U

v

wefpsoytsmtmonpsettcct

SE

Neuroscience 191 (2011) 129–138

REVIEW

EFFECTS OF ENDOGENOUS AND EXOGENOUS ESTROGENEXPOSURES IN MIDLIFE AND LATE-LIFE WOMEN ON EPISODIC

MEMORY AND EXECUTIVE FUNCTIONS

Kt

CR

U

CR

V. W. HENDERSONa,b* AND R. A. POPATa

aDepartment of Health Research and Policy (Epidemiology), Stanfordniversity, Stanford, CA, USA

bDepartment of Neurology and Neurological Sciences, Stanford Uni-ersity, Stanford, CA, USA

Abstract—Cognitive aging affects episodic memory and ex-ecutive functions, and these vulnerable domains are postulatedto be modulated by endogenous and exogenous estrogen ex-posures. In midlife and late-life women without dementia, estro-gen effects on cognition can be examined through associationswith concentrations of serum estrone and estradiol and throughclinical trials of estrogen-containing hormone therapy. To thisend, we reviewed published studies including at least 100women (larger studies are less prone to publication bias) ad-dressing associations between estrogen levels and perfor-mance on neuropsychological tests of episodic memory or ex-ecutive functions (including working memory; seven studies),or that reported results of placebo-controlled clinical trials ofhormone therapy with objective measures within these cogni-tive domains (eight studies). Results were considered sepa-rately for midlife and late-life (age>65 years) women. There

ere no consistent associations between endogenous serumstrogen concentrations and episodic memory or executiveunctions in naturally menopausal midlife women or in olderostmenopausal women. Clinical trial findings suggested noubstantial impact of exogenous estrogens on episodic mem-ry or executive functions over time frames of up to severalears. A quantitative synthesis of clinical trial results supportedhe inference of absence of effect. This overall conclusion of noubstantial effect on episodic memory or executive functionsight reassure women concerned by potential adverse cogni-

ive consequences of menopause or of relatively short-termidlife hormone therapy. There was no apparent window ofpportunity during which exogenous hormones might benefitear-term cognition, but included studies provided limitedower to identify such a window. Conclusions are tempered bymall numbers of studies, imprecise estimates of long-termstrogen exposures, and narrow range of neuropsychologicalests. Long-term (late-life) cognitive consequence of midlife es-rogen exposures are poorly addressed by current data, as areognitive consequences of surgical menopause and cognitiveonsequences of exogenous estrogens during the menopauseransition.

This article is part of a Special Issue entitled: Neuroactiveteroids: Focus on Human Brain. © 2011 IBRO. Published bylsevier Ltd. All rights reserved.

*Correspondence to: V. W. Henderson, Stanford University School ofMedicine, 259 Campus Drive, Stanford, CA 94305-5405, USA.E-mail address: vhenderson@stanford.edu (V. W. Henderson).

Abbreviations: SMD, standardized mean difference; Trails-B, TrailMaking Test, Part B.

0306-4522/11 $ - see front matter © 2011 IBRO. Published by Elsevier Ltd. All righdoi:10.1016/j.neuroscience.2011.05.059

129

ey words: aging, estrogen, executive functions, hormoneherapy, menopause, memory.

Contentsognition, cognitive aging, and dementia 130eview and synthesis of human studies of estrogen

exposures 130Endogenous exposures 131

Endogenous exposures in midlife women 131Endogenous exposures in older postmenopausal women 131

Exogenous exposures 131Exogenous exposures in midlife women 133Exogenous exposures in older postmenopausal women 134Quantitative synthesis of clinical trial findings 134

nresolved issues 134Surgical menopause 135Exogenous estrogen exposures during the menopause

transition 135Cognitive outcomes in late-life 135onclusions 135eferences 136

Natural menopause reflects the permanent loss of ovarianfollicles, defined clinically by 12 months of amenorrheafollowing a final menstrual period. It represents a nearcomplete cessation in ovarian hormone secretion (Souleset al., 2001). Menopause is thus characterized by sharpdecrements in concentrations of the primary ovarian estro-gens (estradiol and estrone) and progesterone (Trévoux etal., 1986; Burger et al., 2007). For many women, themenopause transition is punctuated by fluctuations andinstability in serum estrogen levels (Burger et al., 2007).Such changes in the internal endocrine milieu could po-tentially affect cognition during midlife and beyond, and arecommonly believed to do so (Hogervorst et al., 2009). Anunderstanding of potential cognitive consequences forwomen, however, is of necessity derived largely from hu-man observational research, where causality is difficult toinfer. It is possible, however, to address other importantquestions about cognition and menopause experimentally.In particular, evidence for exogenously administered hor-mones, such as estrogen-containing hormone therapy inpostmenopausal women, is available from placebo-con-trolled clinical trials.

This review of cognitive effects of estrogens in womenemphasizes endogenous and exogenous exposures thatoccur during and after midlife. The focus is on episodic

memory and executive functions, two cognitive domainsts reserved.

darplpauiiipcv2twtdat

sp

oPtstlviiprlmIWaats

epepcgeOwfm2amc2eia

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138130

viewed as vulnerable to effects of aging (Deary et al.,2009). Age or temporal proximity to the menopause ishypothesized to modify some actions of estrogens in thebrain, as reflected in the so-called critical window, or win-dow of opportunity, hypothesis (Resnick and Henderson,2002; Sherwin, 2009), and we consider studies in midlifewomen separately from those of older women.

COGNITION, COGNITIVE AGING, ANDDEMENTIA

The term cognition refers to brain processes by whichknowledge is acquired, stored and used. Cognition encom-passes attention and concentration, learning and memory,language, complex perceptual and motor abilities, and plan-ning, judgment and reasoning. Although brain organizationfor cognitive function is not necessarily modular, a first ap-proach can consider one cognitive domain as at least partiallyindependent of another. A caveat is that optimal performanceon a particular cognitive task draws upon various mentalresources, and no neuropsychological test can quantify cog-nitive function strictly delimited to a single domain.

Cognitive abilities vary with age. In many instances,age-associated changes involve an erosion of skills. Thisis a process that begins insidiously in early adult life ormiddle age, accelerates during old age, and affects bothepisodic memory and executive functions (Deary et al.,2009; Salthouse, 2009). Cognitive aging is usually concep-tualized as differing from pathological processes—for ex-ample, the accumulation of neurofibrillary tangles and neu-ritic plaques in Alzheimer’s disease—that lead to severecognitive impairments characteristic of dementia, but thedistinction is not absolute. For both episodic memory andexecutive functions, early or excessive decline is viewedas an early, preclinical indicator of Alzheimer’s disease(Albert et al., 2001; Grober et al., 2008).

Episodic memory refers to the ability to recall in aeliberate manner recent episodes or events. It is oftenssessed by delayed recall tasks, such as the ability toecall items from a word list or narrative details from aaragraph story, tested after some delay of minutes or

onger. Normal episodic memory performance dependsarticularly on integrity of the hippocampus and adjacentreas of the medial temporal lobes (Squire, 2009). Exec-tive functions are controlled, non-automatic processes

nvolved in problem solving and other goal-directed behav-ors. These include selective attention, judgment, reason-ng, planning, initiation, inhibition, and monitoring of taskerformance. This group of interrelated cognitive pro-esses is thought particularly, but not exclusively, to in-olve the lateral prefrontal cortex (Cummings and Miller,007). Working memory, viewed as one aspect of execu-ive functions, refers to the cognitive demands involvedith the temporary storage and manipulating of informa-

ion. Working memory can be assessed with such tests asigit span backward, Letter-Number Sequencing, serialddition, n-back tests, and Brown-Peterson distraction

ests. For the digit span backward test, a woman hears a

tring of single-digit numbers, which she immediately re-eats in reverse order.

