The cognitive effects of conjugated equine estrogens depend on whether menopause etiology is transitional or surgical

Abstract

The question of whether to take hormone therapy (HT) will impact every woman as she enters reproductive senescence. In women, studies suggest that ovarian hormone loss associated with menopause has deleterious cognitive effects. Results from clinical studies evaluating whether estrogen-containing HT mitigates these effects, and benefits cognition, are discrepant. Type of menopause, surgical vs. transitional, impacts cognitive outcome in women. However, whether type of menopause impacts cognitive effects of HT has not been methodically tested in women or an animal model. We used the 4-vinylcyclohexene diepoxide rodent model of ovarian follicle depletion, which mimics transitional menopause, and the traditional rat model of menopause, ovariectomy, to cognitively test the most commonly prescribed estrogen therapy in the United States, conjugated equine estrogens (Premarin). Here we show conjugated equine estrogens benefited cognition in surgically menopausal rats, but, in contrast, impaired cognition in transitionally menopausal rats. Androstenedione, released from the residual transitional menopausal ovary, was positively associated with impaired performance, replicating our previous findings in 4-vinylcyclohexene diepoxide animals. The current findings are especially salient given that no clinical study testing cognition has methodically separated these two populations of menopausal women for analysis. That we now show surgical vs. transitional modes of menopause result in disparate cognitive effects of HT has implications for future research and treatments optimizing HT for menopausal women.

Figure 1

Mean ± se working memory correct (WMC) errors (A), working memory incorrect (WMI) errors (B), and reference memory (RM) errors (C) for each group (OVX, n = 9; OVX-CEE, n = 9; VCD, n = 12; VCD-CEE, n = 13) for the lattermost testing block (block 4). The significant hormone × menopause type interaction for each error type indicates that OVX animals given CEE made fewer WMC, WMI, and RM errors than those given oil, while VCD animals given CEE made more WMC, WMI, and RM errors compared with those given oil (P < 0.05). Therefore, treatment with CEE impaired animals undergoing transitional hormone loss but benefited animals undergoing abrupt hormone loss.

Figure 2

Mean ± se working memory correct (WMC) errors for each group (OVX, OVX-CEE, VCD, and VCD-CEE) during the lattermost testing block (block 4). The inset shows mean ± se WMC errors for each group for trial 4. There was a significant trial × hormone × menopause type interaction for WMC errors, indicating that the pattern of change as trials increased differed with CEE treatment and menopause type. Further analyses revealed a hormone × menopause type interaction for trial 4, the trial with the highest working memory load. The data indicate that OVX animals given CEE were better able to handle the highest memory load than those given oil, and VCD animals given CEE were less able to handle the highest memory load than those given oil (P < 0.05). CEE therefore enhanced the ability to handle a more demanding memory load in animals with surgical, abrupt hormone loss but impaired the ability to handle a more demanding memory load in animals that had experienced transitional hormone loss.

Figure 3

Mean ± se total errors on the DMS plus maze 8-h postdelay trial. OVX-CEE animals made fewer errors relative to OVX animals after an 8-h delay (P < 0.005). CEE did not impact postdelay performance if animals had undergone transitional menopause via VCD treatment.

Figure 4

Mean ± se distance on the MWM for d 1–5. Group comparisons set post hoc revealed that on d 3 OVX-CEE rats showed decreased distance scores relative to OVX rats (P = 0.05). There was no CEE effect on this day for VCD-treated animals (P = 0.32). Thus, CEE enhanced performance in OVX rats but not VCD rats on d 3 of MWM testing, indicating faster learning by the OVX-CEE group.

Figure 5

Mean ± se serum androstenedione levels for each group. Androstenedione levels were higher in VCD animals relative to OVX animals (P < 0.01).

Figure 6

Regression analysis indicated that in VCD animals (VCD and VCD-CEE), higher androstenedione was associated with more WRAM working memory correct, working memory incorrect, and reference memory errors on the last testing block (P < 0.05). For ease of presentation, here we present only the combined graph with total errors (P < 0.01).