Bench to Bedside: Longitudinal and Epidemiological Studies

III. Longitudinal and Epidemiological Studies

Levels of Endogenous Sex Steroids and Risk for Brain Aging—Dr. Lenore Launer

Dr. Launer's presentation outlined the following:

1. Endogenous levels of estrogen, MRI brain changes, and risk for dementia in women (findings from the Rotterdam study)
2. Endogenous levels of estrogen, testosterone, and risk for MRI brain changes, dementia, and neuropathologic outcomes in men (findings from the Honolulu Asian Aging Study [HAAS])
3. Changes in the association of hormone levels with stage of dementia  
4. Both the HAAS and Rotterdam study results are more consistent with the trial data

Specific points included:

• Rotterdam study: (1) an increase in reproductive years (menarche–menopause) predicted a higher incidence of dementia, and (2) on MRI, as the estradiol increased, hippocampal volume decreased.
• HAAS-men
• Slight increase in Alzheimer's disease with higher estrogen levels

Estrogen Therapy and Risk of AD in the Baltimore Longitudinal Study of Aging (BLSA)—Dr. Claudia Kawas

Dr. Kawas's presentation outlined the following:

1. Key methods and results of the BLSA included: (1) prospective, observational study with long follow-up period (16 years); (2) reduced risk (approximately 50 percent if ET ever used, as compared to never used); (3) there was no effect for duration of ET use; (4) the most commonly used compounds reported in the BLSA were unopposed estrogens (in the form of Premarin); and (5) education was adjusted for, but there may be other unknown confounders.
2. Advantages and disadvantages of observational studies, case control studies, cohort studies, clinical trials, etc. 
3. Each study type provides important information, and each study type also has its disadvantages.
4. Implications of findings of the BLSA for future studies in humans.

Specific points included:

• BLSA, prospective study
  • Protective effect, ever versus never, relative risk (RR) 0.46, for protection from dementia
  • No dose effect
  • No duration effect
  • No route effect
• Leisure World
  • Approximately 14,000 women
  • HT showed a risk reduction for dementia by about one-half
  • Of 5,000 women, 2,200 were alive and older than 90 years of age
  • For mortality:
    • Approximately 6 percent reduction
    • Effect of duration (no reduction with less than 3 years of use)
    • Last use effect present to about 15 percent

A Population-Based Longitudinal Study of Cognitive Functioning in the Menopausal Transition: Study of Women's Health Across the Nation (SWAN)—Dr. Peter Meyer

Dr. Meyer's presentation highlighted the following:

1. The Chicago cohort of the SWAN study is a population-based cohort of 868 women. Of these, 802 had valid cognitive assessments at one or more of their annual interviews from baseline through follow-up year 4.  
2. Used digit span backwards and the symbol digit modality test as brief tests of working memory and perceptual speed.
3. Findings: (1) at baseline, the proportion of women reporting problems with memory during the previous 2 weeks showed a clear association with age and menopausal status; and (2) there was no indication of a decline in working memory or in perceptual speed during the 5 years of the study related to age or menopausal status.
4. Implications of findings from the SWAN study for future studies in humans.

Specific points included:

• There was no real change in digit span and symbol digit modality test during menopause.
• There was no difference in rate of change.

Estrogen Exposures in Midlife, Memory, and Dementia—Dr. Victor Henderson

Dr. Henderson's presentation outlined the following:

1. Midlife estrogen exposures do not have a substantial effect on episodic memory for women undergoing natural menopause. 
2. Early exposure to menopausal hormone therapy (HT) may be associated with Alzheimer's disease risk reduction.

Specific points included:

• Discussion about the Melbourne Women's Midlife Health Project (N=326 women; surgical menopause excluded).
• There was no difference on immediate or delayed word list task (10 items) by premenopausal, perimenopausal, or postmenopausal status.
• There was no relationship between endogenous estradiol and immediate or delayed recall.
• There was no difference in immediate or delayed recall based on HT use (never, past users, current users).
• Post hoc analyses suggested better immediate recall for current HT users who initiated HT prior to their final menstrual period when compared to those who initiated HT after their final menstrual period.
• Conclusion from Melbourne Women's Midlife Health Project: there is no evidence that menopausal status or other measures of midlife estrogen exposure affect verbal episodic memory as assessed by word list learning.
• Discussion about Multi-Institutional Research in Alzheimer Genetic Epidemiology (N=426 cases, 524 controls).
• HT was associated with a reduced risk of Alzheimer's disease.
• There was a significant interaction with age.
• In age tertiles, HT was significantly associated with reduced risk in the youngest tertile only. 
• Among other possibilities, findings raise the question of whether early HT exposure might reduce Alzheimer's disease risk.

Are Epidemiologic Studies Worthless as an Indication of Trial Results, or Does “Nixon's Law” (Timing Is Everything) Prevail?—Dr. John Breitner

Dr. Breitner's presentation outlined the following:

1. Review epidemiological literature suggesting protection with HT.
2. Discuss possible reasons for contrast with trials data (e.g., inadequate control on socioeconomic status [SES]).
3. Another explanation: timing of exposure in relation to onset of dementia, a late-stage phenomenon in the pathogenesis of Alzheimer's disease. With proper consideration of this issue, Cache County data and the WHIMS data are concordant. 

Specific points presented:

• Cache County shows a decreased risk of Alzheimer's disease with HT, which is time dependent on when hormones were used. 
  • 0–10 years—HR 2.22
  • 3–10 years—HR 0.22
  • Greater than 10 years—HR 0.17
• Suggestion: HT is protective years before dementia, but crosses over to RR of 0 at about 10 years before dementia.

Methodologic Challenges in the Nonexperimental Study of Hormones and Dementia—Dr. Diana Petitti

Dr. Petitti's presentation outlined the following:

1. Proxy reports of estrogen use are unreliable (use is underreported).
2. Women with dementia/cognitive impairment do not accurately report their own current or past exposure to estrogen (use is underreported).
3. Computer-stored prescription data can be used to mitigate these reporting biases in observational studies.
4. Empiric results from a study based on computer-stored data will be presented to illustrate studies consistent with data.

Specific points included: 

• Women (1,944) entered the study with estrogen use based on pharmacy records.
• Classification was based on a telephone interview of cognitive status and the Dementia Questionnaire.
• Mean duration of E (29 years) or E + P (21 years), for women older than 75 years of age.
• Age at start for hormone use was between 46 and 61 years of age.
• Slightly higher RR for both types of therapy for all dementia:
  • E-1.26
  • E + P-1.46

Discussion—Longitudinal and Epidemiological Studies

• Other trials that were unsuccessful based on basic or epidemiological studies included vitamin E, Beta carotene, and low fat-high fiber.
• SES must be considered in the selection of subjects.
• More observational studies are needed; case-control studies may not provide information about dementia.
• Users of alcohol (1–3 drinks per day) have better memory.
• Start ET early.
• There is a long incubation time for dementia.
• Underlying neuron health.
• The question of oral contraceptive use is a "black box:" supraphysiologic levels of E in first pills are commercially available. How can the findings of the Cache County study be reconciled with the Petitti Women's Memory Study?