Neuropsychological tasks commonly used to assessther executive functions include the Trail Making Test,art B (Trails-B; divided attention and response alterna-

ion) and item generation (fluency) tests (initiation andelf-regulation). The former is a timed pencil and paperask that involves the ability to alternate between circlesabeled sequentially by number and by letter. A commonerbal fluency task entails the generation of names of

tems belonging to a particular category during a fixed timenterval, such as generating as many animal names asossible over a period of 1 min. Another fluency taskequires the generation of names starting with a specifiedetter of the alphabet. Other executive function tasks are

aze and tower tests (planning), the Stroop Color Wordnterference Test (selective attention and set changing), the

isconsin Card Sorting Test (set changing), and digit-symbolnd symbol-digit substitution tests (scanning and trackingspects of attention, and working memory). Details of these

ests are found in standard texts of neuropsychological as-essment (Lezak et al., 2004; Spreen et al., 2006).

There is a strong biological rational for consideringstrogen effects on cognitive functions that implicate hip-ocampus and frontal neocortex. In animal studies, forxample, estradiol influences synaptic plasticity in the hip-ocampus and prefrontal cortex and improves aspects ofognitive function believed to implicate these cortical re-ions (Woolley and McEwen, 1993; Rapp et al., 2003; Haot al., 2006; Mukai et al., 2006; Liu et al., 2008; Foy, 2011).ther support comes from functional neuroimaging studieshere estrogens appear to modulate medial temporal and

rontal lobe activity during cognitive task performance (Ber-an et al., 1997; Joffe et al., 2006; Berent-Spillson et al.,010; Maki et al., 2011). Animal work supports consider-tion of whether cognitive effects of estrogen exposuresight vary by age. Effects of estradiol on synaptic plasticity

an differ between young and aged rats (Yildirim et al.,008) and primates (Hao et al., 2007), and behavioralffects of estradiol are influenced by timing of treatment

nitiation after ovariectomy (Daniel et al., 2006; Bohaceknd Daniel, 2010).

REVIEW AND SYNTHESIS OF HUMAN STUDIESOF ESTROGEN EXPOSURES

We systematically identified two sets of published studiesinvolving midlife and late-life women, according to criteriadescribed below. The first set examined associations be-tween serum concentrations of estrone or estradiol (en-dogenous exposures) and cognitive function, and the sec-ond set reported placebo-controlled trials of estrogen-con-taining hormone therapy with or without a progestagen(exogenous exposures). Because smaller studies are morevulnerable to publication bias (Easterbrook et al., 1991), wereviewed only reports involving midlife and late-life cohorts ofat least 100 women or clinical trials involving at least 100postmenopausal women. We pursued a quantitative synthe-

sis where published data rendered this approach feasible.

Hi1tswdftwech1spo(

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138 131

Studies for review were identified by one author from asearch of the MEDLINE computerized database and biblio-graphic reviews of relevant articles, and one author indepen-dently verified eligibility of identified studies.

We considered midlife women separately from late-lifewomen. Midlife represents the interval between the onsetof the menopause transition (Soules et al., 2001) and thebeginning of old age, commonly taken to be 65 years.Menopause before age 40 years is viewed as premature,(Luborsky et al., 2003), and we chose this age as the lowerboundary for mean ages of study participants. Late-lifewas defined by a mean participant age of at least 65 years.

Endogenous exposures

For endogenous exposures, we reviewed findings fromcohorts of at least 100 women not using hormone therapyand without dementia or other identified cognitive disorder,also including an objective neuropsychological measure ofepisodic memory or executive functions. Seven observa-tional studies met our selection criteria (Table 1). Becauseof heterogeneity in study design and the manner in whichresults were reported, we were unable to pool results toevaluate associations of endogenous estrogen exposureswith cognitive outcomes. A quantitative synthesis was thusnot feasible.

Endogenous exposures in midlife women. Relationsbetween cognitive function and serum concentrations ofestrone and estradiol have been investigated in three co-horts of midlife women. These include research in thepopulation-based Melbourne Women’s Midlife Health Proj-ect (Henderson et al., 2003; Ryan et al., 2011, in press),the population-based Betula Project in Sweden (Herlitz etal., 2007) and the Study of Women’s Health Across theNation, a multi-ethnic voluntary cohort in the United States(Luetters et al., 2007).

Findings from the Melbourne cohort failed to suggeststrong associations between estrogen levels and episodicmemory or executive functions. Initial analyses of pre-menopausal and postmenopausal midlife women not usinghormone therapy showed no significant relations betweenepisodic memory (word list recall) and the concentration oftotal estradiol or the free estradiol index (an index basedon the ratio of estradiol to sex hormone binding globulin)(Henderson et al., 2003). Analyses 3 years later focusedon midlife women who had by that time undergone naturalmenopause (Ryan et al., in press). Principal componentanalysis of neuropsychological test scores suggested fourcognitive factors, including verbal episodic memory (wordlist recall tasks) and nonverbal episodic memory (facerecognition tasks). Concentrations of estrone, total estra-diol, and free (nonprotein bound) estradiol were not linkedto these memory factors in cross sectional and 2-yearlongitudinal analyses (Ryan et al., in press). In exploratoryfollow-up analyses, the estrone concentration was posi-tively related to scores on one of the executive functionsmeasures (the Tower of London test) (Ryan et al., 2011).This finding, however, may have occurred by chance, as

estrogen levels were not associated with other executive

function tests (digit span backward, Letter-Number se-quencing, Trails-B, category fluency, and Symbol DigitModalities Test) (Ryan et al., 2011, in press).

Total estradiol levels were similarly unrelated to epi-sodic memory in midlife women from Sweden (incidentalrecall of test materials (Herlitz et al., 2007)) or the UnitedStates (paragraph recall (Luetters et al., 2007)). In thesecohorts, estradiol was also unrelated to executive func-tions as assessed by digit span backward (Luetters et al.,2007), category fluency (Herlitz et al., 2007), and the Sym-bol Digit Modalities Test (Luetters et al., 2007).

Endogenous exposures in older postmenopausal women.Findings for older postmenopausal women are inconsis-tent (Table 1). A Dutch study reported a positive relationbetween serum estradiol and episodic memory (den Heijeret al., 2003), whereas two US studies suggested no asso-ciations or inverse associations between some estrogenmeasures and some aspects of executive function (Yaffeet al., 1998; Barrett-Connor and Goodman-Gruen, 1999;Laughlin et al., 2010).

In the first of these, the Rotterdam Study, cross sec-tional analyses described better episodic memory (wordlist delayed recall) for women in the two highest tertiles ofbioavailable and free estradiol compared to women in thelowest tertile; however, associations were not significantwhen estradiol was analyzed as a continuous variable (den

eijer et al., 2003). The Study of Osteoporotic Fracturesncluded two executive function measures (Yaffe et al.,998). Test scores did not differ by serum estradiol quar-ile, but women in the higher estrone quartiles had lowercores on the Digit Symbol Modalities Test compared toomen in lower quartiles. At follow-up, changes in estra-iol were unrelated to test scores, but reductions in per-ormance on the Trails-B test were larger for women in thewo higher estrone quartiles (Yaffe et al., 1998). Amongomen in the Rancho Bernardo retirement community,strone and estradiol levels were generally unrelated toognitive performance 4 years after serum was obtain forormone analyses (Barrett-Connor and Goodman-Gruen,999). Follow-up after an additional 4 years (8 years aftererum was obtained) showed higher estrone levels to beositively associated with greater decline during the sec-nd 4 year interval on one of two executive function taskscategory fluency but not Trails-B) (Laughlin et al., 2010).

