University of Southern California Dissertations and Theses

Association between Blood Biomarkers of Alzheimer’s Disease and Cognition in Postmenopausal Women

(USC Thesis Other)

Association between Blood Biomarkers of Alzheimer’s Disease and Cognition in Postmenopausal Women

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2025 Tiansheng Jin
Association between Blood Biomarkers of Alzheimer’s Disease and Cognition in
Postmenopausal Women
by
Tiansheng Jin
A Thesis Presented to the
FACULTY OF THE USC KECK SCHOOL OF MEDICINE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(BIOSTATISTICS)
May 2025

ii
TABLE OF CONTENTS
List of Tables ................................................................................................................................ iii
List of Figures................................................................................................................................ iv
Abstract .......................................................................................................................................... v
Introduction .................................................................................................................................... 1
Chapter 1: Methods ........................................................................................................................ 4
Study design ............................................................................................................. 4
AD biomarkers ......................................................................................................... 5
Cognitive composite scores ..................................................................................... 5
Statistical analysis .................................................................................................... 6
Chapter 2: Results .......................................................................................................................... 9
Univariable analysis of baseline associations ........................................................ 18
Multivariable analysis of baseline associations ..................................................... 18
Longitudinal analysis ............................................................................................. 21
Chapter 3: Discussion .................................................................................................................. 27

References .................................................................................................................................... 30

iii
List of Tables
Table 1: Demographic and Clinical Characteristics of the ELITE Participants at Baseline ......... 9
Table 2: Levels of AD Biomarkers in the ELITE Participants at Baseline ................................. 11
Table 3: Cross-sectional Univariable Associations between Demographic and Clinical
Properties and Cognitive Composite Scores at Baseline ............................................................. 12
Table 4: Cross-sectional Association between AD Blood Biomarkers and Demographic
and Clinical Factors at Baseline ................................................................................................... 15
Table 5: Cross-sectional Univariable Associations between AD Biomarkers and Cognitive
Composite Scores at Baseline ...................................................................................................... 18
Table 6: Cross-sectional Multivariable Analysis of Associations between AD Biomarkers
and Cognitive Composite Scores at Baseline .............................................................................. 19
Table 7: Cross-sectional Stratified Multivariable Analysis of Associations between AD
Biomarkers and Cognitive Composite Scores at Baseline ........................................................... 20
Table 8: Longitudinal Analysis of Associations between Baseline AD Blood Biomarkers
and Longitudinal Measures of Cognition .................................................................................... 21
Table 9: Longitudinal Analysis of Associations between 2.5-year AD Biomarkers Changes
and Slopes of Global Composite Scores (Univariable and Multivariable) .................................. 22
Table 10: Longitudinal Analysis of Associations between 2.5-year AD Biomarkers Changes
and Slopes of Verbal Memory Scores (Univariable and Multivariable) ..................................... 22
Table 11: Longitudinal Analysis of Associations between 2.5-year AD Biomarkers Changes
and Slopes of Executive Function Scores (Univariable and Multivariable) ................................ 23

iv
List of Figures
Figure 1: Boxplot of ptau181 by baseline and 2.5-year follow-up .............................................. 23
Figure 2: Boxplot of GFAP by baseline and 2.5-year follow-up ................................................. 24
Figure 3: Boxplot of Aβ40 by baseline and 2.5-year follow-up .................................................. 24
Figure 4: Boxplot of Aβ42 by baseline and 2.5-year follow-up .................................................. 25
Figure 5: Boxplot of NfL by baseline and 2.5-year follow-up .................................................... 25
Figure 6: Boxplot of ratio of Aβ42/40 by baseline and 2.5-year follow-up ................................ 26

v
Abstract
Background: Alzheimer’s disease (AD) and dementia are global public health priorities.
Worldwide, over 55 million people are currently living with Alzheimer's disease. In the US about
7 million people are suffering from AD and this number is expected to increase to 14 million by
2060. AD is usually not diagnosed until symptomatic, even though neurodegenerative processes
start decades before the symptoms. However, it is not practical to conduct PET diagnostic or CSF
analysis to identify patients pre-symptomatically on a large scale. Recent years, research involving
blood biomarkers including ptau181, GFAP, Aβ40, Aβ42, and NfL have shown to be associated
with AD. However, knowledge is limited on plasma AD biomarkers in postmenopausal women
and if and how those biomarkers are associated with cognition.
Methods: We conducted a post hoc analysis in healthy postmenopausal women, who participated
in the Early versus Late Intervention Trial with Estradiol (ELITE), a randomized controlled trial
evaluating the effect of hormone therapy on carotid artery atherosclerosis and cognition over an
average 4.8 years. Cognitive skills were assessed at baseline, 2.5 years, and 5 years of the end of
trial using a battery of 14 tests. Plasma levels of Aβ40, Aβ42, ptau181, GFAP, and NFL were
measured at baseline and 2.5-year visit using single molecule array (Simoa) technology. Blood
biomarker levels at baseline and cognitive scores from 3 domains including global cognition,
executive functions, and verbal memory were tested on using linear mixed effects models. Changes
in the blood biomarkers over 2.5 and cognitive skills were also evaluated for association using
mixed effects models.
Results: Baseline plasma concentrations of GFAP and NfL were significantly higher in late
postmenopausal women compared to the early postmenopausal women (p < 0.02), Aβ40 levels

vi
were significantly lower in the late group (p = 0.0003). The Aβ42/40 ratio was statistically
significantly higher in the late postmenopausal group (p = 0.04).
In longitudinal analysis, baseline GFAP levels were significantly inversely associated with
executive memory (β (SE) = -0.0003 (0.00013); p = 0.03). Baseline plasma NfL concentration was
inversely associated with executive memory with a borderline statistical significance (β (SE) = -
0.0013 (0.00069); p = 0.07). Baseline plasma ptau181 concentration showed inverse association
with global cognition with borderline statistical significance (β (SE) = -0.0013 (0.00069); p = 0.08).
Conclusion: Our study showed that some plasma AD biomarkers were associated with cognitive
decline over 2.5 years in cognitively healthy postmenopausal women. These data indicate a
potential use of AD risk blood biomarkers in cognitively healthy people for screening individuals
at risk of cognitive decline. Further research is warranted to assess the implications of these results
for dementia and Alzheimer’s disease in women and men.
Keywords: Alzheimer's disease, AD blood biomarkers, Cognitive composite scores, Longitudinal
analysis, Postmenopausal women

