 months (Sih et al., 1997). Testosterone improved bilateral grip strength and increased hemoglobin, but was not associated with significant changes in prostate specific antigen (PSA) or with changes in lipoprotein lipid values. Neither were there changes on tests of verbal or visual memory or in verbal fluency. By chance, the T group had higher scores on these cognitive tests at baseline and their performance had not changed at posttreatment. However, a serious methodological problem concerns the fact that blood was sampled and cognitive testing was administered 14 to 17 days postinjection, at a time when Free T levels were at their nadir. Spatial skills were not tested in this study. Recently, we conducted a longitudinal study on a group of healthy elderly men, a group of women estrogen users, and a group of female estrogen nonusers, who were an average of 72 years of age. Most returned for a second test session 18 months to 2 years later, when they were approximately 74 years of age (Carlson & Sherwin, submitted). While plasma E-2 levels were significantly higher in the estrogen users than in the men and the estrogen-nonusers at both test times, plasma E-2 levels of the men were significantly higher than those of the women who did not use estrogen at both the 72- and the 74-year-old test times. Of course, plasma T levels were higher in the men, compared with both female groups, but their average T level fell within the lower third of the normal range of male values. Both the males and the female estrogen users also had higher scores on the Digit Span test, a measure of verbal memory, coincident with their higher E-2 levels. These findings show that, in absolute terms, 72- and 74-year-old healthy men had three times the concentration of E-2 in plasma and better scores on some tests of verbal memory, compared with age-matched women who did not use estrogen. Although these findings suggest that estrogen may be important for maintaining verbal memory in older men, as well as in older women, the correlational nature of this study precludes causal statements. In summary, while the evidence showing that estrogen maintains verbal memory in postmenopausal women is fairly consistent, the findings on the possible role of sex-steroid hormones on aspects of cognition in men are inconclusive. One explanation for the inconsistency between the male studies are the inherent methodological problems (failure to use a comprehensive battery of neuropsychological tests in any given study, failure to determine that T levels were in the physiological range during posttreatment testing, failure to measure all aspects of cognition). Another possible explanation, however, is the failure of these investigators to consider that T is aromatized to E-2, both peripherally and in brain tissue (Roof & Havens, 1992). Since the testicular production of its steroids decreases with increasing age but never ceases entirely, male brains have E-2 available lifelong. It is, therefore, possible that our 72- and 74-year-old men (Carlson & Sherwin, submitted) performed better on a test of verbal memory than age-matched women estrogen nonusers because of their higher levels of plasma E-2 and not because of their higher levels of plasma T. The finding that T replacement therapy given to hypogonadal men caused an increase in scores on a test of verbal fluency (Alexander et al., 1998) provides support for this hypothesis. Additional evidence comes from the positive relationship between E-2 levels and visual memory in young men (Kampen & Sherwin, 1996) and the enhancement of spatial memory by intrahippocampal injection of E-2 to male rats (Alexander et al., 1994). It may, therefore, be the case that it is E-2 aromatized from T, and not T itself, that maintains aspects of cognition in elderly men. None of the extant studies could address this possibility since they failed to measure E-2 levels or to administer E-2 to men. Interestingly, women with AD not taking estrogen replacement therapy performed significantly worse than men with AD on verbal fluency, when age and severity of dementia were controlled (Roof & Havens, 1992). Although plasma levels of E-2 were not measured in that study, we have reported that elderly men with AD also have higher E-2 levels than age-matched women estrogen nonusers with AD (Carlson & Sherwin, submitted).
