Early Initiation of Hormone Therapy in Menopausal Women Is Associated with Increased Hippocampal and Posterior Cingulate Cholinergic Activity

Abstract

Context:

The role of ovarian hormones in maintaining neuronal integrity and cognitive function is still debated. This study was undertaken to clarify the potential relationship between postmenopausal hormone use and the cholinergic system.

Objective:

We hypothesized that early initiated hormone therapy (HT) preserves the cholinergic system and that estrogen therapy (ET) would be associated with higher levels of acetylcholinesterase activity in the posterior cingulate cortex and hippocampus compared to estrogen plus progestin therapy (EPT) or no HT.

Design and Setting:

We conducted a cross-sectional study at a university teaching hospital.

Patients:

Fifty postmenopausal women (age, 65.2 ± 0.7 yr) with early long-term HT (n = 34; 13 ET and 21 EPT) or no HT (n = 16) participated in the study.

Interventions:

There were no interventions.

Main Outcome Measure:

We measured cholinergic activity (acetylcholinesterase) in the hippocampus and posterior cingulate brain regions as measured by N-[¹¹C]methylpiperidin-4-yl propionate and positron emission tomography as a marker of cholinergic function.

Results:

Significant effects of treatment on cholinergic activity measures were obtained in the left hippocampus (F = 3.56; P = 0.04), right hippocampus (F = 3.42; P = 0.04), and posterior cingulate (F = 3.76; P = 0.03). No significant effects were observed in a cortical control region. Post hoc testing identified greater cholinergic activity in the EPT group compared to the no-HT group in the left hippocampus (P = 0.048) and posterior cingulate (P = 0.045), with a nonstatistically significant trend in the right hippocampus (P = 0.073).

Conclusions:

A differential effect of postmenopausal ET and EPT on cholinergic neuronal integrity was identified in postmenopausal women. The findings are consistent with a preservation of cholinergic neuronal integrity in the EPT group.

The role of estrogen in maintaining neuronal integrity and cognitive function is of great significance. In women, evidence suggests that postmenopausal hormone therapy (HT) is associated with improved or preserved cognitive function; however, not all studies support this (for reviews, see Refs. 1 and 2). The Women's Health Initiative Memory Study (WHIMS) found an increased risk of dementia in hormone users (3) and a potential differential effect of hormone combinations (3–5). Specifically, the combined HT (conjugated equine estrogens plus medroxyprogesterone acetate) group demonstrated an increased risk of dementia (4), whereas the CEE group demonstrated a trend but no significant difference for greater dementia risk compared with controls (3). This contrasts with previous prospective observational studies suggesting that HT reduces the risk or delays the onset of Alzheimer's disease (AD) (6–8). The Women's Health Initiative Study of Cognitive Aging (WHISCA) also demonstrated different neuropsychological effects of the hormone preparations (9, 10): spatial processing improved with combination therapy and worsened with estrogen only.

To explain discrepancies between observational and randomized trials, it has been suggested that the timing of HT treatment (from menopause onset) may be an important factor in neuroprotection (11–13). Data from animal models have shown that early HT after ovariectomy, compared with delayed treatment, more effectively preserves hippocampal CA1 synaptic density in rats (14) and maintains cognitive function in rats and nonhuman primates (15, 16). In women, the cognitive effects of estrogen initiated early in menopause have not been well studied. A reassessment of randomized studies, evaluating treatment early or later in menopause, suggests that hormone use early, rather than later, in menopause may provide cognitive benefit (2). Furthermore, a follow-up study of women randomized to HT or placebo for 2–3 yr in early postmenopause found reduced risk of cognitive impairment in those who previously received HT (17). Likewise, early initiators of HT performed better than late initiators on the Mini-Mental State Examination (MMSE) and during an attentional task (18).

The source of such neuroprotective effects of HT on cognitive function is not fully understood; however, the cholinergic system, known to be critically involved in cognition, memory, and the aging process, is a major brain target for hormone activity (19, 20). Estrogen receptors are present throughout the cerebral cortex (21), as well as within the nuclei of the basal forebrain, a major source of cholinergic innervation (22). Estrogen provides trophic support to cholinergic cells and regulates various markers of cholinergic function, including choline acetyltransferase and acetylcholine release (22–28).