1
Introduction
The World Health Organization (WHO) has identified Alzheimer's disease (AD) as a global public
health priority. As a major cause of dependence, disability and mortality,
1 AD is a progressive
neurodegenerative disorder that is characterized by memory loss and difficulties with thinking,
language and problem-solving.
2 AD cases are increasing worldwide. According to a WHO report
there were a total 24 million AD cases in 2020, and the total number of people with dementia is
estimated to increase 4 times by 2050 worldwide.
3 In the United States, an estimated 6.07 million
adults ages 65 and older have clinical AD in 2020, which will increase to 13.85 million by 2060.6
Observational studies have identified a wide range of potential risk factors for AD, including race,
gender, cardiovascular risk factors (hypertension, diabetes, obesity, etc.), psychosocial factors
such as depression, lifestyles (physical activity, smoking, alcohol consumption), head injuries,
infections, environmental factors (heavy metals, trace metals, and others), and most importantly,
genetic factors and increasing age.
3-5 The incidence of dementia increases with age, approximately
5-8% are affected over age of 65, which increases to 25-50% for people ≥ 85 years.
2 The
prevalence of AD for men overall is lower than that for women by 19-29%.2
There are currently only three classes of drugs approved to treat AD, including Aducanumab
(Aduhelm), Lecanemab (Leqembi) and Donanemab. Aducanumab is a monoclonal antibody that
targets amyloid beta plaques, a hallmark of Alzheimer's; the drugs work by reducing these plaques
in the brain. Lecanemab is another monoclonal antibody that targets and clears amyloid beta
plaques in the brain. Donanemab is a monoclonal antibody targeting amyloid plaques, particularly
a modified form of amyloid-beta plaques (the N3pG epitope). Unfortunately, despite the positive

2
effect of these medications, they are effective only in relieving the symptoms of AD, and do not
cure or prevent the disease.3
Early diagnosis or screening is critical for disease prevention. However, AD is usually not
diagnosed until symptomatic, even though neurodegenerative processes are known to start 20 to
30 years before the onset of symptoms.
7 The primary limitations of the standard AD biomarker
diagnostics (core CSF biomarkers and PET scan) are high cost and invasive nature of the tests,
which make them difficult to apply routinely in clinical practice.
Recent studies have shown potential for utilization of blood biomarkers in early diagnosis of AD.
For instance, the single-type proteins including amyloid beta(Aβ)40, Aβ42, Aβ oligomers (AβO)
in amyloid precursor protein metabolism, total tau (t-tau), tau phosphorylated at threonine 181
(ptau181), tau phosphorylated at threonine 217 (ptau217), tau phosphorylated at threonine 231
(ptau231) or a combination of these markers, such as the Aβ42/Aβ40 ratio or the ptau181/Aβ42
ratio have been linked with AD.
11 Apart from the Aβ and tau proteins, recent research has
documented other blood biomarker candidates, such as the neurofilament light (NfL) protein for
neurodegeneration, brain-derived neurotrophic factor (BDNF), and glial fibrillary acidic protein
(GFAP)8-10, which is expressed by numerous cell types of the central nervous system including
astrocytes and ependymal cells during development.36 Further understanding of the correlation
between blood biomarkers and AD risk may offer an simpler and more affordable access for early
diagnosis. For earlier prevention of AD onset, it is important to evaluate the association between
blood biomarkers of AD risk and cognitive function cognitively in healthy individuals. Our
primary objective in this study is to evaluate the associations between blood biomarkers of AD
risk and cognitive function among postmenopausal women. Of note, menopausal transition and

3
early menopause (natural or surgical) have been associated with decreased verbal memory and
dementia, respectively.

4
Chapter 1: Methods
Study design
This study is a post hoc analysis of postmenopausal women who participated in the Early versus
Late Intervention Trial with Estradiol (ELITE). Details of the ELITE methods and primary results
have been published.12 Briefly, ELITE was a randomized, double-blinded, placebo-controlled,
factorial designed clinical trial, conducted from June 2005 through February 2013 with a median
follow-up of 4.8 (range from 0.5 to 6.7) years.
12 A total of 643 postmenopausal women were
initially divided into early (< 6 years since menopause) and late (≥ 10 years past menopause)
postmenopausal groups. Within each postmenopausal group, women were randomized to hormone
therapy (HT) or placebo. The primary outcome of ELITE was common carotid artery
atherosclerosis progression measured as intima-media thickness (CIMT).
The effect of HT on cognitive function was also evaluated as a secondary trial outcome.
14 The
participants were from the general population in the Greater Los Angeles17 area meeting the
inclusion criteria of absence of menses for at least 6 months or bilateral oophorectomy, with serum
estradiol level below 25 pg/mL, and no clinical history of cardiovascular disease or diabetes13
.
In this post hoc study, outcome variables were three cognitive function weighted average of
component z-scores (global cognition, executive functions, and verbal memory) calculated from a
battery of 14 tests measured at baseline and two post-randomization visits (around 2.5 years and 5
years after baseline).
13 Five blood biomarkers of AD risk including ptau181, GFAP, Aβ40, Aβ42,
and NfL were measured in stored samples from baseline and 2.5-year visit. Since several recent
studies have demonstrated that the blood Aβ42/40 ratio is a reliable tool for the assessment of

5
amyloid pathology in the brain,3 we tested the association of Aβ42/40 ratio on the cognitive
composite scores. Socio-demographic information was collected at baseline. Clinical measures
relevant to the ELITE primary objectives including body mass index (BMI), blood pressure and
serum levels of cholesterol, were collected at each study visit during the trial.
AD biomarkers
Five biomarkers of AD risk were measured in the stored plasma samples from ELITE participants
at baseline and 2.5-year visit. Measured biomarkers included Aβ40, Aβ42, ptau181, GFAP, and
NfL. Plasma samples were preprocessed by diluting 4 times and running in duplicate. The ptau181
concentration was measured by Quanterix, (Billerica, MA, USA) using the Simoa® Human
ptau181 Advantage V2.1 assay kit (Product # 104111). The Simoa® ptau181Advantage kits were
used according to manufacturer’s instructions with commercial availability. GFAP, Aβ40, Aβ42,
and NfL were measured with the likewise pre-processed diluted sample at Quanterix (Billerica,
MA, USA) using their commercially available Simoa® Neuro 4-Plex E Advantage kit (Product #
103670) under the manufacturer’s instructions. All assays were performed on the Simoa HD-X
analyzer using Simoa technology.15,16 The mean of replicates for each sample was calculated.
Individual measures were included in the analysis if the coefficient of variation across replicates
was < 25%.
Cognitive composite scores
Three cognitive composite scores were calculated as the linear sums of the standardized test scores
within each domain, with each standard test score inversely weighted by its correlation with other
contributing cognitive tests.18 A comprehensive neuropsychological battery emphasizing
standardized tests sensitive to age-associated change in middle-aged and older adults were used