When analyzed categorically, women in the two highesttertiles of estrone and bioavailable estradiol (bioavailableestradiol is not bound to sex hormone binding globulin)showed higher odds of being in the lowest tertile of fluencydecline compared to women in the lowest estrogen tertiles(for each, the odds ratio was 1.8 and 95% confidenceintervals were 1.0–3.1) (Laughlin et al., 2010). Again, therewere no significant associations with scores on the Trails-Btest.

Exogenous exposures

For exogenous exposures, we selected randomized, pla-cebo-controlled, double-blind clinical trials of systemic es-

trogen-containing hormone therapy involving at least 100

Table 1. Associations between serum estrogens and tests of episodic memory or executive functionsa

Study Design Type ofmenopauseb

Number ofwomen

Mean ageor range

Estrogen Episodicmemory

Executive functions

Workingmemory

Trails-B Fluency Other

Midlife cohort (mean age�64 y)Henderson et al.

(2003)C Natural 326 57 y E1; total E2 and

free E2 indexNS — — — —

Herlitz et al. (2007) C Both 242 45–55 y Total E2 NS — — NS —Luetters et al. (2007) C Natural 1657 50 y Total E2 NS NS — — NSRyan et al. (2011, in

press)C, L Natural 148 60 y E1; total and free

E2NS NS NS NS NS; E1�c

Late life cohort (mean age�65 y)Yaffe et al. (1998) C, L Both 532 72 y E1; total E2 — — NS; E1�d — NS; E1�d

Barrett-Connor andGoodman-Gruen(1999)

L Both (presumed) 393 74 y E1; total andbioavailable E2

NS NS NS NS —

den Heijer et al.(2003)

C Both 210 70 y Total, bioavailableand free E2

NS; E2�e — — — —

Laughlin et al.(2010)f

L Both 343 69 y E1; total andbioavailable E2

— — NS NS; E1�, E2�f —

C, cross sectional; E1, estrone; E2, estradiol; L, longitudinal; NS, nonsignificant probability P�0.05; Trails-B, Trail Making Test, Part B. Significant associations between estrogen level and cognitivetest are represented by � (positive association) or � (negative association).a Studies of midlife and late-life women without dementia or identified cognitive disorder, not using hormone therapy, sample size of at least 100 women, and objective measures of episodic memoryor executive functions. Episodic memory tasks were word-list recall of unrelated (Barrett-Connor and Goodman-Gruen, 1999; Henderson et al., 2003; Herlitz et al., 2007; Laughlin et al., 2010; Ryanet al., in press) or related (Ryan et al., in press) words, paragraph recall (Luetters et al., 2007), incidental recall of verbal information and study tasks (Herlitz et al., 2007), figure recall (Barrett-Connorand Goodman-Gruen, 1999), and face recognition (Herlitz et al., 2007; Ryan et al., in press). Working memory tasks were digit span backwards (Luetters et al., 2007; Ryan et al., 2011),Letter-Number Sequencing (Ryan et al., 2011), and serial subtraction of sevens and spelling WORLD backward (Barrett-Connor and Goodman-Gruen, 1999). Other executive function tests werethe Tower of London (Ryan et al., 2011) and digit-symbol or symbol-digit substitution tests (Yaffe et al., 1998; Luetters et al., 2007; Ryan et al., 2011).b Surgical menopause was defined by hysterectomy status.c Postmenopausal women only; includes women reported in (Henderson et al., 2003). Estrone was positively associated with Tower of London performance (P�0.02) on cross sectional analyses;most executive function associations were nonsignificant.d Women in the two higher estrone quartiles had worse Digit Symbol Modalities Test scores compared to women in two lower quartiles in cross sectional analyses (analysis of covariance P�0.004).Women in the two higher estrone quartiles showed greater declines on the Trail Making Test compared to women in the two lower quartiles (analysis of covariance P�0.01). Other executive functionassociations were not significant.e For bioavailable and free estradiol, but not total estradiol, recall scores were significantly higher for women in the two higher estradiol tertiles than in the lowest tertile (each P�0.05). Other episodicmemory associations were not significant.f Same women reported in (Barrett-Connor and Goodman-Gruen, 1999). Higher baseline estrone was a significant predictor of greater decline between 4 and 8 y later on category fluencyperformance (P�0.04). Women in the highest tertile of estrone (P�0.04) and bioavailable estradiol (P�0.04) had a significantly greater odds of decline in category fluency compared to women inthe lowest tertile. Other executive function associations were not significant.

V.

W.

Henderson

andR

.A

.P

opat/

Neuroscience

191(2011)

129–138

132

T

T

M

dm2rst

(K

(ures were0.01).

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138 133

women without dementia or other identified cognitive dis-order, of at least 4 weeks treatment duration and includingan objective neuropsychological measure of episodicmemory or executive functions. Findings in these trials aremore robust for late-life women, where samples sizestended to be larger and treatment durations longer, thanfor midlife women. Eight clinical trials met selection forreview (Table 2).

We were able to synthesize results from these studiesusing meta-analytic methods (Table 3). Different neuro-psychological tests were used to assess the cognitivedomains of interest. To reduce heterogeneity, we selectedtests that were most frequently used across the eight trials.For episodic memory, we analyzed word-list learning (de-layed recall), the most commonly reported test within thisdomain (seven trials). For working memory, which repre-sents one aspect of executive functioning, the most com-monly reported test was digit span backward (three trials).Two other executive function tasks were used in at leasttwo eligible studies, Trails-B (two trials) and verbal fluency(category fluency, letter fluency, or both; eight trials). Theprimary outcome in each of these task-specific analyseswas change in memory or executive function score be-tween baseline and final assessment. The effect size ofinterest was the standardized mean difference (SMD) in

Table 2. Hormone therapy and episodic memory or executive functio

Study Type ofmenopauseb

Numberof women

Meanage

rials involving predominately midlife womenMaki et al. (2007) Natural 180 52 yKocoska-Maras et al. (2011) Natural 133d 59 y

rials involving predominately late life womenGrady et al. (2002) Both 1063 67 yViscoli et al. (2005) Both 461 70 yAlmeida et al. (2006) Surgical 115 74 yResnick et al. (2006) Natural 1416 71 yYaffe et al. (2006) Natural 417 67 yResnick et al. (2009) Surgical 886 74 y

CEE, conjugated equine estrogens; E2, estradiol; MPA, medroxyaking Test, Part B. Significant between-group comparisons are repr

a Randomized placebo-controlled trials involving at least 100 midlifeuration of at least 4 wk, and objective measures of episodic memory oemory tasks were word-list recall of unrelated (Yaffe et al., 2006; Ko009; Maki et al., 2007) words, paragraph recall (Yaffe et al., 2006; Maecall (Resnick et al., 2006, 2009; Yaffe et al., 2006; Maki et al., 2007pan backwards (Resnick et al., 2006, 2009; Maki et al., 2007), and Dhan the Trail Making Test and verbal fluency were reported in these

b Surgical menopause was defined by hysterectomy status.c Active treatment was with oral conjugated equine estrogens 0.625 mResnick et al., 2009) medroxyprogesterone acetate 2.5 mg/d, oraocoska-Maras et al., 2011), or very low dose transdermal estradiol 0

d For this three arm trial; the sample size represents women in the ese Category fluency was better in the placebo group (P�0.02).f Face recognition (immediate recall) was significantly better in the estdelayed recall) or immediate and delayed recall of unrelated words.