In the ovariectomized primate model, both short-term and long-term HT preserved cholinergic fibers (29–31). In women, manipulation of cholinergic activity can alter cognitive functioning. For instance, anticholinergic therapy after a medically induced menopause resulted in more false-positive errors in verbal recognition and reduced frontal functional magnetic resonance imaging (fMRI) activation (32). Furthermore, in most animal and human models, the detrimental cognitive effects of anticholinergics are attenuated by pretreatment or cotreatment with estrogen (33–36), although not in all (37).

Studies have attempted to link cholinergic system functioning to HT in women using neuroimaging techniques that allow noninvasive study of cholinergic synaptic densities. A study of postmenopausal estrogen therapy (ET) and brain muscarinic receptor density using single photon emission computed tomography (SPECT) and (R,R)[¹²³I] I-QNB showed that long-term ET, compared with no ET, was associated with higher muscarinic receptor concentrations in the hippocampus, left striatum, frontal cortex, and thalamus. Furthermore, peripheral estradiol levels correlated with muscarinic receptor densities (38). In a previous pilot study, we examined the relationship between postmenopausal ET and the cholinergic system using SPECT and [¹²³I]iodobenzovesamicol, labeling the presynaptic vesicular acetylcholine transporter, a measure of cholinergic terminal density. In that work, we showed that the length of HT use was positively associated with greater concentrations of cholinergic synaptic terminals in multiple cortical regions, and that the ET group had higher cholinergic synaptic concentrations in the posterior cingulate region (an associate cognitive region) than the estrogen plus progestin therapy (EPT) group. Most likely because of the small sample size, we were unable to identify a difference in cholinergic density between the HT and no-HT therapy groups (39).

The current cross-sectional study was undertaken to examine this question in greater detail and to clarify the potential relationship between postmenopausal hormone use and the cholinergic system. For this purpose, we quantified acetylcholinesterase (AChE) activity, a surrogate of cholinergic functional capacity, using the positron emission tomography (PET) radioligand N-[¹¹C]methylpiperidin-4-yl propionate ([¹¹C]PMP) (40–42). We studied healthy postmenopausal women treated with ET, EPT, or no HT. We studied women who initiated HT within 2 yr of menopause and hypothesized that initiating HT before the development of central neurovascular pathology would preserve neurochemical systems involved in cognition. We further expected that ET would be associated with higher levels of AChE concentrations in the posterior cingulate cortex and hippocampus compared with EPT or control groups.

Subjects and Methods

Subjects

As part of a comprehensive evaluation of menopausal HT, 50 healthy right-handed postmenopausal women, 60 yr or older, were recruited by advertisement (43). Menopause was defined as the absence of menstrual periods for 1 yr, the onset of severe symptoms after hysterectomy, or the time of hysterectomy with bilateral oophorectomy. Women included in the study either had never used hormones (n = 16) or had taken HT continuously for at least 10 yr (n = 34; 13 ET and 21 EPT). The hormone group began treatment within 2 yr of menopause and included both current hormone users and women who had recently stopped HT. All individuals on HT used an identical dose and preparation of estrogen: 0.625 mg/d CEE (Premarin; Wyeth Ayerst, Philadelphia, PA), with or without cyclic or continuous medroxyprogesterone acetate (Provera; Pfizer, New York, NY; or Prempro, Wyeth Pharmaceuticals, Philadelphia, PA).

Eleven of the 34 hormone-treated women were currently taking hormones (four of 13 women in the ET group and seven of 21 women in the EPT group). For those no longer taking hormones, mean time since hormone use ended was 2.2 ± 0.2 yr. All women in the ET group had undergone hysterectomy, including nine of 13 women (69.2%) with bilateral oophorectomy.

Subjects underwent an initial phone screen followed by a complete medical, psychiatric, and neurological history and physical exam in the Medical Clinical Research Unit. A neuropsychological battery of tests was given including: 1) Mini-Mental State Examination (44), a brief screening measure of dementia; 2) Shipley Institute of Living Scale (45), a short estimate of intellectual power; and 3) Geriatric Depression Rating Scale (46).