6
for cognitive assessment.18,19 Briefly, contributing tests for the executive functions composite
score were determined using a principal components analysis of baseline scores: Symbol Digit
Modalities Test (complex scanning and visual tracking, attention, psychomotor speed), Trail
Making Test part B (visuomotor tracking, planning, cognitive flexibility, psychomotor speed),
Shipley Abstraction Scale (concept formation), Letter-Number Sequencing (working memory,
attention, concentration), and category fluency (animal naming). The verbal memory composite
score was determined using California Verbal Learning Test (verbal episodic memory, word list
learning, concept formation) and East Boston Memory Test (verbal episodic memory, logical
memory), immediate and delayed recall for both. The global cognitive composite score was
calculated as the weighted average of all neuropsychological measures mentioned above as well
as Judgment of Line Orientation (visuospatial perception), Block Design, Visual Memory (Faces
I, immediate recall and Faces II, delayed recall).13,20 Cognitive tests were assessed at baseline, at
about 2.5 years (mean 33 months, SD 2.5, range 29–50), and at end of the study (mean 57 months,
SD 5.8, range 36–77).21
Covariates: The following demographic and clinical characteristics were categorized into
meaningful subgroups for the analysis. Annual income was categorized into four categories: less
than $40,000, $40,000 to $69,999, $70,000 to $99,999, ≥$100,000. Educational level was
categorized in three categories: high school graduate or less, trade or business school after high
school or some college, bachelor’s degree or higher than that. The ApoE4 genotype is the most
well-characterized genetic risk factor for AD. ApoE genotype was classified using an additive
model ApoE4++ (E4/E4), ApoE4+ (E2/E4, E3/E4), or ApoE4- (E2/E2, E2/E3, E3/E3).
Statistical analysis

7
A total of 550 women were included in the baseline analysis after data cleaning. T-tests were used
to verify if the mean AD biomarker levels are statistically significantly different between early and
late postmenopausal groups. Then, we conducted cross-sectional univariable analysis to evaluate
the demographic and clinical correlates of the plasma AD biomarkers and cognitive scores at
baseline in 550 women participating in the ELITE using simple linear regression.
Known correlates of cognitive skills including age, race, BMI, smoking, and ApoE genotype were
considered as potential confounders. Those factors were then tested for their confounding role by
comparing the β estimates between models with and without the potential confounder. All of those
5 factors changed the β estimates for at least one association between AD biomarkers and cognitive
scores by >10% and were included as covariates in the analyses.
A cross-sectional multivariable analysis was performed between each cognitive composite score
and each AD blood biomarker of AD risk at baseline. Separate models were fitted for each
cognitive composite variable. The covariates included age, race, smoking (categorized by never
smoking, former smoker and current smoker), body mass index (BMI), and ApoE genotype.
Furthermore, we evaluated the multivariable associations between cognitive composite scores and
AD blood biomarkers stratified by time since menopause: within 6 years of menopause (early
postmenopausal) and 10 years or more after menopause (late postmenopausal).12
Associations between the AD biomarkers and the cognitive composite scores over time were
assessed using linear mixed effects models, in which the time-varying measures of cognition were
the dependent variable. The primary explanatory variables were follow-up time in years, blood
level of AD biomarkers at baseline, and their interaction with follow-up time. The follow-up time
variable is the duration in years from the baseline visit to each follow-up visit of cognitive

8
assessments. (i.e. zero years for baseline, 2.5 years for second measurement, 5 years for third
measurement). Separate models were fitted to test the association of each AD biomarker level at
baseline with each of the 3 cognition variables. The baseline AD biomarker variable was included
in the mixed effects model as a continuous variable. The rate of cognitive score progression
associated with a per unit of baseline AD biomarker levels was estimated from the regression
coefficient associated with the interaction terms between cognition scores and the follow-up time.
The covariates included in the models were age, race, income, smoking, systolic blood pressure,
and LDL-cholesterol.
We conducted an exploratory analysis evaluating the association of the changes in AD biomarker
levels from baseline to the 2.5-year measure, and the longitudinal measures of the cognition scores.
Mixed effects models like the primary analysis were used with the exception of explanatory
variable being the change in the AD biomarker levels from baseline to 2.5 year.
Statical significance was set at 0.05 for all analysis. All statistical analyses were performed using
SAS version 9.4 (SAS Institute, Inc., Cary, NC).

9
Chapter 2: Results
A total of 550 women with AD biomarker data were included in the baseline analysis. T-tests were
used to compare these 550 women to the entire ELITE cohort on demographic and clinical factors
and there was no statistically significant difference of their means. The demographic and clinical
characteristics of the participants are included in Table 1. On average (SD), the women were 60.6
(7) years old. The majority (68%) were non-Hispanic Whites, followed by 16% Hispanic, then
Black and Asian, each 8%. Family annual income distribution was relatively even. The majority
of the participants were well-educated, non-smokers, and low alcohol users. The participants were
overweight with average (SD) BMI 27.3 (5.41) kg/m2
. The average (SD) DBP (diastolic blood
pressure) and SBP (systolic blood pressure) were within normal range for age based on American
Heart Association guideline. The lipid profile was well managed in the 18% of participants using
antihyperlipidemic medication. Thirty one percent of women were ApoE4+, of which 27% were
E4 heterozygous and 4% were E4 homozygous; 45% were within 6 years of menopause, 14% had
hysterectomy, and 11% had surgical menopause. Approximately, 70% of the women used
hormones in the past.
Table 1
Demographic and Clinical Characteristics of the ELITE Participants at Baseline
Total Sample
Variables a N = 550
Age (year) 60.57 (6.91)
Race
Non-Hispanic While 376 (68.4)
Black 49 (8.9)
Hispanic 76 (13.8)
Asian 49 (8.9)
Family Annual Income
Less than $ 40,000 122 (22.2)