g Based on annual rates of change, three of four word list recall measwas significantly better in the conjugated equine estrogens group (P�

the change scores between treatment and placebo groups

in these trials. Because studies differed on the interven-tions used, duration of treatment, and participant charac-teristics, we calculated a pooled, or summary, SMD usinga random-effects model. Statistical heterogeneity for eachsummary SMD was assessed with the Q statistic (Q sta-tistic with P�0.05 is considered heterogeneous) and the I2

index. This statistic describes the proportion of total vari-ability between studies due to heterogeneity beyondchance; I2 values greater than 50% are considered heter-ogeneous (Higgins and Thompson, 2002; Borenstein etal., 2009). We do not report meta-analytic results sepa-rately for midlife and late-life women, as only two trialsinvolved midlife women. However, we conducted a sensi-tivity analysis by excluding the studies of midlife women toensure the robustness of our pooled estimates. Analyseswere performed with Comprehensive Meta-Analysis ver-sion 2 software (Biostat, Englewood, NJ, USA).

Exogenous exposures in midlife women. For midlifewomen, only two trials meeting our criteria have beenreported. Both involved naturally menopausal women. Re-sults from each suggest no significant treatment effects onmeasures of episodic memory or executive functions (Ta-ble 2). In the first of these, women 45–55 years of age withcognitive complaints were randomized for 4 months to

en without dementia: randomized placebo-controlled trialsa

tcDuration Episodic memory Executive functions

Workingmemory

Trails-B Fluency

A 4 mon NS NS — NS4 wk NS — — NS

A 4 y NS — NS CEE–e

3 y NS NS — NS20 wk NS; E2�f — — NS

A 4 y NS; CEE�; CEE�g — — NS2 y NS — NS NS2.7 y NS — — NS

rone acetate; NS, non-significant probability P�0.05; Trails-B, Trailby � (favoring active treatment) or � (favoring placebo treatment).

women without dementia or identified cognitive disorder, treatmente functions. All identified trials used a parallel groups design. Episodicras et al., 2011) or related (Almeida et al., 2006; Resnick et al., 2006,007), incidental recall of naming test items (Viscoli et al., 2005), figurerecognition (Almeida et al., 2006). Working memory tasks were digit

al Recognition (Viscoli et al., 2005). No executive function tests other

(Grady et al., 2002; Resnick et al., 2006; Maki et al., 2007) or withoutol 1 mg/d (Viscoli et al., 2005) or 2 mg/d (Almeida et al., 2006;/d (Yaffe et al., 2006).nd placebo arms.

up (P�0.02), but differences were not significant on face recognition

significantly better in the placebo group (each P�0.02); figure recall

ns in wom

Activetreatmen

CEE�MPE2

CEE�MPE2E2CEE�MPE2CEE

progesteesentedor late-lifer executivcoska-Maki et al., 2), or faceisk Spatistudies.

g/d withl estradi.014 mgtradiol a

radiol gro

active treatment with conjugated equine estrogens com-

wtesdin(ph(h(wI

eahr0S9

S %.

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138134

bined with medroxyprogesterone acetate or to treatmentwith placebo (Maki et al., 2007). Active treatment in thesecond study was with oral estradiol; eligible women wereaged 50–65 years, and outcomes were assessed at 4weeks (Kocoska-Maras et al., 2011).

Exogenous exposures in older postmenopausalomen. As with younger women, clinical trial results in

his age older group provide no evidence for importantffects of estrogen-containing hormone therapy on epi-odic memory or executive functions (Table 2). Treatmenturations ranged from 4 weeks to 4 years. These trials

ncluded women with pre-existing vascular disease (coro-ary heart disease (Grady et al., 2002) or ischemic strokeViscoli et al., 2005)), relatively healthy surgically meno-ausal women (surgical menopause here being defined byysterectomy rather than bilateral oophorectomy)Almeida et al., 2006; Resnick et al., 2009), and relativelyealthy women who had undergone natural menopauseResnick et al., 2006; Yaffe et al., 2006). Two of these trialsere ancillary studies derived from the Women’s Health

nitiative (Resnick et al., 2006, 2009).

Quantitative synthesis of clinical trial findings. Pooledffect sizes for standardized mean differences were smallnd did not provide evidence of a significant effect oformone therapy on episodic memory (word list delayedecall SMD��0.05, 95% confidence interval [CI] �0.13,.03) or on executive functions (digit span backwardMD��0.04, 95% CI �0.12, 0.05; Trails-B SMD��0.02,5% CI �0.21, 1.66; verbal fluency SMD��0.06, 95% CI

�0.13, 0.02) (Table 3). Sensitivity analyses in which find-ings from the two midlife trials were excluded did notsubstantially changes these results, implying that effects ofhormone therapy on these particular tasks were similar in

Table 3. Meta-analysis of larger randomized placebo-controlled trialswithout dementiaa

Cognitive domain(neuropsychological task)

Studies included, (number)

Episodic memory (word listdelayed recall)

Grady et al. (2002); Almeida et al. (2(2005); Resnick et al. (2006, 2009(2006); Maki et al. (2007) (seven

Executive function, workingmemory (digit spanbackwards)

Resnick et al. (2006); Maki et al. (20(2009) (three studies)

Executive function, responsealternation (Trail MakingTest, Part B)

Grady et al. (2002); Yaffe et al. (200

Executive function, initiationand self-regulation (verbalfluency; category or letter)

Grady et al. (2002); Almeida et al. (2(2005); Resnick et al. (2006); YafMaki et al. (2007); Resnick et al.Maras et al. (2011) (eight studies)

For letter fluency only, including Maki et al. (2007); Resnick et al. (2�0.14, 0.04), P�0.26; Q�3.1, P�0.37; I2�4%. CI, confidence intera See text for selection criteria.b Only two studies are included; the I2 value suggests heterogeneity,c For Resnick et al. (2006, 2009), analyses assume correlations betwd For category fluency only, including Grady et al. (2002); Viscoli e

MD��0.06 (95% CI �0.14, 0.03), P�0.20; Q�6.6, P�0.16; I2�40

midlife and late-life women.

UNRESOLVED ISSUES

Some populations of women and some cognitive out-comes are poorly addressed in observational and clinicaltrial research described above. Unresolved issues involvewomen with induced menopause (especially menopauseinduced by bilateral oophorectomy), effects of hormoneexposures during the menopause transition, and long-termcognitive consequences of midlife estrogen exposures(Table 4).

one therapy and episodic memory or executive functions in women

Random-effects modelpooled SMD (95% CI),probability

Q statistic,probability

I2

index

scoli et al.et al.

