Serum estrogen levels, cognitive performance, and risk of cognitive decline in older community women

OBJECTIVE: To determine the association between serum estrogen levels, cognitive performance, and risk of cognitive decline in older women. DESIGN: Prospective cohort study with an average follow-up of 5 years. SETTING: Clinical centers in Baltimore, MD, Minneapolis, MN, Portland, OR, and the Monongahela Valley in Pennsylvania. PARTICIPANTS: 532 women aged 65 years or older who were the controls from two nested case-control studies in the ongoing Study of Osteoporotic Fractures. OUTCOME MEASURES:Three cognitive tests - a modified Mini-Mental Status Exam, Digit Symbol, and Trails B - were administered at study initiation and were then repeated approximately 5 years later. Estrone and estradiol levels were determined by radioimmunoassay at two laboratories from baseline stored serum. RESULTS: The characteristics of the women in the four serum estrogen quartiles did not differ except that body weight and change in weight since age 50 increased directly with higher quartile of serum estrogen (P < .001, for both estrone and estradiol). Initial cognitive performance on all three tests did not differ consistently by quartile of estradiol or by the estradiol to estrone ratio. Women in the higher estrone quartiles had 15% lower (worse) scores on Digit Symbol compared with the lower quartiles (P = .004) but there was no difference by quartile on the modified MMSE or on Trails B. Cognitive function test scores declined over the 5 years of follow-up. There was no difference in amount of change by quartile of estradiol, but women in the higher estrone quartiles had greater reduction of scores on Trails B compared with those in the lower quartiles (P = .012), even after adjusting for age, education, depression, stroke history, weight, and change in weight since age 50. The age-adjusted odds of cognitive decline (defined as tenth percentile of women with the largest decline in cognitive performance) did not vary across quartile of estrone or estradiol. CONCLUSIONS: Endogenous estrogens are not associated consistently with cognitive performance or risk of cognitive decline on a selected battery of cognitive tests in older community-dwelling women. Worse performance on two cognitive tests among women with higher estrone levels was surprising and warrants further investigation.