10
From $ 40,000 to $ 69,999 125 (22.7)
From $ 70,000 to $ 99,999 113 (20.6)
Equal to or more than $ 100,000 190 (34.6)
Educational Level
High school graduate or less 18 (3.3)
Business school or some college 157 (28.6)
Bachelor’s degree or higher 375 (68.2)
Smoking Status
Never 331 (60.2)
Former 200 (36.4)
Current 19 (3.5)
Alcohol Use
None 282 (51.3)
>0 199 (36.2)
>1 49 (8.9)
>2 20 (3.6)
Body Mass Index (kg/m2
) 27.28 (5.41)
Diastolic Blood Pressure (mmHg) 75.1 (7.1)
Systolic Blood Pressure (mmHg) 117.9 (12.3)
Antihyperlipidemic Use 101 (18.4)
Total Cholesterol (mg/dl) 223.7 (33.5)
HDL Cholesterol (mg/dl) 66.0 (17.6)
LDL Cholesterol (mg/dl) 136.4 (31.4)
Triglycerides (mg/dl) 106.8 (54.9)
ApoE4
E2/E2 or E2/E3 or E3/E3 379 (68.9)
E2/E4 or E3/E4 150 (27.3)
E4/E4 21 (3.8)
Years since Menopause 10.5 (7.7)
Menopause Stratum
<6 years 248 (45.1)
>=10 years 302 (54.9)
Hysterectomy 79 (14.4)
Type of Menopause
Natural 488 (88.7)
Surgical 62 (11.3)

11
Previous Hormone Use 387 (70.4) a Mean (SD) for continuous variables; n (%) for categorical variables.
The average (SD) levels of the five AD biomarkers and Aβ42/40 ratio measured at baseline are
reported in Table 2. The biomarker concentrations are compared between early and late
postmenopausal women. Compared to the early menopause group, GFAP (p = 0.02) and NfL (p
= 0.003) concentrations were significantly higher in late postmenopausal women, whereas Aβ40
levels were significantly lower in the late postmenopausal group (p = 0.0003). The Aβ42/40 ratio
was statistically significantly higher in the late postmenopausal group (p = 0.04). Concentrations
of ptau181 and Aβ42 did not differ by time since menopause.
Table 2
Levels of AD Biomarkers in the ELITE Participants at Baseline
Total Sample Early Postmenopausal
Group
Late Postmenopausal
Group p-value a
Biomarkers (pg/ml) N = 550 N = 248 N = 302
ptau181 20.89 (7.58) b 20.21 (7.41) 21.44 (7.69) 0.058
GFAP 105.39 (53.17) 99.64 (56.83) 110.11 (49.57) 0.023
Aβ40 80.57 (16.14) 83.35 (17.18) 78.28 (14.87) 0.0003
Aβ42 5.92 (1.35) 6.01 (1.35) 5.84 (1.35) 0.14
NfL 13.18 (9.43) 11.62 (7.72) 14.46 (10.47) 0.0003
Ratio of Aβ42/40 0.074 (0.013) 0.073 (0.012) 0.075 (0.013) 0.04
a Test if the mean AD biomarker levels are statistically significantly different between early and late postmenopausal groups. b Mean (SD)
Associations between participant characteristics and cognitive composite scores are summarized
in Table 3. Age, race, smoking status, BMI, years since menopause stratum, and ApoE genotype
showed statistically significant associations cognitive composite scores. Age was significantly
inversely associated with all 3 cognitive measures (p < 0.01). Compared to non-Hispanic Whites,

12
Blacks, Hispanics, and Asian women had significantly lower global, verbal and executive memory
scores (all p < 0.0001). Higher income was associated with higher cognitive skills for all 3
measures (all p < 0.001). Higher level of education was associated with lower cognitive skills (p
for trend < 0.001 for all 3 cognitive scores). Compared to never smokers, current smokers had
significantly lower global cognitive scores (p < 0.03). Compared to nonalcohol users, there was a
positive association between greater alcohol use and higher global cognition and executive
memory (p for trend < 0.02). Higher HDL-cholesterol levels were associated with greater global
cognition and executive memory. LDL-cholesterol levels were significantly inversely associated
with global cognition. Higher triglyceride levels were significantly inversely associated with all 3
cognition measures (all p < 0.001). Women who had surgical menopause had significantly lower
verbal memory (p < 0.001).
Table 3
Cross-sectional Univariable Associations between Demographic and Clinical Properties and Cognitive
Composite Scores at Baseline
Variables Global Composite Verbal Memory Executive Function
β (SE) p-value β (SE) p-value β (SE) p-value
Age -0.032 (0.012) 0.007 -0.029 (0.0085) 0.001 -0.044 (0.0084) <.0001
Race <.0001a <.0001a <.0001a
Non-Hispanic
White - - - - - -
Black -1.03 (0.26) 0.0001 -0.27 (0.20) 0.18 -0.75 (0.19) <.0001
Hispanic -1.39 (0.22) <.0001 -0.79 (0.16) <.0001 -1.36 (0.16) <.0001
Asian -1.26 (0.27) <.0001 -0.82 (0.20) <.0001 -0.77 (0.19) <.0001
Family Annual
Income <.0001a <.0001a <.0001a
Less than $ 40,000 - - - - - -
From $ 40,000 to
$ 69,999 0.88 (0.22) <.0001 0.41 (0.17) 0.015 0.62 (0.16) 0.0001
From $ 70,000 to
$ 99,999 0.80 (0.23) 0.0005 0.51 (0.17) 0.003 0.80 (0.17) <.0001
Equal to or more
than $ 100,000 1.55 (0.20) <.0001 0.79 (0.15) <.0001 1.28 (0.15) <.0001
Educational
Level <.0001a <.0001a <.0001a