Screening laboratory tests included electrolytes, glucose, complete blood count, TSH, and estradiol. Exclusion criteria included acute or uncorrected medical illnesses, the use of centrally acting medications, intermittent estrogen use, phytoestrogen supplements, smoking within the last 5 yr, inability to tolerate scanning procedures, and contraindications to magnetic resonance imaging (MRI).

After a full description of the study, written informed consent was obtained. All procedures were approved by the University of Michigan's institutional review board and the radiation safety committee.

Scanning procedures and image processing

Subjects were positioned in the scanner gantry, and an iv line was placed in an antecubital vein. A light forehead restraint was used to minimize intrascan head movement. Small head movements during the emission scans were corrected using an automated computer algorithm (47).

AChE activity was defined with PET to calculate tracer kinetic estimates of the local hydrolysis rate of [¹¹C]PMP. The [¹¹C]PMP radioligand, an acetylcholine analog that is hydrolyzed by AChE (48), was prepared by N-[11C] methylation of piperidin-4-yl propionate in high radiochemical purity (42). Blood-brain barrier transport rate (K₁R) and AChE activity measures (k₃) were calculated using a kinetic modeling approach that does not require arterial plasma sampling (49, 50). The average injected radioactivity was 18.4 ± 1.5 mCi (mean ± sd).

Emission data were collected as a sequence of 17 image dynamic PET frames over an 80-min scanning period using a Siemens ECAT Exact HR+ scanner (Siemens, Knoxville, TN) operated in three-dimensional mode with septa retracted. Images were reconstructed using Fourier rebinning and the iterative OSEM routine (four iterations, 16 subsets) resulting in images with a full-width at half-maximum resolution of approximately 5.5 mm both in-plane and axially. Attenuation correction was performed using a 6-min transmission scan ([⁶⁸Ge] source) obtained before the emission study, also with iterative reconstruction of the blank/transmission data followed by segmentation of the attenuation image and reprojection.

Anatomical MRI scans were acquired axially using a 3T whole-body MRI scanner (General Electric, Milwaukee, WI) equipped with a standard head coil. A T1-weighted coronal image set was acquired with a spoiled gradient recalled three-dimensional volumetric acquisition [repetition time = 9.6, echo time = 3.3, inversion recovery preparation = 200 msec, flip angle = 17°, bandwidth = 15.63, 24-cm field of view, 1.5-mm slice thickness, 106–110 slices, 256 × 256 matrix, and two excitations].

Parametric K₁R and k₃ images were coregistered to the subject's magnetic resonance (MR) images using SPM2 software (Wellcome Department of Cognitive Neurology, London, UK). Because spatial normalization of older brains to standard templates generated from young adults is not ideal, we used a minimal deformation template (MDT2) derived from 25 older normal subjects for the anatomical normalization of the PET and MR images, developed by the Imaging of Dementia and Aging Laboratory at the University of California, Davis (51). The quality of coregistration and normalization was confirmed for each subject individually by comparing the transformed MR and PET images to each other and the MDT2 template.

Analysis

Volumes of interest (VOI) in the posterior cingulate and bilateral hippocampus were defined based on a priori hypotheses regarding hormonal effects on AChE functioning in the aging brain. Based on the results from a previous study (39), we extracted data from the posterior cingulate using a 20-mm diameter spherical region of interest with voxel coordinates x, y, z = 0, −50, and 27, respectively. For the hippocampal VOI, we used hippocampal templates created specifically for the MDT2 template [anatomical boundaries described in Sun et al. (51)]. Data for a control region not typically affected in the dementias was also examined to test the specificity of the findings. This VOI was generated using the gray matter cortical boundaries of the motor and premotor cortex [Brodmann areas (BA) 4 and 6] and medial frontal cortex (BA 8–11) using predefined templates applied in stereotactic space (52).

Possible differences in regional cholinergic activity indices between groups were examined by ANOVA or analysis of covariance (ANCOVA) with Tukey post hoc testing. Pearson correlations were analyzed to determine the relationship between regional cholinergic activity, age, and covariates of interest.

