Cognitive Function in Postmenopausal Women Treated with Raloxifene

Abstract & commentary

Synopsis: Treatment with raloxifene 60 mg or 120 mg for 3 years did not affect overall cognitive scores in postmenopausal women with osteoporosis.

Source: Yaffe K, et al. N Engl J Med. 2001;344:1207-1213.

Cognitive function was studied in 7705 postmenopausal women with osteoporosis who were enrolled in the Multiple Outcomes of Raloxifene Evaluation (MORE) trial. The primary outcomes were bone mineral density and vertebral fractures. Postmenopausal women who met criteria for osteoporosis (14,674 were excluded) were enrolled via 180 sites in 25 countries and randomized to receive either raloxifene 60 mg or 120 mg or placebo for 3 years. There were roughly 2500 women in each arm. Cognitive testing was done using the Short Blessed Test for orientation, concentration, and memory; the Trails Making Test for visuospatial scanning, sequential processing, motor speed, executive function, and attention; and the Word List Memory and Recall tests for memory if the subject spoke English, French, or Spanish (4424 women).

The mean age of the women at enrollment was 66 years old. There were no significant differences among the 3 treatment groups on any of the cognitive tests at baseline or after 3 years. More women in the raloxifene groups than in the placebo group reported new or worsened hot flashes, but there was no correlation between presence or absence of hot flashes and cognitive performance.

COMMENT by Sarah L. Berga, MD

The main conclusion of this study is that short-term raloxifene exposure does not cause cognitive decline in relatively young postmenopausal women with osteoporosis. This study does not tell us whether long-term raloxifene use would increase, decrease, or not alter the risk of dementia. Less than 5% of the enrollees developed dementia while participating in the trial. The low incidence likely reflects the relatively young age of the women. This study also does not tell us about mood, sleep, libido, and other central nervous system outputs that comprise quality of life. We learn nothing about how raloxifene use might compare to estrogen use. While the study might, at first blush, be interpreted as providing reassurance about the null cognitive effects of raloxifene use, I would caution against such optimism. Further, to determine the effect of any intervention on risk of dementia, one would have to start women on the agent in question before age 70 and follow them at least 10 years. The best study design would be to randomize women to an estrogen, a selective estrogen receptor modulator (SERM), and a placebo within 1-2 years of menopause and then to compare the effects of "never" use to "always" use 30-40 years later. This is not a study likely to be done, so we will have to try to discern risks and benefits based on other investigative strategies. There are those who argue that a hiatus in exposure to an estrogen for several years sets in motion a neurodegenerative process that late in life exposure to estrogen cannot undo. Expecting to undo dementia once it is clinically apparent seems analogous to expecting late in life exposure to "undo" established kyphosis from osteoporosis. Because of the inherent plasticity of the brain, by the time dementia is evident, more than 30% of the neurons are dead. It seems ridiculous to expect estrogen or SERM use to resurrect dead neurons as much as it seems unrealistic to expect estrogen to restore vertebral bodies to their original shape once compressed. Like many age-related infirmities, the best treatment for dementia is prevention.

Is there reason to suspect that raloxifene may be less beneficial for CNS than estrogen? I believe that there is. To make a complicated story short, the brain bits that subserve memory and other cognitive functions have estrogen receptors. They express both ERa and ERb, but it appears that ERa occupation is critical to maintaining key neurotransmitter systems and for protecting the brain from both oxidative and anoxic injury. Thus, while raloxifene may have some estrogenic action in the brain, it is unlikely to match the panoply of physiological actions that the cognate ligand, estradiol, confers by fully occupying ER (a, b, and membrane).1

We are forced to make educated guesses based on the available information. My educated guess based on the extant animal and primate literature is that the most neuroprotection ligand will be an estrogen, not a SERM that is an ERa antagonist. Phytoestrogens apparently occupy ERb and not ERa, so while they do not antagonize ERa-mediated activities, they also are unlikely to provide full neuroprotection based on what we understand today. Our job is to help our patients "pick their poison" based on their risk factors, underlying conditions, and personal preferences. It is a difficult task even when the patient is well and with few complaints. Stay tuned for an evolving story.

Dr. Berga is Professor and Director, Division of Reproductive Endocrinology and Infertility, University of Pittsburgh, Pittsburgh, Pa.


1. Neele SJM, et al. J Clin Endocrinol Metab. 2001;86: 1422-1424.