13
High school
graduate or less - - - - - -
Some college
education -1.81 (0.43) <.0001 -0.67 (0.32) 0.039 -1.21 (0.32) 0.0002
College graduate
or more -0.79 (0.17) <.0001 -0.53 (0.13) <.0001 -0.62 (0.12) <.0001
Smoking Status 0.012a 0.13a 0.018 a
Never - - - - - -
Former 0.27 (0.16) 0.097 0.074 (0.12) 0.54 0.28 (0.12) 0.021
Current -0.94 (0.42) 0.028 -0.57 (0.31) 0.068 -0.40 (0.31) 0.20
Alcohol Use 0.042a 0.27a 0.040a
None - - - - - -
>0 0.42 (0.17) 0.014 0.086 (0.12) 0.49 0.21 (0.12) 0.088
>1 0.29 (0.28) 0.31 0.19 (0.21) 0.36 0.38 (0.21) 0.070
>2 0.78 (0.43) 0.069 0.57 (0.32) 0.070 0.68 (0.31) 0.031
Body Mass Index -0.019 (0.014) 0.19 -0.014 (0.011) 0.18 -0.020 (0.011) 0.065
Diastolic Blood
Pressure -0.023 (0.011) 0.034 -0.015 (0.0081) 0.058 -0.0037 (0.0081) 0.65
Systolic Blood
Pressure -0.019 (0.0063) 0.0023 -0.015 (0.0046) 0.0013 -0.012 (0.0046) 0.013
Antihyperlipidem
ic Use 0.79a 0.74a 0.64a
No - - - - - -
Yes 0.054 (0.20) 0.79 0.049 (0.15) 0.74 -0.069 (0.15) 0.64
Total Cholesterol -0.0043
(0.0024) 0.074 -0.0036 (0.0018) 0.043 -0.0015 (0.0018) 0.39
HDL Cholesterol 0.013 (0.0045) 0.003 0.0060 (0.0033) 0.071 0.0099 (0.0033) 0.003
LDL Cholesterol
-0.0057
(0.0026) 0.03 -0.0033 (0.0019) 0.085 -0.0024 (0.0019) 0.21
Triglycerides
-0.0056
(0.0014) <.0001 -0.0044 (0.0011) <.0001 -0.0041 (0.0011) 0.0001
ApoE4 0.43a 0.51 a 0.21 a
E2/E2 or E2/E3 or
E3/E3 - - - - - -
E2/E4 or E3/E4 -0.20 (0.18) 0.27 -0.15 (0.13) 0.25 -0.13 (0.13) 0.31
E4/E4 -0.33 (0.42) 0.43 -0.050 (0.31) 0.87 -0.47 (0.31) 0.12
Years since
Menopause -0.030 (0.010) 0.0036 -0.027 (0.0075) 0.0004 -0.038 (0.0074) <.0001
Menopause
Stratum 0.012a 0.0048a <.0001a
<6 years - - - - - -
>=10 years -0.39 (0.16) 0.012 -0.33 (0.11) 0.0048 -0.60 (0.11) <.0001

14
Hysterectomy 0.56a 0.11a 0.16a
No - - - - - -
Yes -0.13 (0.22) 0.56 -0.26 (0.17) 0.11 -0.23 (0.17) 0.16
Type of
Menopause 0.44a 0.0089a 0.31a
Natural - - - - - -
Surgical -0.19 (0.25) 0.44 -0.48 (0.18) 0.0089 -0.19 (0.18) 0.31
Previous
Hormone Use 0.30a 0.48a 0.0011a
No - - - - - -
Yes -0.18 (0.17) 0.30 -0.090 (0.13) 0.48 -0.41 (0.12) 0.0011 a Global P-value. b β (SE): beta-estimates (standard error) from linear regression.
The associations between demographic and clinical characteristics and AD blood biomarkers at
baseline are presented in Table 4. All 5 biomarkers of AD risk are significantly positively
associated with age. Compared to non-Hispanic White women, Black and Hispanic women had
significantly lower levels of Aβ40, Aβ42, and NfL levels. Hispanic women had significantly lower
GFAP and ptau181 levels compared to non-Hispanic White women. Current smokers had lower
levels of GFAP, Aβ40 and Aβ42 (all p < 0.01). GFAP, Aβ42 and NfL were significantly negatively
associated with BMI. As for ApoE genotype, ApoE-positive women had lower levels of Aβ42 and
lower Aβ42/40 ratio. Women having longer years since menopause (10 years or more after
menopause) had higher levels of ptau181, GFAP, NfL, and lower levels of Aβ40, Aβ42 (but higher
Aβ42/40 ratio).

Table 4
Cross
15
-sectional Association between AD Blood Biomarkers and Demographic and Clinical Factors at Baseline
Variables Log of ptau181 Log of GFAP Aβ40 Aβ42 Log of NfL Ratio of Aβ42/40
β (SE) p- value β (SE) p- value β (SE) p- value β (SE) p- value β (SE) p- value β (SE) p- value
Age 0.007 (0.002) 0.001 0.024 (0.002) <.0001 0.13 (0.10) 0.20 0.001 (0.008) 0.91 0.029 (0.003) <.0001 -0.0001 (0.00008) 0.17
Race 0.0004 0.0093 <.0001 0.0073 <.0001 0.22
Non-Hispanic White - - - - - - - - - - - -
Black -0.08 (0.048) 0.097 -0.097 (0.062) 0.12
-10.34
(2.41) <.0001 -0.54 (0.20) 0.0079
-0.29
(0.06) <.0001 0.003 (0.002) 0.17
Hispanic -0.17 (0.04) <.0001 -0.15 (0.05) 0.0041 -5.46 (1.99) 0.006 -0.43 (0.17) 0.011
-0.20
(0.05) 0.0002 -0.001(0.002) 0.51
Asian -0.026 (0.048) 0.59 0.052 (0.06) 0.40 -3.69 (2.41) 0.13 -0.043 (0.20) 0.83
-0.068
(0.064) 0.29 0.0026 (0.0019) 0.17
Family Annual
Income 0.38 0.74 0.061 0.085 0.0071 0.16
Less than $ 40,000 - - - - - - - - - - - -
From $ 40,000 to
$ 69,999
-0.06 (0.041) 0.15 -0.05 (0.052) 0.34 -1.96 (2.05) 0.34 -0.17 (0.17) 0.33
-0.14
(0.06) 0.01 -0.0006 (0.002) 0.72
From $ 70,000 to
$ 99,999
-0.057 (0.042) 0.18 -0.021 (0.053) 0.69 -0.63 (2.10) 0.76 -0.33 (0.18) 0.063
-0.18
(0.06) 0.002 -0.003 (0.002) 0.04
Equal to or more than
$ 100,000
-0.019 (0.037) 0.62 -0.046 (0.048) 0.34 2.78 (1.86) 0.14 0.055 (0.16) 0.73
-0.07
(0.05) 0.15 -0.002 (0.002) 0.18
Educational Level 0.62 0.2292 0.83 0.74 0.48 0.44
Bachelor’s degree or
higher
- - - - - - - - - - - -
High school graduate
or less
-0.046 (0.078) 0.55 0.17 (0.099) 0.088 0.22 (3.90) 0.95 0.19 (0.33) 0.55 0.11 (0.10) 0.29 0.0021 (0.0030) 0.48
Business school or
some college
-0.026 (0.031) 0.40 0.00084 (0.039) 0.98 -0.91 (1.54) 0.55 0.073 (0.13) 0.57 -0.020 (0.041) 0.63 0.0014 (0.0012) 0.25
Smoking Status 0.82 0.024 0.0046 0.0014 0.12 0.44
Never - - - - - - - - - - - -
Former 0.0093 (0.029) 0.75 0.019 (0.036) 0.59 1.38 (1.43) 0.34 0.11 (0.12) 0.36 0.034 (0.039) 0.38 0.0001(0.001) 0.94
Current 0.044 (0.076) 0.56 -0.25 (0.096) 0.0099
-11.27
(3.78) 0.003 -1.066 (0.32) 0.001
-0.17
(0.10) 0.09 -0.004 (0.003) 0.21
Alcohol Use 0.38 0.71 0.23 0.095 0.45 0.05

