Understanding the cognitive impact of the contraceptive estrogen Ethinyl Estradiol: tonic and cyclic administration impairs memory, and performance correlates with basal forebrain cholinergic system integrity

Abstract

Ethinyl Estradiol (EE), a synthetic, orally bio-available estrogen, is the most commonly prescribed form of estrogen in oral contraceptives, and is found in at least 30 different contraceptive formulations currently prescribed to women as well as hormone therapies prescribed to menopausal women. Thus, EE is prescribed clinically to women at ages ranging from puberty to reproductive senescence. Here, in two separate studies, the cognitive effects of cyclic or tonic EE administration following ovariectomy (Ovx) were evaluated in young female rats. Study I assessed the cognitive effects of low and high doses of EE, delivered tonically via a subcutaneous osmotic pump. Study II evaluated the cognitive effects of low, medium, and high doses of EE administered via a daily subcutaneous injection, modeling the daily rise and fall of serum EE levels with oral regimens. Study II also investigated the impact of low, medium and high doses of EE on the basal forebrain cholinergic system. The low and medium doses utilized here correspond to the range of doses currently used in clinical formulations, and the high dose corresponds to doses prescribed to a generation of women between 1960 and 1970, when oral contraceptives first became available. We evaluate cognition using a battery of maze tasks tapping several domains of spatial learning and memory as well as basal forebrain cholinergic integrity using immunohistochemistry and unbiased stereology to estimate the number of choline acetyltransferase (ChAT)-producing cells in the medial septum and vertical/diagonal bands. At the highest dose, EE treatment impaired multiple domains of spatial memory relative to vehicle treatment, regardless of administration method. When given cyclically at the low and medium doses, EE did not impact working memory, but transiently impaired reference memory during the learning phase of testing. Of the doses and regimens tested here, only EE at the highest dose impaired several domains of memory; tonic delivery of low EE, a dose that corresponds to the most popular doses used in the clinic today, did not impact cognition on any measure. Both medium and high injection doses of EE reduced the number of ChAt-immunoreactive cells in the basal forebrain, and cell population estimates in the vertical/diagonal bands negatively correlated with working memory errors.

Keywords: Contraceptive; Ethinyl Estradiol; Hormone; Memory.

Figure 1

WRAM Performance Study I. During the asymptotic phase of testing, there was a Trial × Treatment interaction for both WMC [F_(4,52)=4.14; p<0.01] and WMI [F_(6,78)= 4.23; p<0.01] errors. Two group planned comparisons showed that the high EE group committed more WMC [F_(2,36)=3.38; p<0.05] and WMI [F_(3,54)=5.06; p<0.01] errors than the vehicle group as working memory load increased. High EE animals also made more WMI and WMC errors than low EE animals on trial 4, the trial with the highest working memory load (Fisher, p<0.05). There was also a main effect of Treatment for WMI errors across all trials [F_(2,24)= 4.99; p<0.05] with the high EE group making more errors than the vehicle group [F_(1,18)= 4.38; p=0.05], and the low EE group (Fisher, p<0.05). There were no differences between the low EE and vehicle groups, nor were there any group differences in RM errors.

Figure 2

MM Performance Study I. a) There were no group differences for days 1–3 of MM testing (p>0.05). b) For the probe trial, a higher percent distance was spent in the previously platformed quadrant vs. the opposite quadrant [F_{(1, 26)}=86.84, p<0.0001] with no percent distance by treatment interaction, indicating that all groups spatially localized the platform by the end of testing.

Figure 3

WRAM Performance Study II. During the asymptotic phase of testing, there was a Trial × Treatment interaction for WMC errors [F_(6,64)= 2.82; p<0.05] with the high EE treated animals making more errors than vehicle treated animals as working memory load increased [F_(2,32)= 5.78; p<0.01]. Post-hoc analyses also showed that the high EE group committed more WMC errors than the low EE (Fisher, p<0.05) and medium EE (Fisher, p<0.05) animals at the highest working memory load. There were no differences in WMC errors between the vehicle group and the low or med EE group. There were no differences for WMI or RM errors.

Figure 4

MM Performance Study II. a) There was a Treatment × Day interaction for MM testing [F_(6,64)=2.21; p=0.05]. Further analyses revealed a main effect of Treatment [F_(3,32)= 3.22; p<0.05] for Day 1 of MM, whereby the vehicle group performed better than each of the low EE [F_(1,16)= 6.84; p<0.05], med EE [F_(1,16)= 8.51; p<0.05], and high EE [F_(1,16)= 9.47; p<0.01] groups. There was no Treatment × Trial interaction for Day 1, indicating that this effect was present across all trials and was not carried by the initial exposure to the task. There were no effects of Treatment during days 2–3 of MM testing. b) A higher percent distance was spent in the previously platformed quadrant vs. the opposite quadrant [F_(1,32)= 374.33; p<0.0001] for the probe trial, with no percent distance by treatment interaction, indicating that all groups spatially localized the platform equally by the end of testing.

Figure 5

ChAT Cell Population Estimates Study II. a) There was a main effect of Treatment [F_(3,12)=3.66; p<0.05] in the VDB with a larger ChAT-IR cell population estimate in the vehicle-treated animals than both the med EE- (Fisher, p<0.05) and high EE-treated animals (Fisher, p=0.05). ChAT-IR cell population estimates in the low EE group did not differ from those in the vehicle group. ChAT-IR cell population estimates in the low EE group did not differ from those in the vehicle group, however the ChAT-IR cell population estimate in the low EE group were higher than in the med EE group (Fisher, p<0.05), and marginally higher than in the high EE group (Fisher, p=0.09). b) There were no effects of Treatment on ChAT-IR cell population estimates in the MS. c) There was a main effect of Region, whereby there were fewer cells overall in the MS relative to the VDB [F_(1,12)=96.49, p<0.001]. d) There was a positive correlation between ChAT-IR cell counts in the VDB and MS of the basal forebrain (r=0.56, p<0.05). e) There was a negative correlation between ChAT-IR cell population estimates in the VDB and the average number of WMC errors on trial 4 during the asymptotic portion of WRAM testing [r= −0.55; p<0.05].