Abstract
Cardiovascular Risk factors, Aging and Dementia (CAIDE) is a Finnish population-based study. 731 cognitively normal women had self-reported hormone therapy (HT) data in 1998 as: no use, use ≤5 years, and >5 years. Information on type of HT was only available from 1995–1998 (Prescription Register). Cognition was assessed in 1998 and 2005–2008. Long-term (>5 years) HT use, especially use of estradiol alone among women having hysterectomy with bilateral oophorectomy, was associated with better episodic memory in 1998, but not in 2005–2008. Although a strong evidence for protective effect of estradiol on cognition was not observed in our study, improved global cognition among long-term users suggests that long-term postmenopausal HT may be beneficial for some cognitive domains.
INTRODUCTION
Use of hormone therapy (HT) has been related to improvement in global cognition and different cognitive domains in observational studies [1, 2], yet findings from recent clinical trials have not been convincing [3–5].
We explored the associations between estradiol-based HT and cognitive decline in different domains in a population-based prospective cohort during8-year follow-up.
METHODS
Study population
Cardiovascular Risk factors, Aging and Dementia (CAIDE) study is a longitudinal, population based study carried out in Eastern Finland. The participants were examined in midlife within the framework of the North Karelia project and the FINMONICA (Finnish Multinational Monitoring of Trends and Determinants in Cardiovascular Disease) study in 1972, 1977, 1982, or 1987. Individuals who were alive, aged 65–79 years and living in the Kuopio and Joensuu regions at the end of 1997 were invited to the first re-examination in 1998 (baseline visit for this study). A second re-examination was conducted in 2005–2008 (follow-up visit for this study). Both re-examinations included a self-administered questionnaire on sociodemographic characteristics, health-related behaviors, and medical history. Specially trained nurses ensured that questionnaires were fully completed.
A three-step protocol for dementia diagnosis was applied at both re-examinations: screening phase, clinical phase, and differential diagnostic phase. In 1998, participants with ≤24 points on the Mini-Mental State Examination (MMSE) at screening were referred to the clinical phase for further examinations. In 2005–2008, participants with ≤24 points on MMSE, or with decrease ≥3 points on MMSE since 1998, or with <70% delayed recall in the Consortium to Establish A Registry for Alzheimer’s disease (CERAD) word list, or with informant concerns regarding the participant’s cognition were referred to the clinical phase. The clinical phase involved comprehensive neurological, cardiovascular, and neuropsychological examinations. The differential diagnostic phase included brain imaging (MRI/CT), blood tests, cerebrospinal fluid (CSF) analysis if needed, and electrocardiogram. A review board consisting of the physician, neuropsychologist, and a senior neurologist ascertained the primary diagnosis based on all information.
Dementia was diagnosed using the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria.
For mild cognitive impairment (MCI), a modified version of the Mayo Clinic Alzheimer’s Disease Research Center criteria was used. This study includes women without dementia or MCI at the first re-examination in 1998, and who responded to the HT use questionnaire in 1998 (n = 731). Mean follow-up time between first and second re-examination for this CAIDE subsample was 8.3 years.
Ethical approval
The CAIDE study was approved by the local ethics committee (University of Kuopio and Kuopio University Hospital, Kuopio, Finland), and written informed consent was obtained from all participants.
Cognitive assessments
A comprehensive battery of neuropsychological tests was carried out to assess several cognitive domains at both assessments including: (1) MMSE as a measure of global cognition; (2) immediate word recall test (one word list) for episodic memory; (3) The Stroop test to assess executive functioning; (4) Verbal expression assessed by category fluency test; (5) Bimanual Purdue Pegboard Test and the letter digit substitution test, with the mean of their normalized scores as a measure of psychomotorspeed [6].
Hormone therapy use
HT use was measured with a self-administered questionnaire in 1998: “How long have you been using HT during your life? Self-reported HT use (n = 731) was classified as: never use (n = 488), use for ≤5 years (n = 116), and >5 years (n = 127). Data on type of HT was obtained from the Finnish Prescription Register with following ATC codes: G03C for estrogen and G03F (estrogen and progestogen) HT. The Prescription Register started in 1995 in Finland, and data recorded from 1995–1998 was used in the present study. Classification by HT duration was thus not possible for register data. HT type, obtained from the register, was classified as follows: no self-reported or recorded HT (n = 463), estrogen alone (n = 156), or combination HT (n = 41). As HT type was not ascertainable before 1995, women with self-reported HT but no recorded HT in the register (n = 125) were excluded from analyses focusing on HT type. Non-physiological HT is not used in Finland and thus all HT was estradiol-based.
Other assessments
History of gynecological surgery was inquired in the self-report questionnaire. Co-morbidity index was calculated from medical conditions recorded in the Finnish Hospital Discharge Register before 1998. Following conditions with corresponding scores were included: myocardial infarction, heart failure, coronary heart disease, stroke/transient ischemic attacks, diabetes, asthma or chronic obstructive pulmonary disease (score of 1); kidney failure and malignant neoplasm (score of 2) [7]. APOE genotype was assessed from blood leucocytes using polymerase chain reaction and Hhal digestion. Women were classified as APOE ɛ4 carriers and non-carriers.