16
None - - - - - - - - - - - -
>0 0.030 0.45
-0.0026
(0.038) 0.95 -0.36 (1.49) 0.81 -0.25 (0.12) 0.046
-0.020
(0.040) 0.62 -0.003 (0.001) 0.02
>1 -0.047 (0.050) 0.35
-0.0099
(0.063) 0.88 -2.79 (2.49) 0.26 -0.38 (0.21) 0.069
-0.063
(0.067) 0.35 -0.002 (0.002) 0.22
>2 -0.076 (0.075) 0.31 -0.11 (0.095) 0.24 -6.81 (3.73) 0.068 -0.35 (0.31) 0.27
-0.14
(0.10) 0.16 0.0026 (0.0029) 0.36
Body Mass Index
-0.00070
(0.0025) 0.78
-0.015
(0.0032) <.0001 -0.24 (0.13) 0.056
-0.025
(0.011) 0.021
-0.016
(0.0034) <.0001
-0.000088
(0.000098) 0.37
Diastolic Blood
Pressure
-0.0046
(0.0019) 0.017
-0.0037
(0.0024) 0.13
-0.11
(0.096) 0.27
-0.011
(0.0081) 0.16
-0.011
(0.0026) <.0001
-
0.000042 (0.00
0075)
0.58
Systolic Blood
Pressure
-0.0023
(0.0011) 0.042 0.0019 (0.0014) 0.19 -0.059 (0.056) 0.28 -0.0052 (0.0047) 0.26 -0.00072 (0.0015) 0.63 -0.0000093 (0.000043) 0.83
Antihyperlipidemic
Use 0.75 0.042 0.090 0.026 0.17 0.43
No - - - - - - - - - - - -
Yes -0.011 (0.036) 0.75 0.091 (0.045) 0.042 3.01 (1.77) 0.090 0.33 (0.15) 0.026 0.065 (0.048) 0.17 0.0011 (0.0014) 0.43
Total Cholesterol
-0.00069
(0.00042) 0.10
-0.000075
(0.00053) 0.89
-0.040
(0.021) 0.054
-0.0035
(0.0018) 0.048
-0.000012
(0.00056) 0.98
-
0.0000060 (0.0
00016)
0.71
HDL Cholesterol 0.0017 (0.00080) 0.031 0.0035 (0.0010) 0.0006 -0.061 (0.040) 0.13 0.0027 (0.0034) 0.42 0.0043 (0.0011) <.0001 0.000079 (0.000031) 0.010
LDL Cholesterol
-0.0012
(0.00045) 0.01
-0.00091
(0.00057) 0.11
-0.026
(0.022) 0.25
-0.0041
(0.0019) 0.028
-0.00078
(0.00060) 0.20
-
0.000024 (0.00
0017)
0.16
Triglycerides
-0.00029
(0.00025) 0.25
-0.00043
(0.00032) 0.18
-0.0024
(0.013) 0.85
-0.0011
(0.0011) 0.30
-0.00092
(0.00034) 0.0068
-0.000012
(0.000009) 0.21
ApoE4 0.50 0.24 0.16 <.0001 0.51 <.0001
E2/E2 or E2/E3 or
E3/E3
- - - - - - - - - - - -
E2/E4 or E3/E4 -0.067 (0.031) 0.98 0.027 (0.039) 0.49 -0.28 (1.55) 0.86 -0.36 (0.13) 0.0054
-0.027
(0.042) 0.51
-0.0041
(0.0012) 0.0005
E4/E4 0.063 (0.072) 0.24 0.15 (0.092) 0.11 -6.90 (3.61) 0.057 -1.25 (0.30) <.0001
-0.10
(0.097) 0.31 -0.011 (0.0028) 0.0001

17
Years since
Menopause
0.00533
(0.00180) 0.0031 0.012 (0.0022) <.0001
-0.23
(0.090) 0.0094
-0.0097
(0.0076) 0.20 0.019 (0.0023) <.0001 0.0011 (0.000070) 0.20
Menopause Stratum 0.027 0.0048 0.0002 0.14 <.0001 0.040
<6 years - - - - - - - - - - - -
>=10 years 0.061 (0.028) 0.027 0.099 (0.035) 0.0048 -5.06 (1.37) 0.0002 -0.17 (0.12) 0.14 0.22 (0.036) <.0001 0.0022 (0.0011) 0.040
Hysterectomy 0.35 0.94 0.028 0.33 0.29 0.11
No - - - - - - - - - - - -
Yes 0.037 (0.039) 0.35
-0.0038
(0.050) 0.94 -4.31 (1.96) 0.028 -0.16 (0.16) 0.33 0.056 (0.053) 0.29 0.0024 (0.0015) 0.11
Type of Menopause 0.42 0.38 0.017 0.44 0.20 0.040
Natural - - - - - - - - - - - -
Surgical 0.035 (0.045) 0.42 0.049 (0.055) 0.38 -5.20 (2.17) 0.017 -0.14 (0.18) 0.44 0.0748 (0.058) 0.20 0.0035 (0.0017) 0.040
Previous Hormone
Use 0.54 0.11 0.24 0.77 0.0016 0.38
No - - - - - - - - - - - -
Yes 0.019 (0.030) 0.54 0.062 (0.038) 0.11 -1.77 (1.51) 0.24 -0.036 (0.13) 0.77 0.13 (0.040) 0.0016 0.0010 (0.0012) 0.38
a β (SE): beta-estimates (standard error) from linear regression.