Results

Demographic and baseline information for study participants is described in Tables 1 and 2. The average age of participants was 65.2 ± 0.7 yr. The neuropsychological data showed normal-range IQ and absence of dementia and depression in all groups. There were no differences between the three groups in neuropsychological test results, age, or education. Post hoc two-group comparisons show that age at HT initiation and duration of hormone use differed between the ET and the combined EPT groups, with the ET group initiating HT earlier and maintaining in treatment longer. There were no statistical differences between current and past hormone users.

Table 1.

Demographic and baseline information for study participants

	Mean ± sd			P t-Testa	Mean ± sd Never treated	P ANOVAb
	All subjects	ET	EPT
n	50	13	21		16
Age (yr)	65.2 ± 4.8	66.2 ± 4.3	64.4 ± 4.8	0.30	65.6 ± 5.1	0.56
Education (yr)	16.6 ± 2.4	15.8 ± 2.3	17.3 ± 2.6	0.11	16.3 ± 2.0	0.20
Age began HT	47.7 ± 4.0	44.8 ± 3.5	49.4 ± 3.3	0.00
Years on HT	15.3 ± 5.6	18.6 ± 6.3	13.3 ± 4.1	0.01
Mini-Mental State Examination	28.6 ± 1.6	28.3 ± 1.8	28.8 ± 1.3	0.34	28.8 ± 1.9	0.65
Shipley Estimated IQ	114.3 ± 9.2	110.2 ± 9.7	113.7 ± 8.6	0.29	118.3 ± 8.3	0.06
Geriatric Depression Rating Scale	0.7 ± 0.9	1.1 ± 1.3	0.5 ± 0.8	0.10	0.6 ± 0.8	0.19

^at-Test comparison between ET and EPT groups.

^bANOVA comparison between all three treatment groups.

Table 2.

Demographic and baseline information for current and past HT users

	Current users (mean ± sd)	Past users (mean ± sd)	t-Testa (P)
n	11	23
Age (yr)	63.7 ± 4.0	65.7 ± 4.9	0.24
Education (yr)	15.5 ± 2.7	17.3 ± 2.3	0.06
Age began HT	46.8 ± 4.5	48.1 ± 3.8	0.40
Years on HT	15.1 ± 6.4	15.4 ± 5.4	0.87
Mini-Mental State Examination	28.4 ± 1.8	28.7 ± 1.3	0.66
Shipley Estimated IQ	111.8 ± 10.6	112.7 ± 8.5	0.80
Geriatric Depression Rating Scale	0.6 ± 1.1	0.7 ± 1.0	0.72
Time since HT was stopped (yr)		2.2 ± 1.0

^at-Test comparison between current and past hormone users.

Significant differences in cholinergic activity measures between the three groups were detected by ANOVA in the hippocampus and posterior cingulate (Table 3 and Fig. 1). Post hoc testing with Tukey revealed higher cholinergic activity in the EPT group compared with the no-HT group in the left hippocampus (P = 0.048) and posterior cingulate (P = 0.045), with a trend in the same direction for the right hippocampus (P = 0.073) (Fig. 2). No significant differences in cholinergic activity were observed for the reference region (Table 3). t-Test comparisons of current and former HT groups revealed no significant differences in any of these regions. In a correlational analysis that included all subjects, age was not related to regional cholinergic activity. Additionally, in the hormone group, there was no association identified between age of hormone initiation and regional cholinergic activity.

Table 3.

Average AChE activity with PMP PET

Region	Mean ± sd				P
	All subjects PMP	ET PMP	EPT PMP	No-HT PMP	EPT vs. no-HTa	ET vs. no-HTa	ET vs. EPTb	ANOVAc
n	50	13	21	16
Left hippocampus	0.047 ± 0.005	0.046 ± 0.006	0.049 ± 0.004	0.045 ± 0.005	0.048	0.954	0.121	0.036
Right hippocampus	0.043 ± 0.005	0.041 ± 0.006	0.045 ± 0.005	0.041 ± 0.005	0.073	1.000	0.123	0.041
Posterior cingulate	0.028 ± 0.002	0.027 ± 0.003	0.029 ± 0.002	0.027 ± 0.003	0.045	0.974	0.024	0.031
Reference region	0.034 ± 0.003	0.033 ± 0.005	0.035 ± 0.003	0.034 ± 0.002	0.430	0.872	0.086	0.205

PMP measure is k₃ (min⁻¹).

