Background Previous studies have found no association between serum concent
rations of total oestradiol and cognitive function, but these measurements
may not reflect concentrations of hormone available to the brain. We tested
the hypothesis that concentrations of non-protein-bound (free) and loosely
bound (bioavailable) sex hormones are associated with cognitive function i
n older women.
Methods We measured cognitive performance with a modified mini mental statu
s examination (mMMSE) at baseline (1986-88) and 6 years later in 425 women
(65 years or older). Concentrations of non-protein-bound and bioavailable o
estradiol and total and non-protein-bound testosterone were measured by RIA
in serum samples taken at baseline.
Findings Initial cognitive scores did not differ by tertile of non-protein-
bound oestradiol, bioavailable oestradiol, or testosterone. Cognitive impai
rment (a decrease of 3 points or more in mMMSE score) occurred in five (5%)
of 94 women in the high tertile for non-protein-bound oestradiol and in 17
(16%) of 106 in the low tertile (odds ratio 0.3 [95% CI 0.1-0.8]). After a
djustment for age, years of education, body-mass index, current oestrogen u
se, history of surgical menopause, and baseline mMMSE score, the odds ratio
was 0.3 (0.1-0.9), The results were similar for bioavailable oestradiol (f
ive [5%] vs 15 [15%]; adjusted odds ratio 0.3 [0.1-1.0]). There was no asso
ciation between risk of cognitive impairment and serum testosterone.
Interpretation Women with high serum concentrations of non-protein-bound an
d bioavailable oestradiol, but not testosterone, were less likely to develo
p cognitive impairment than women with low concentrations. This finding sup
ports the hypothesis that higher concentrations of endogenous oestrogens pr
event cognitive decline.