�0.05 (�0.13, 0.03), P�0.20 8.5, P�0.20 30%

nick et al. �0.04 (�0.12, 0.05), P�0.40 1.5, P�0.47 0%

studies) �0.02 (�0.21, 0.17), P�0.83 2.6, P�0.11 62%b

scoli et al.2006);ocoska-

�0.06 (�0.13, 0.02), P�0.13 8.8, P�0.27 20%

9); Kocoska-Maras et al. (2011) (four studies): SMD��0.05 (95% CI, standardized mean difference.

ugh the Q statistic value does not differ significantly from chance.gory and letter fluencies of 0.5.04); Resnick et al. (2006, 2009); Yaffe et al. (2006) (five studies):

Table 4. Cognitive effects of midlife and late-life estrogen exposures:barriers to understanding

Endogenous exposures and memory or executive functionoutcomes

Human research is observationalRestricted range of endogenous estrogen levelsRelatively few studies, especially longitudinal studiesImprecise estimates of long-term exposuresExecutive functions difficult to assess with single testsUnknown role of neurosteroids

Exogenous estrogens and memory or executive function outcomesSmall number of clinical trials in midlife womenExecutive functions difficult to assess with single testsComplexity due to different hormone interventions

Different estrogens, doses and formulationsUse of a progestagenDifferent progestagens, doses and formulationsDifferent administration schedules (e.g. sequential or

continuous combined)Issues and populations poorly addressed by existing data

Surgical menopauseMenopause transition

of horm

006); Vi); Yaffe

studies)07); Res

6); (two

006); Vife et al. ((2009); Kc,d

006, 200val; SMD

even thoeen catet al. (20

Late-life consequences of midlife exposures

rsp

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138 135

Surgical menopause

In the United States, 28% of women undergo hysterectomyby age 54 years (Merrill, 2008), and somewhat over half ofthese women undergo bilateral oophorectomy at the timeof hysterectomy (Whiteman et al., 2008). Surgically meno-pausal women are poorly represented in research summa-rized in Tables 1–3. Some of these studies involved natu-ally menopausal women; others co-mingled women withurgical and natural menopause, or defined surgical meno-ause by the presence or absence of a uterus.

Distinctions between natural and surgical menopauseinclude acuity (surgical menopause begins abruptly afterbilateral oophorectomy), age (by definition, surgical meno-pause is performed during a woman’s reproductive years,before the time that natural menopause would have other-wise occurred), health and life-style factors (women in theWomen’s Health Initiative who had undergone hysterec-tomy were less physically active and more likely to beobese than women who had experienced natural meno-pause (Stefanick et al., 2003)), and hormonal effects(midlife testosterone levels decline after surgical meno-pause but not after natural menopause (Davison et al.,2005)). In a rodent model, cognitive effects of an estrogendiffer between surgical and transitional menopause(Acosta et al., 2010).

No large clinical trials of hormone therapy have beenconducted among women with surgical menopausal de-fined by bilateral oophorectomy. Significant benefit for ver-bal episodic memory (paragraph recall (Sherwin, 1988;Phillips and Sherwin, 1992), verbal paired associate learn-ing (Phillips and Sherwin, 1992)) but not nonverbal epi-sodic memory (figure recall (Phillips and Sherwin, 1992)) isreported from two small trials in which estradiol treatmentbegan immediately after surgery. Executive functions werenot assessed. Surgical menopause was linked to an ele-vated risk of late-life cognitive impairment in one observa-tional study (Rocca et al., 2007) but not in another (Kritz-Silverstein and Barrett-Connor, 2002).

Exogenous estrogen exposures during themenopause transition

Many women who initiate hormone therapy for vasomotorsymptoms do so during the menopause transition prior to afinal menstrual period (Brett and Chong, 2001). Thesewomen are generally not considered for clinical trials withcognitive outcomes, which have focused exclusively onpostmenpausal women. Exploratory analyses in the Studyof Women’s Health Across the Nation and the MelbourneWomen’s Midlife Health Project imply that hormone ther-apy initiated prior to natural menopause may benefit epi-sodic memory (Henderson et al., 2003; Greendale et al.,2009; Maki et al., 2011) and executive functions (Green-dale et al., 2009), although benefit may not be sustainedafter menopause (Greendale et al., 2009).

Cognitive outcomes in late-life

Late-life consequences of midlife estrogen exposures are

of major concern, particularly outcomes pertaining to Alz-

heimer’s disease and other forms of dementia. The criticalwindow hypothesis suggests that higher estrogen expo-sures at a younger age, closer to the time of menopause,reduce the risk of Alzheimer’s disease, but exposures laterin life do not (Resnick and Henderson, 2002; Sherwin,2009); other cognitive outcomes might be affected as well(MacLennan et al., 2006). In ancillary studies from theWomen’s Health Initiative involving women aged 65–79years, randomized allocation to conjugated equine estro-gens plus medroxyprogesterone acetate increased de-mentia risk in women with a uterus (hazard ratio 2.1, 95%CI 1.2, 3.5) (Shumaker et al., 2003); there was a similartrend after allocation to conjugated equine estrogens alonein women without a uterus (hazard ratio 1.5, 95% CI 0.8,2.7) (Shumaker et al., 2004). To the contrary, observa-tional results generally suggest that early hormone therapyexposures reduce Alzheimer’s disease risk (Zandi et al.,2002; Henderson et al., 2005; Whitmer et al., 2011) butthese associations may be skewed by selection bias(healthy user bias) or other biases (reviewed in (Hender-son, 2006)). The lack of heterogeneity for cognitive out-comes among clinical trials involving midlife and late-lifewomen (Table 3) fails to support the critical window hy-pothesis, but the number of trials is small, limiting power toobserve an effect, and the range of cognitive outcomes islimited. The very long interval between exposure and out-come indicates that it will be very difficult to assess cog-nitive effects of midlife hormone therapy on such late-lifeoutcomes as episodic memory, executive functions, orAlzheimer’s disease incidence.

CONCLUSIONS

Observational research on endogenous estrogen expo-sures and clinical trial research on exogenous exposuressuggest two general conclusions regarding cognitiveoutcomes.

First, there are no convincing associations betweenendogenous serum estrogen concentrations and episodicmemory or executive functions in naturally menopausalmidlife women or in older postmenopausal women (Table1). Hints in older women that endogenous estrogen levelshave an inverse association with cognition (e.g., Laughlinet al., 2010) are contradicted by other findings and requireconfirmation. Second, randomized clinical trials in midlifeand late-life women without dementia do not show a sig-nificant impact of exogenous estrogens—beneficial orharmful—on neuropsychological measures of episodicmemory or executive functions over time frames rangingfrom 4 weeks to 4 years (Table 2). A quantitative synthesisof trial results similarly failed to discern significant effectswithin these cognitive domains (Table 3). This overall con-clusion of no effect may be reassuring to women con-cerned by potential adverse cognitive effects of meno-pause but disappointing to women seeking cognitive aug-mentation from hormone therapy. Findings support currentguidelines, which do not recommend hormone therapy to

improve cognition or prevent dementia (Board of the Inter-

ant2otutaemp

mtgcoitspeet

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138136

national Menopause Society et al., 2007; North AmericanMenopause Society, 2010).

There are limitations to even these modest conclusions(Table 4). For endogenous exposures, the number of stud-ies is relatively small; study designs are often cross sec-tional or hormone concentrations are often assessed at asingle point in time; and neuropsychological tests in thesestudies may be insufficiently sensitive to episodic memoryand executive functions skills. Indeed, performances onexecutive function tasks are not necessarily strongly cor-related (Kafer and Hunter, 1997; Miyake et al., 2000; Ryanet al., 2011), and studies incorporating only one or twoexecutive function tests may fail to capture important as-pects of what is understood by this term. For episodicmemory, even a widely validated task such as word listdelayed recall may be insensitive to aspects of this domain(Bird and Burgess, 2008), and delayed recall performancemay be influenced by attentional, language and planningdisturbances, in addition to memory impairment. More-over, because most women in these studies are postmen-pausal, conclusions pertain only to the range of relativelylow levels of estrone and estradiol typical of the postmeno-pausal state. Another caveat is that hormone levels at onepoint in time provide an imprecise estimate of long-term,cumulative hormone exposures. Among women with os-teoporosis, bone mineral density, a marker of cumulativeestrogen exposures, is positively associated with bettercognitive outcomes (Yaffe et al., 1999).