18
Univariable analysis of baseline associations
In unadjusted univariable cross-sectional analysis, higher baseline Aβ40 was associated with
higher global cognitive scores (β (SE) = 0.011 (0.004); p = 0.015). Increased Aβ40 was also
significantly positively associated with higher executive function (β (SE) = 0.0097 (0.003); p =
0.003) as shown in Table 5. Aβ42 concentration was significantly positively associated with
executive function (β (SE) = 0.01 (0.04); p = 0.01).
Table 5
Cross-sectional Univariable Associations between AD Biomarkers and Cognitive Composite Scores at
Baseline
Variables (pg/ml) Global Composite Verbal Memory Executive Function
β (SE) p-value β (SE) p-value β (SE) p-value
ptau181 0.0039 (0.0097) 0.68 -0.0047 (0.007) 0.51 0.0031 (0.007) 0.66
GFAP -0.0016 (0.001) 0.27 -0.0014 (0.001) 0.17 -0.0014 (0.001) 0.18
Aβ40 0.011 (0.004) 0.015 0.005 (0.003) 0.11 0.0097 (0.003) 0.003
Aβ42 0.10 (0.05) 0.06 0.048 (0.04) 0.22 0.10 (0.04) 0.01
Ratio of Aβ42/40 0.02 (5.66) 0.99 -0.085 (4.17) 0.98 1.47 (4.22) 0.73
NfL -0.004 (0.008) 0.62 -0.0026 (0.006) 0.66 -0.009 (0.006) 0.15 a β (SE): beta-estimates (standard error) from linear regression.
Multivariable analysis of baseline associations
Adjusted for age, race, smoking status, BMI, and ApoE4 genotype, no AD blood biomarkers
showed statistically significant cross-sectional associations with cognitive scores (Table 6).
However, after stratifying by time since menopause, GFAP concentration showed a statistically
significant negative association with verbal memory scores (β (SE) = -0.0034 (0.0016); p = 0.034)
in the late postmenopausal group. There were no statistically significant associations between AD
blood biomarkers and cognitive scores in the early postmenopausal group, as shown in Table 7.

19
Table 6
Cross-sectional Multivariable Analysis of Associations between AD Biomarkers and Cognitive Composite
Scores at Baseline
Variables (pg/ml) Global Composite Verbal Memory Executive Function
β (SE) p-value β (SE) p-value β (SE) p-value
ptau181 -0.0036 (0.0093) 0.70 -0.0079 (0.007) 0.26 0.0013 (0.0067) 0.84
GFAP -0.0012 (0.0014) 0.41 -0.00069 (0.0011) 0.52 -0.000028 (0.001) 0.98
Aβ40 0.004 (0.0043) 0.36 0.0022 (0.0032) 0.50 0.0055 (0.0031) 0.07
Aβ42 0.014 (0.053) 0.79 0.0027 (0.04) 0.95 0.037 (0.038) 0.33
NfL -0.0058 (0.0081) 0.47 -0.0013 (0.0061) 0.83 -0.0051 (0.0058) 0.38
Ratio of Aβ42/40 -2.60 (5.47) 0.63 -2.29 (4.09) 0.58 -1.92 (3.89) 0.62 a β (SE): beta-estimates (standard error) from multivariable analysis. b Age, race, smoking status and BMI and ApoE4 genotype were adjusted in the model.

20
Table 7
Cross-sectional Stratified Multivariable Analysis of Associations between AD Biomarkers and Cognitive Composite Scores at Baseline
Variables
(pg/ml) Global Composite Verbal Memory Executive Function
Early Postmenopausal
Group
Late Postmenopausal
Group
Early Postmenopausal
Group
Late Postmenopausal
Group
Early Postmenopausal
Group
Late Postmenopausal
Group
β (SE)
pvalue β (SE) p- value β (SE) p- value β (SE) p- value β (SE) p- value β (SE) p- value
ptau181 0.00091 (0.015) 0.95 -0.0043 (0.012) 0.72 0.0022 (0.011) 0.84 -0.013 (0.0093) 0.18 -0.0021 (0.011) 0.84 0.00085 (0.0085) 0.92
GFAP -0.0005 (0.0021) 0.81 -0.0027 (0.002) 0.18 0.0013 (0.0015) 0.38 -0.0034 (0.0016) 0.034 0.00062 (0.0014) 0.67 -0.0018 (0.0015) 0.22
Aβ40 0.0044 (0.0067) 0.51 -0.0026 (0.0061) 0.67 0.004 (0.0047) 0.41 -0.0029 (0.0048) 0.55 0.0056 (0.0047) 0.23 0.00054 (0.0044) 0.90
Aβ42 -0.06 (0.085) 0.48 0.052 (0.068) 0.44 -0.037 (0.06) 0.53 0.019 (0.053) 0.72 0.012 (0.059) 0.84 0.045 (0.049) 0.36
NfL -0.012 (0.015) 0.42 -0.00028 (0.0096) 0.98 0.0048 (0.011) 0.65 -0.0039 (0.0076) 0.60 -0.0028 (0.01) 0.78 -0.0047 (0.0069) 0.50
Ratio of
Aβ42/40 -11.41 (8.90) 0.20 10.09(7.09) 0.16 -10.64 (6.29) 0.092 6.47 (5.59) 0.25 -3.66 (6.20) 0.56 4.82 (5.11) 0.35
a β (SE): beta-estimates (standard error) from stratified multivariable analysis.
b Age, race, smoking status, BMI and ApoE genotype were adjusted in the model.