Statistical analysis
Cognitive test scores were log-transformed because of skewed distribution. Continuous variables are presented as medians with 95% confidence intervals (CIs) and their range (minimum-maximum). The associations between categorical variables and categories of HT use by duration were assessed by chi-square tests and between continuous variables and HT use by linear regression.
The associations between categories of HT use by duration and cognitive tests at baseline and follow-up examination (8.3 years later) were investigated with multivariable linear regression. Model 1 was adjusted for age, education, and APOE status. Model 2 was additionally adjusted for hysterectomy and co-morbidity index. The analyses of cognitive test performance in follow-up examination were adjusted for follow-up time and cognition at baseline examination. The same models were applied in secondary analyses on register-recorded data on type of HT from 1995–1998; additional analyses were stratified by type of gynecological surgery (no surgery; hysterectomy with bilateral oophorectomy; or hysterectomy alone or with unilateral oophorectomy). The results are presented as beta coefficients with p-values. Data was analyzed using Stata 12.0 (Stata Corp LP, College Station, TX).
RESULTS
Women using estradiol-based HT for >5 years were younger, had more education, and were more likely to have undergone hysterectomy than never and ≤5 years users (Table 1). Estradiol use was not associated with APOE status and co-morbidities. Women who had used estradiol-based HT for >5 years had better scores in global cognition, episodic memory, and psychomotor speed tests at baseline than women who had used estradiol HT for less than five years or non-users. Long-term users also performed better in verbal expression test at baseline than short-term users. Association between long-term use and better global cognition and psychomotor speed remained during follow-up (Table 1).
The multivariable-adjusted associations between estradiol-based HT use and cognitive tests are shown in Table 2. A positive trend in global cognition and episodic memory among women using HT for >5 years in 1998 was observed. These associations were no longer evident in the follow-up measurements, although the point estimate for global cognition was of similar magnitude as in 1998. However, due to smaller number of participants the confidence intervals were wider and included also the null value (Table 2). In secondary analyses considering the HT type used from 1995–1998 and type of hysterectomy, better memory scores in 1998 were found in women using estradiol alone with (β, 95% confidence interval (CI) of 0.70, 0.06–1.35, p = 0.03). The association was not observed among women who used estradiol and progestogen concomitantly (0.12, –1.03–1.26, p = 0.84). When the analyses were stratified by type of hysterectomy, combination HT use was not associated with better memory scores (data not shown), but the association between estradiol and memory was observed only among women who had underwent hysterectomy with bilateral oophorectomy (1.74, 0.35–3.13, p = 0.015). These associations were no longer evident in 2005–2008.
DISCUSSION
The overall findings from our prospective cohort study do not provide strong evidence for protective effect of estradiol-based HT against cognitive decline. However better global cognition and episodic memory in 1998 among >5 year estradiol users suggest selective beneficial impact of estradiol on cognition which was particularly evident among women who had undergone radical hysterectomy with bilateral oophorectomy.
Our findings are consistent with the recent clinical trials where no overall significant effect of use of conjugated equine estrogens (CEE) based HT among 50–55-year-old young postmenopausal women, and use of neither oral CEE nor transdermal estradiol around menopause, with cognition was observed [3, 5]. While our findings are in contrast with the Women’s Health Initiative Memory Study (WHIMS), where CEE-based HT use had negative impact on verbal memory [8]. One possible explanation for difference in results can be the use of estradiol in our study in contrast to CEE-based HT in WHIMS [8] or it can possibly be due to underlying healthy user bias in our study. Long term persistent decline in cognition with use of CEE-based HT depicts harmful effects of start of HT in late postmenopause >65 years of age [9]. Strengths of our study are; population based cohort, long follow up time, relatively homogenous population, use of established neuropsychological tests to validate cognitive status, and control for various confounders. Limitations are: self-reported HT use, lack of ascertainment of age at start of HT and menopausal status at baseline, type and time of initiation of HT in relation to menopause onset. The number of combination users was small, which would decrease the power to detect an association. However, the β point estimate which is not affected by sample size, was lower in combination HT group than for women who used estrogen only.
HT has the potential for neuroprotection [10]; however, late life cognitive functioning is influenced by multiple factors acting across lifespan. Further studies are needed to underpin the differential effects of estradiol-based HT on cognition.
Footnotes
ACKNOWLEDGMENTS
Bushra Imtiaz has received funding from the Doctoral Program of Molecular Medicine, University of Eastern Finland. The authors wish to acknowledge the following funding: VPH-DARE@IT (FP7-ICT-2011-9-601055) and MIND-AD, Academy of Finland, 291803, EU Joint Programme— Neurodegenerative Disease Research (JPND) projects, Axa Research Fund, Center of Innovative Medicine (CIMED) grants as well as VTR funding from Kuopio University Hospital forMIND-AD.
