Results from several population-based studies are inconsistent about the role that estrogen therapy plays in the prevention of cognitive decline in elderly women. It is suggested that these inconsistencies may be explained by the preferential use of estrogen by healthier and better-educated women. An association between higher serum concentrations of endogenous estradiol and a decrease in the risk of cognitive decline would support the hypothesis that estrogen concentrations may be protective. On the other hand, total hormone concentration may not support an appropriate assessment of biologic activity of the hormones because non–protein-bound (free) and loosely bound (bioavailable) estrogen cross the blood-brain barrier more readily and correlate with cognitive function. Testosterone is converted to estrogen by aromatase enzymes that are present throughout the body, but the relationship between serum testosterone concentration levels and cognitive function in elderly women has not been studied. Yaffe and colleagues studied the association between serum concentration levels of estradiol (non–protein-bound and bioavailable) and testosterone concentration levels (total and non–protein-bound) and the risk of cognitive impairment in elderly women.

Participants included 9,704 elderly, predominantly white women living in four U.S. cities who were enrolled in a long-term study of osteoporosis. Women who were at least 65 years of age, were mobile and had no history of hip replacements entered the study. Baseline assessment (occurring between 1986 and 1988) included clinical history, physical examination, laboratory testing and cognitive assessment using a modified Mini Mental State examination (mMMSE). Participants were examined every two years and completed yearly questionnaires. As part of a case-cohort study of hormonal risk factors for breast cancer, hormone concentrations were measured in baseline serum samples in women who developed breast cancer during the study and in randomly selected control subjects. The analyses reported here are based on 425 women who underwent measurements of cognitive function and serum sex hormones. Over the course of the six-year study, 116 of the 425 women developed breast cancer. At baseline, none of these women had reported a history of dementia.

The women were grouped by their initial hormone levels. The initial cognitive test scores did not differ by hormone level for any of the types of hormone evaluated. Of the 425 women enrolled, 39 (9 percent) died before repeat cognitive testing. Of the survivors, mMMSE scores were available for 292 women at the end of the six-year study period. Overall, the average score declined 0.5 points over six years, but significant differences were found between groups. The score for women in the high tertile group for non–protein-bound estradiol declined 0.1; scores for those in the middle and lower tertiles declined 0.7 and 0.6, respectively. Overall, significant cognitive impairment (a decrease of three points or more on the mMMSE after six years) occurred in 37 women (13 percent). Cognitive impairment occurred in 17 of 106 women (16 percent) in the lowest tertile of the non–protein-bound estradiol group, but in only five of 94 women (5 percent) in the highest tertile group (see accompanying table). Results for bioavailable estradiol were similar. No differences were evident in the proportion of women with cognitive impairment or the risk of impairment, by tertile of total estradiol or non–protein-bound testosterone concentration levels. After multivariate analysis of cognitive impairment with hormone concentrations as continuous variables, the authors estimated a 50 percent lower risk of cognitive decline for every standard deviation increase in non-protein–bound estradiol. This is equivalent to a 0.5 odds ratio. Similar results were found for bioavailable estradiol, but no relationship was found between the risk of cognitive impairment and testosterone concentration levels.

The authors conclude that women with higher concentration levels of non–protein-bound and bioavailable estradiol were less likely to develop cognitive decline; however, no relationship was found with testosterone concentration levels. The authors indicated that their findings add to a growing body of evidence relating estrogen to the prevention of cognitive impairment in older women, but acknowledge that further studies are needed in this complex area.