^aTukey honestly significant difference post hoc test.

^bANCOVA controlling for years on HT.

^cANOVA comparison between all three treatment groups.

Fig. 1.

Brain images of AChE activity (k₃ values, min⁻¹) in the selected regions of interest.

Fig. 2.

ET, EPT, and no-HT group comparisons of AChE activity.

Given that there were differences in cholinergic activity between groups and differences in years of total HT between the two hormone groups (ET and EPT), ANCOVA was performed to see whether this measure of estrogen exposure accounted for the regional cholinergic differences. Therefore, a final ANCOVA analysis included all three subject groups and years on HT as covariate, with results maintaining significant effects of treatment in the posterior cingulate (P = 0.024).

Discussion

The present report describes relationships between early initiation, long-term use of HT, and cholinergic activity, as well as the differential effects of conjugated equine estrogens and combination conjugated equine estrogens/medroxyprogesterone acetate. We identified an 8–10% greater AChE activity in the hippocampus and posterior cingulate in the EPT group compared with the no-HT group, consistent with a preservation of cholinergic neuronal integrity in that sample.

There are a number of mechanisms through which gonadal steroids may maintain cognitive function. In animal models, estrogen has been shown to affect neuronal function by modulating neurotransmission, acting as a neuroprotectant and antioxidant, increasing neurite branching and synaptogenesis, increasing cerebral blood flow, regulating β-amyloid production, and affecting trophic factors (27, 53–59). Recently, it was reported that estradiol and progesterone are potent regulators of mitochondrial function in the brain (60). Estradiol interacts with IGF-I to activate neuronal survival pathways (61), and activation of either ERα or ERβ may promote neuroprotection (62). Furthermore, genomic studies suggest that polymorphisms of ERα increase the risk of cognitive impairment (63), and that an interaction may exist between ERα polymorphisms and ApoE alleles in conferring this increased risk (64). Thus, estrogen may impact brain function through multiple mechanisms.

In this study, we focused on the effect of HT on the cholinergic neurotransmitter system. We used AChE activity to measure cholinergic neuronal function (40–42). Our finding of a positive effect of hormone use on cholinergic activity is consistent with both human and animal studies. An effect of postmenopausal HT on cholinergic function is supported by data showing trophic effects of estrogen and progesterone on cortical cholinergic neurons in nonhuman primates (31) and prior studies showing similar effects in cultured hippocampal cells (28). More recently, in ovariectomized monkeys, long-term estrogen replacement (2 yr) preserved cholinergic fibers in the prefrontal cortex, whereas placebo treatment resulted in a significant decrease (29). In addition, PET studies in primates using [¹⁸F]fluorobenzyltrozamicol, a cholinergic radiotracer selective for the vesicular acetylcholine transporter, have shown a reduction in transporter concentrations (reflecting cholinergic presynaptic density) in the monkey striatum 3 yr after ovariectomy and sustained increases in binding after initiating ET (66). Furthermore, a SPECT study of human brain muscarinic receptor concentrations demonstrated increased receptor availability in the hippocampus with ET (38).

Contrary to our previous pilot study (39) in which we were unable to detect a difference between the HT and no-HT groups, here we identified increased cholinergic activity in the EPT group, compared with both the no-HT and ET groups. In the present study, the ET group had an onset of menopause (after hysterectomy and oophorectomy in all cases) that was earlier than that of the no-HT and EPT groups. This may have influenced cholinergic functional integrity by reducing natural exposure to gonadal steroids.

Additionally, a neurochemical explanation may relate to differing neural growth actions of estrogen and progesterone. Estrogen-inducible progesterone receptors are present in the hippocampus (68). Studies report contradictory effects of progesterone on brain areas critical for cognition (69). Some data suggest a beneficial effect of progesterone (70–72). Goodman et al. (72) showed that progesterone reduced neuronal vulnerability to excitotoxic, metabolic, and oxidative injuries. However, other studies report down-regulation of dendritic spine growth when estrogen is combined with progesterone (73, 74), reversal of estrogen-induced increases in neurotrophins by progesterone (73, 75), and induction of spatial memory deficits in rat and human models (76, 77). Likewise, studies demonstrate that the synthetic progestin medroxyprogesterone acetate does not provide neuroprotection from excitotoxicity (78) and impairs memory in ovariectomized rats (79).