For clinical trial results, there are similar limitations withrespect to number of studies and selection of neuropsy-chological instruments. Interpretations are potentially lim-ited as well by use in different trials of different formulationsof different estrogens at differing doses, administered ondiffering schedules for differing durations. A progestagen,used in some trials but not others, has the potential tomodify cognitive effects of an estrogen in women (Rice etal., 2000) and laboratory animals (Lowry et al., 2010). Formidlife women in particular, trial durations have thus farbeen relatively short. Results anticipated from two larger,longer duration hormone therapy trials may provide addi-tional insights on estrogen effects in midlife women. Theseare the Early versus Late Intervention Trial with Estrogen(ELITE, clinicaltrials.gov identifier NCT00114517) and theKronos Early Estrogen Prevention Study (KEEPS, clinical-trials.gov identifier NCT00623311).

In understanding these largely negative observationalnd clinical trial results, it may be important that estradiol isot only an ovarian steroid, it is also a neurosteroid syn-hesized within the central nervous system (Mukai et al.,010; Srivastava et al., 2010). The widespread distributionf aromatase in neurons in the hippocampus and neocor-ex supports the view that local estrogen synthesis contrib-tes to synaptic plasticity (Yague et al., 2008). At leastheoretically, brain-synthesized estradiol may be at leasts important in modulating cognition as ovarian derivedndogenous estrogens or exogenous estrogens in hor-one therapy. Circulating levels may therefore be weak

redictors of neural effects that modulate cognition.

There remains a pressing need for preclinical and hu-an studies on the relationship between the menopause

ransition and midlife exposures to estrogens, progesta-ens and related compounds, and the risks of age-relatedognitive disorders (Asthana et al., 2009). Future studiesf estrogen exposures and cognition might consider how to

ncorporate these findings into the design and interpreta-ion of observational and experimental research. Under-tudied populations include women with induced meno-ause and women in the menopause transition. Potentialffects of midlife hormone exposures on Alzheimer’s dis-ase risk and other late-life cognitive outcomes merit fur-her study.

Note added in proof: Two other large studies in late-life womendescribe associations between serum estrogens and tests of ep-isodic memory or executive functions. Almeida et al. (2005) re-ported no significant associations between total estradiol or es-trone and episodic memory (word list recall) or executive functions(fluency). Yaffe et al. (2007) reported no significant associationbetween tertiles of bioavailable estradiol and episodic memory(word list recall), but women in the highest tertile were significantlyless likely to show decline two years later than women in thelowest tertile.

REFERENCES

Acosta JI, Mayer LP, Braden BB, Nonnenmacher S, Mennenga SE,Bimonte-Nelson HA (2010) The cognitive effects of conjugatedequine estrogens depend on whether menopause etiology is tran-sitional or surgical. Endocrinology 151:3795–3804.

Albert MS, Moss MB, Tanzi R, Jones K (2001) Preclinical prediction ofAD using neuropsychological tests. J Int Neuropsychol Soc7:631–639.

Almeida OP, Lautenschlager N, Vasikaram S, Leedman P, Flicker L(2005) Association between physiological serum concentration ofestrogen and the mental health of community-dwelling postmeno-pausal women age 70 years and over. Am J Geriatr Psychiatry13:142–149.

Almeida OP, Lautenschlager NT, Vasikaran S, Leedman P, Gelavis A,Flicker L (2006) A 20-week randomized controlled trial of estradiolreplacement therapy for women aged 70 years and older: effect onmood, cognition and quality of life. Neurobiol Aging 27:141–149.

Asthana S, Brinton RD, Henderson VW, McEwen BS, Morrison JH,Schmidt PJ, Frontiers Proposal for Estrogen and Cognitive AgingWork Groups (2009) Frontiers proposal. National Institute of Aging“bench to bedside: estrogen as a case study.” Age (Dordr)31:199–210.

Barrett-Connor E, Goodman-Gruen D (1999) Cognitive function andendogenous sex hormones in older women. J Am Geriatr Soc47:1289–1293.

Berent-Spillson A, Persad CC, Love T, Tkaczyk A, Wang H, ReameNK, Frey KA, Zubieta JK, Smith YR (2010) Early menopausalhormone use influences brain regions used for visual workingmemory. Menopause 17:692–699.

Berman KF, Schmidt PJ, Rubinow DR, Danaceau MA, Van Horn JD,Esposito G, Ostrem JL, Weinberger DR (1997) Modulation ofcognition-specific cortical activity by gonadal steroids: a positron-emission tomography study in women. Proc Natl Acad Sci U S A94:8836–8841.

Bird CM, Burgess N (2008) The hippocampus supports recognitionmemory for familiar words but not unfamiliar faces. Curr Biol18:1932–1936.

Board of the International Menopause Society, Pines A, Sturdee DW,Birkhäuser MH, Schneider HP, Gambacciani M, Panay N (2007)IMS updated recommendations on postmenopausal hormone ther-

apy. Climacteric 10:181–194.

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138 137

Bohacek J, Daniel JM (2010) The beneficial effects of estradiol onattentional processes are dependent on timing of treatment initia-tion following ovariectomy in middle-aged rats. Psychoneuroendo-crinology 2010:694–705.

Borenstein M, Hedges LV, Higgins JPT, Rothstein HR (2009) Intro-duction to meta-analysis. West Sussex, UK: John Wiley & Sons.

Brett KM, Chong Y (2001) Hormone replacement therapy: knowledgeand use in the United States. Hyattsville, MD: National Center forHealth Statistics.

Burger HG, Hale GE, Robertson DM, Dennerstein L (2007) A review ofhormonal changes during the menopausal transition: focus onfindings from the Melbourne Women’s Midlife Health Project. HumReprod Update 13:559–565.

Cummings JL, Miller BL (2007) Conceptual and clinical aspects of thefrontal lobes. In: The human frontal lobes: functions and disorders,2nd ed (Miller BL, Cummings JL, eds), pp 12–21. New York:Guilford Press.

Daniel JM, Hulst JL, Berbling JL (2006) Estradiol replacement en-hances working memory in middle-aged rats when initiated imme-diately after ovariectomy but not after a long-term period of ovarianhormone deprivation. Endocrinology 147:607–614.

Davison S, Bell R, Donath S, Montalto J, Davis SR (2005) Androgenlevels in adult females: changes with age, menopause, and oo-phorectomy. J Clin Endocrinol Metab 90:3847–3853.

Deary IJ, Corley J, Gow AJ, Harris SE, Houlihan LM, Marioni RE,Penke L, Rafnsson SB, Starr JM (2009) Age-associated cognitivedecline. Br Med Bull 92:135–152.

den Heijer T, Geerlings MI, Hofman A, de Jong FH, Launer LJ, PolsHAP, Breteler MMB (2003) Higher estrogen levels are not associ-ated with larger hippocampi and better memory performance. ArchNeurol 60:213–220.

Easterbrook PJ, Berlin JA, Gopalan R, Matthews DR (1991) Publica-tion bias in clinical research. Lancet 337:867–872.

Foy MR (2011) Ovarian hormones, aging and stress on hippocampalsynaptic plasticity. Neurobiol Learn Mem 95:134–144.

Grady D, Yaffe K, Kristof M, Lin F, Richards C, Barrett-Connor E(2002) Effect of postmenopausal hormone therapy on cognitivefunction: the Heart and Estrogen/Progestin Replacement Study.Am J Med 113:543–548.

Greendale GA, Huang M-H, Wight RG, Seeman T, Luetters C, AvisNE, Johnston J, Karlamangla AS (2009) Effects of the menopausetransition and hormone use on cognitive performance in midlifewomen. Neurology 72:1850–1857.

Grober E, Hall CB, Lipton RB, Zonderman AB, Resnick SM, Kawas C(2008) Memory impairment, executive dysfunction, and intellectualdecline in preclinical Alzheimer’s disease. J Int Neuropsychol Soc14:266–278.

Hao J, Rapp PR, Janssen WG, Lou W, Lasley BL, Hof PR, MorrisonJH (2007) Interactive effects of age and estrogen on cognition andpyramidal neurons in monkey prefrontal cortex. Proc Natl Acad SciU S A 104:11465–11470.

Hao J, Rapp PR, Leffler AE, Leffler SR, Janssen WG, Lou W, McKayH, Roberts JA, Wearne SL, Hof PR, Morrison JH (2006) Estrogenalters spine number and morphology in prefrontal cortex of agedfemale rhesus monkeys. J Neurosci 26:2571–2578.

Henderson VW (2006) Estrogen-containing hormone therapy and Alz-heimer’s disease risk: understanding discrepant inferences fromobservational and experimental research. Neuroscience 138:1031–1039.

Henderson VW, Benke KS, Green RC, Cupples LA, Farrer LA (2005)Postmenopausal hormone therapy and Alzheimer’s disease risk:interaction with age. J Neurol Neurosurg Psychiatry 76:103–105.

Henderson VW, Dudley EC, Guthrie JR, Burger HG, Dennerstein L(2003) Estrogen exposures and memory at midlife: a population-based study of women. Neurology 60:1369–1371.

Herlitz A, Thilers P, Habib R (2007) Endogenous estrogen is notassociated with cognitive performance before, during, or after

menopause. Menopause 14:425–431.

Higgins JP, Thompson SG (2002) Quantifying heterogeneity in ameta-analysis. Stat Med 21:1539–1558.

Hogervorst E, Henderson VW, Gibbs RB, Brinton RD, eds (2009)Hormones, cognition and dementia: state of the art and emergenttherapeutic strategies. New York: Cambridge University Press.

Joffe H, Hall JE, Gruber S, Sarmiento IA, Cohen LS, Yurgelun-Todd D,Martin KA (2006) Estrogen therapy selectively enhances prefrontalcognitive processes: a randomized, double-blind, placebo-con-trolled study with functional magnetic resonance imaging in peri-menopausal and recently postmenopausal women. Menopause13:411–422.

Kafer KL, Hunter M (1997) On testing the face validity of planning/problem-solving tasks in a normal population. J Int NeuropsycholSoc 3:108–119.

Kocoska-Maras L, Zethraeus N, Rådestad AF, Ellingsen T, vonSchoultz B, Johannesson M, Hirschberg AL (2011) A randomizedtrial of the effect of testosterone and estrogen on verbal fluency,verbal memory, and spatial ability in healthy postmenopausalwomen. Fertil Steril 95:152–157.

Kritz-Silverstein D, Barrett-Connor E (2002) Hysterectomy, oophorec-tomy, and cognitive function in older women. J Am Geriatr Soc50:55–61.

Laughlin GA, Kritz-Silverstein D, Barrett-Connor E (2010) Endogenousoestrogens predict 4-year decline in verbal fluency in postmeno-pausal women: the Rancho Bernardo Study. Clin Endocrinol (Oxf)72:99–106.

Lezak MD, Howieson DB, Loring DW (2004) Neuropsychological as-sessment, 4th ed. New York: Oxford University Press.

Liu F, Day M, Muñiz LC, Bitran D, Arias R, Revilla-Sanchez R, GrauerS, Zhang G, Kelley C, Pulito V, Sung A, Mervis RF, Navarra R,Hirst WD, Reinhart PH, Marquis KL, Moss SJ, Pangalos MN,Brandon NJ (2008) Activation of estrogen receptor-beta regulateshippocampal synaptic plasticity and improves memory. Nat Neu-rosci 11:334–343.

Lowry NC, Pardon LP, Yates MA, Juraska JM (2010) Effects of long-term treatment with 17 beta-estradiol and medroxyprogesteroneacetate on water maze performance in middle aged female rats.Horm Behav 58:200–207.

Luborsky JL, Meyer P, Sowers MF, Gold EB, Santoro N (2003) Pre-mature menopause in a multi-ethnic population study of the meno-pause transition. Hum Reprod 18:199–206.

Luetters C, Huang MH, Seeman T, Buckwalter G, Meyer PM, Avis NE,Sternfeld B, Johnston JM, Greendale GA (2007) Menopause tran-sition stage and endogenous estradiol and follicle-stimulating hor-mone levels are not related to cognitive performance: cross-sec-tional results from the study of women’s health across the nation(SWAN). J Womens Health 16:331–344.

MacLennan AH, Henderson VW, Paine BJ, Mathias J, Ramsay EN,Ryan P, Stocks NP, Taylor AW (2006) Hormone therapy, timing ofinitiation, and cognition in women aged older than 60 years: theREMEMBER pilot study. Menopause 13:28–36.

Maki PM, Dennerstein L, Clark M, Guthrie J, Lamontagne P, FornelliD, Little D, Henderson VW, Resnick SM (2011) Perimenopausaluse of hormone therapy is associated with enhanced memory andhippocampal function later in life. Brain Res 1379:232–243.

Maki PM, Gast MJ, Vieweg A, Burriss SW, Yaffe K (2007) Hormonetherapy in menopausal women with cognitive complaints: a ran-domized, double-blind trial. Neurology 69:1322–1330.

Merrill RM (2008) Hysterectomy surveillance in the United States,1997 through 2005. Med Sci Monit 14:CR24–CR31.

Miyake A, Friedman NP, Emerson MJ, Witzki AH, Howerter A (2000)The unity and diversity of executive functions and their contribu-tions to complex “frontal lobe” tasks: a latent variable analysis.Cogn Psychol 41:49–100.

Mukai H, Kimoto T, Hojo Y, Kawato S, Murakami G, Higo S, HatanakaY, Ogiue-Ikeda M (2010) Modulation of synaptic plasticity by brainestrogen in the hippocampus. Biochim Biophys Acta 1800:

1030–1044.

S

T

V

W

W

W

Y

Y

Y

Y

Y

Y

Z

V. W. Henderson and R. A. Popat / Neuroscience 191 (2011) 129–138138

Mukai H, Tsurugizawa T, Ogiue-Ikeda M, Murakami G, Hojo Y, Ishii H,Kimoto T, Kawato S (2006) Local neurosteroid production in thehippocampus: influence on synaptic plasticity of memory. Neu-roendocrinology 84:255–263.

North American Menopause Society (2010) Estrogen and progesto-gen use in postmenopausal women: 2010 position statement of theNorth American Menopause Society. Menopause 17:242–255.