The posterior cingulate cortex, noted in this study to have increased cholinergic density in the EPT group, has been identified as a brain region involved in very early stages of AD. This is an associative cortical region with connections to the hippocampal formation and other brain areas involved in cognitive processing (80–85). It has been implicated in both the encoding and retrieval of episodic memory (86–89) and is thought to link verbal and nonverbal information with prior knowledge (90–93). Profound reductions in glucose metabolism in this region have been identified in patients who presented with isolated memory impairments and later developed AD (94). Furthermore, decreased glucose metabolism in the posterior cingulate, as well as in the parietal, temporal, and frontal cortices, is present in established AD (95, 96).

In a study of functional connectivity of the hippocampus during short-term memory tasks in young and older populations with PET and a regional cerebral blood flow marker, the posterior cingulate was found to be differentially activated as a function of age. The younger group activated a neural network that included the prefrontal cortex, fusiform gyrus and posterior cingulate, whereas the older subjects activated more anterior regions, but not the posterior cingulate (97). Furthermore, an effect of estrogen in this and other cortical brain regions has been demonstrated during working memory tasks in a placebo-controlled fMRI study. Specifically, the administration of conjugated equine estrogens for 21 d in postmenopausal women was associated with increased prefrontal cortical and posterior cingulate activation during the retrieval component of a working memory task (98). These data suggest that the posterior cingulate cortex undergoes functional changes during the aging process that are influenced by estrogen and have cognitive implications.

The hippocampus, a region central to cognitive function, displayed higher cholinergic activity in hormone users in this study. In AD and mild cognitive impairment, hippocampal changes are prominent and include hypometabolism, morphological alterations, neuronal atrophy, and decreased volume (99–110). In women with AD, estradiol levels in the cerebral spinal fluid are positively correlated with glucose metabolism in the hippocampus (111). Furthermore, in healthy postmenopausal women, estrogen use has been associated with greater gray matter volumes in the hippocampus, amygdala, and multiple cortical areas (112, 113). This hippocampal effect has also been noted for postmenopausal women with susceptible ApoE genes and with HT users having higher hippocampal volumes compared with controls (114). Furthermore, the use of tamoxifen, a selective estrogen receptor modulator that lowers estrogen activity, has been associated with smaller hippocampal volumes compared with women on ET (115).

The timing of onset of HT may be an important factor in neuroprotection (11–13). Our findings, in subjects selected for early initiation of hormone use, show increased cholinergic activity in brain areas critical for cognition (hippocampus and posterior cingulate) and support this critical window concept. Data from the Research into Memory, Brain Function and Estrogen Replacement (REMEMBER) pilot also indicate that early HT initiation may benefit some cognitive domains (global cognition, attention and concentration, and verbal expression), whereas late initiation may be detrimental to cognition (18). Future studies need to examine the relationship between measures of cholinergic integrity and neuropsychological testing data of relevant cognitive domains.

Our study included women with both early initiation and long-term hormone use; therefore we are unable to separate their individual effects. Neural effects of long-term hormone use are less clear in the existing literature but point toward benefits primarily with use in the earlier menopausal years (113, 116, 117). The present cross-sectional study design was also limited by the lack of prospective randomization between the HT and no-HT groups, which may allow for selection biases, and by the lack of longitudinal follow-up to assess cognitive impact. However, effects of early initiation and long-term hormone use are difficult to study in randomized, longitudinal designs because of the length of treatment time required. In addition, this design relies on patient recall on initiation of use, which may not be entirely accurate. Although HT users generally tend to be healthier and more educated (65, 67, 118), our groups were matched for education level and had no major medical illnesses. Furthermore, our hormone group included both current and past hormone users; however, these groups were similar in demographics and the average time off of HT was short (2.2 yr).

In summary, a differential effect of ET and EPT on cholinergic neuronal integrity was identified in postmenopausal women. We found that HT, particularly EPT, influences the survival or plasticity of cholinergic cells in postmenopausal women. Further understanding of the actions of estrogens and progestins in the brain will facilitate the development of individualized HT regimens, appropriate alternatives to standard HT, and medications targeted to prevent cognitive aging.