Phillips SM, Sherwin BB (1992) Effects of estrogen on memory func-tion in surgically menopausal women. Psychoneuroendocrinology17:485–495.

Rapp PR, Morrison JH, Roberts JA (2003) Cyclic estrogen replace-ment improves cognitive function in aged ovariectomized rhesusmonkeys. J Neurosci 23:5708–5714.

Resnick SM, Espeland MA, An Y, Maki PM, Coker LH, Jackson R,Stefanick ML, Wallace R, Rapp SR, Women’s Health InitiativeStudy of Cognitive Aging Investigators (2009) Effects of conju-gated equine estrogens on cognition and affect in postmenopausalwomen with prior hysterectomy. J Clin Endocrinol Metab94:4152–4161.

Resnick SM, Henderson VW (2002) Hormone therapy and risk ofAlzheimer disease: a critical time. JAMA 288:2170–2172.

Resnick SM, Maki PM, Rapp SR, Espeland MA, Brunner R, Coker LH,Granek IA, Hogan P, Ockene JK, Shumaker SA (2006) Effects ofcombination estrogen plus progestin hormone treatment on cogni-tion and affect. J Clin Endocrinol Metab 91:1802–1810.

Rice MM, Graves AB, McCurry SM, Gibbons LE, Bowen JD, McCor-mick WC, Larson EB (2000) Postmenopausal estrogen and estro-gen-progestin use and 2-year rate of cognitive change in a cohortof older Japanese American women: the Kame Project. Arch InternMed 160:1641–1649.

Rocca WA, Bower JH, Ahlskog JE, Grossardt BR, de Andrade M,Melton LJ 3rd (2007) Increased risk of cognitive impairment ordementia in women who underwent oophorectomy before meno-pause. Neurology 69:1074–1083.

Ryan J, Stanczky FZ, Dennerstein L, Mack WJ, Clark MS, Szoeke C,Kildea D, Henderson VW (in press) Hormone levels and cognitivefunction in postmenopausal midlife women. Neurobiol Aging, inpress.

Ryan J, Stanczyk FZ, Dennerstein L, Mack WJ, Clark MS, Szoeke C,Henderson VW (2011) Executive functions in recently postmeno-pausal women: absence of strong association with serum gonadalsteroids. Brain Res 1379:199–205.

Salthouse TA (2009) When does age-related cognitive decline begin?Neurobiol Aging 30:507–514.

Sherwin BB (1988) Estrogen and/or androgen replacement therapyand cognitive functioning in surgically menopausal women. Psy-choneuroendocrinology 13:345–357.

Sherwin BB (2009) Estrogen therapy: is time of initiation critical forneuroprotection? Nat Rev Endocrinol 5:620–627.

Shumaker SA, Legault C, Kuller L, Rapp SR, Thal L, Lane DS, Fillit H,Stefanick ML, Hendrix SL, Lewis CE, Masaki K, Coker LH, Wom-en’s Health Initiative Memory Study (2004) Conjugated equineestrogens and incidence of probable dementia and mild cognitiveimpairment in postmenopausal women: Women’s Health InitiativeMemory Study. JAMA 291:2947–2958.

Shumaker SA, Legault C, Rapp SR, Thal L, Wallace RB, Ockene JK,Hendrix SL, Jones BN III, Assaf AR, Jackson RD, Morley KotchenJM, Wassertheil-Smoller S, Wactawski-Wende J, WHIMS Investi-gators (2003) Estrogen plus progestin and the incidence of demen-tia and mild cognitive impairment in postmenopausal women: theWomen’s Health Initiative Memory Study (WHIMS). JAMA

289:2651–2662.

Soules MR, Sherman S, Parrott E, Rebar R, Santoro N, Utian W,Woods N (2001) Executive summary: stages of reproductive agingworkshop (STRAW). Fertil Steril 76:874–878.

Spreen O, Strauss E, Sherman EMS (2006) A compendium of neuro-psychological tests: administration, norms, and commentary, 3rded. New York: Oxford University Press.

Squire LR (2009) Memory and brain systems: 1969–2009. J Neurosci29:12711–12716.

Srivastava DP, Woolfrey KM, Liu F, Brandon NJ, Penzes P (2010)Estrogen receptor � activity modulates synaptic signaling andstructure. J Neurosci 30:1354–1360.

tefanick ML, Cochrane BB, Hsia J, Barad DH, Liu JH, Johnson SR(2003) The Women’s Health Initiative postmenopausal hormonetrials: overview and baseline characteristics of participants. Am JEpidemiol 13 (9 Suppl):S78–S86.

révoux R, De Brux J, Castanier M, Nahoul K, Soule J-P, Scholler R(1986) Endometrium and plasma hormone profile in the peri-meno-pause and post-menopause. Maturitas 8:309–326.

iscoli CM, Brass LM, Kernan WN, Sarrel PM, Suissa S, Horwitz RI(2005) Estrogen therapy and risk of cognitive decline: results fromthe Women’s Estrogen for Stroke Trial (WEST). Am J ObstetGynecol 192:387–393.

hiteman MK, Hillis SD, Jamieson DJ, Morrow B, Podgornik MN,Brett KM, Marchbanks PA (2008) Inpatient hysterectomy sur-veillance in the United States, 2000 –2004. Am J Obstet Gyne-col 198:e1– e7.

hitmer RA, Quesenberry CP, Zhou J, Yaffe K (2011) Timing ofhormone therapy and dementia: the critical window theory revis-ited. Ann Neurol 69:163–169.

oolley CS, McEwen BS (1993) Roles of estradiol and progesteronein regulation of hippocampal dendritic spine density during theestrous cycle in the rat. J Comp Neurol 336:293–306.

affe K, Barnes D, Lindquist K, Cauley J, Simonsick EM, PenninxB, Satterfield S, Harris T, Cummings SR, Health ABC Investi-gators (2007) Endogenous sex hormone levels and risk of cog-nitive decline in an older biracial cohort. Neurobiol Aging28:171–178.

affe K, Browner W, Cauley J, Launer L, Harris T (1999) Associationbetween bone mineral density and cognitive decline in olderwomen. J Am Geriatr Soc 47:1176–1182.

affe K, Grady D, Pressman A, Cummings S (1998) Serum estrogenlevels, cognitive performance, and risk of cognitive decline in oldercommunity women. J Am Geriatr Soc 46:816–821.

affe K, Vittinghoff E, Ensrud KE, Johnson KC, Diem S, Hanes V,Grady D (2006) Effects of ultra-low-dose transdermal estradiol oncognition and health-related quality of life. Arch Neurol 63:945–950.

ague JG, Wang AC, Janssen WG, Hof PR, Garcia-Segura LM,Azcoitia I, Morrison JH (2008) Aromatase distribution in the mon-key temporal neocortex and hippocampus. Brain Res 1209:115–127.

ildirim M, Janssen WG, Tabori NE, Adams MM, Yuen GS, AkamaKT, McEwen BS, Milner TA, Morrison JH (2008) Estrogen andaging affect synaptic distribution of phosphorylated LIM kinase(pLIMK) in CA1 region of female rat hippocampus. Neuroscience152:360–370.

andi PP, Carlson MC, Plassman BL, Welsh-Bohmer KA, Mayer LS,Steffens DC, Breitner JCS (2002) Hormone replacement therapyand incidence of Alzheimer’s disease in older women: the Cache

County study. JAMA 288:2123–2129.

(Accepted 24 May 2011)(Available online 6 June 2011)