21
Longitudinal analysis
We used linear mixed effects models to test the longitudinal effect of baseline levels of AD biomarkers on
change in cognition scores. Adjusted for age, race, smoking, BMI, and ApoE genotype, executive function
was inversely associated with baseline GFAP level (β-estimate -0.0003 (SE = 0.00013), p = 0.018) (Table
8). This result indicates that with each 1 pg/ml increase in the baseline level of GFAP, executive function
score decreased by 0.0003 (SE = 0.00013) units per year (Table 8). Baseline NfL levels had a borderline
statistically significant inverse association with executive function (β-estimate -0.0013 (SE =
0.00069), p = 0.07), which indicated with 1 pg/ml great level of baseline NfL, the executive
function reduced by -0.0013 (SE = 0.00069) per units per year.
Table 8
Longitudinal Analysis of Associations between Baseline AD Blood Biomarkers and Longitudinal
Measures of Cognition
Variables Global Composite Verbal Memory Executive Function
β (SE)
pvalue β (SE)
pvalue β (SE)
pvalue
Baseline ptau181 0.0014 (0.0092) 0.88 -0.0069 (0.0068) 0.31 0.0028 (0.0067) 0.67
Years*Baseline ptau181 0.0027 (0.0015) 0.075 0.00055 (0.0014) 0.70 -0.00056 (0.00094) 0.55
Baseline GFAP -0.0012 (0.0014) 0.40 -0.00063 (0.001) 0.54 0.00029 (0.001) 0.77
Years*Baseline GFAP -0.00017 (0.00021) 0.40 0.0000065 (0.0002) 0.97 -0.0003 (0.00013) 0.018
Baseline Aβ40 0.0034 (0.0042) 0.42 0.0016 (0.0032) 0.61 0.0055 (0.0031) 0.075
Years*Baseline Aβ40 0.000086 (0.00072) 0.90 0.00052 (0.00068) 0.45 -0.00019 (0.00044) 0.66
Baseline Aβ42 0.025 (0.052) 0.63 -0.00097 (0.039) 0.98 0.035 (0.038) 0.36
Years*Baseline Aβ42 -0.0090 (0.0088) 0.31 0.0043 (0.0083) 0.61 0.0000093 (0.0054) 0.99
Baseline NfL -0.0042 (0.0077) 0.59 -0.0013 (0.0058) 0.82 -0.0051 (0.0056) 0.36
Years*Baseline NfL -0.0006 (0.0011) 0.60 0.000079 (0.0011) 0.94 -0.0013 (0.00069) 0.07
Baseline ratio of Aβ42/40 -3.93 (5.54) 0.48 -3.03 (4.11) 0.46 -1.77 (4.08) 0.66
Years*Baseline ratio of
Aβ42/40 0.54 (0.88) 0.54 0.34 (0.84) 0.68 0.68 (0.54) 0.21 a β (SE): beta-estimates (standard error) mixed effect model. b Age, race, smoking status and BMI and ApoE genotype were adjusted in the model.

22
Table 9
Longitudinal Analysis of Associations between 2.5-year AD Biomarkers Changes and Slopes of Global
Composite Scores (Univariable and Multivariable)
Variables Global Composite
Univariable Multivariable
β (SE) p-value β (SE) p-value
Years*Change of ptau181 -0.0016 (0.0015) 0.30 -0.0015 (0.0016) 0.32
Years*Change of GFAP 0.0005 (0.00029) 0.082 0.00058 (0.00029) 0.047
Years*Change of Aβ40 -0.00065 (0.00071) 0.36 -0.001 (0.00075) 0.17
Years*Change of Aβ42 0.011 (0.011) 0.32 0.0047 (0.012) 0.69
Years*Change of NfL -0.00016 (0.00096) 0.87 -0.00003 (0.00095) 0.97
Years*Change of ratio of
Aβ42/40 4.63 (1.30) 0.0004 4.62 (1.38) 0.0008 a β (SE): beta-estimates (standard error) from mixed effect model. b Age, race, smoking status and BMI and ApoE genotype were adjusted in the model.
Table 10
Longitudinal Analysis of Associations between 2.5-year AD Biomarkers Changes and Slopes of Verbal
Memory Scores (Univariable and Multivariable)
Variables Verbal Memory
Univariable Multivariable
β (SE) p-value β (SE) p-value
Years*Change of ptau181 -0.00094 (0.0014) 0.50 -0.00017 (0.0015) 0.91
Years*Change of GFAP 0.00025 (0.00027) 0.36 0.00028 (0.00028) 0.31
Years*Change of Aβ40 -0.00005 (0.00067) 0.94 -0.00035 (0.00071) 0.62
Years*Change of Aβ42 0.0076 (0.011) 0.47 0.0019 (0.011) 0.87
Years*Change of NfL -0.00016 (0.00090) 0.86 -0.0001 (0.00091) 0.91
Years*Change of ratio of
Aβ42/40 1.35 (1.23) 0.27 1.14 (1.32) 0.39 a β (SE): beta-estimates (standard error) mixed effect model. b Age, race, smoking status and BMI and ApoE genotype were adjusted in the model.

23
Table 11
Longitudinal Analysis of Associations between 2.5-year AD Biomarkers Changes and Slopes of
Executive Function Scores (Univariable and Multivariable)
Variables Executive Function
Univariable Multivariable
β (SE) p-value β (SE) p-value
Years*Change of ptau181 -0.00049 (0.00095) 0.60 0.00013 (0.00096) 0.90
Years*Change of GFAP 0.00035 (0.00018) 0.054 0.00041 (0.00018) 0.02
Years*Change of Aβ40 -0.00022 (0.00045) 0.62 -0.00023 (0.00046) 0.61
Years*Change of Aβ42 0.0045 (0.0071) 0.52 0.0034 (0.0073) 0.64
Years*Change of NfL 0.00072 (0.0006) 0.066 0.00098 (0.00059) 0.094
Years*Change of ratio of
Aβ42/40 1.57 (0.82) 0.057 1.15 (0.85) 0.18 a β (SE): beta-estimates (standard error) mixed effect model. b Age, race, smoking status and BMI and ApoE genotype were adjusted in the model.
Fig. 1. Boxplot of ptau181 by baseline and 2.5-year follow-up.
0 2.5
Visit (Visit 0 is baseline)
20
40
60
80
100
ptau181 (pg/ml)

24
Fig. 2. Boxplot of GFAP by baseline and 2.5-year follow-up.
Fig. 3. Boxplot of Aβ40 by baseline and 2.5-year follow-up.
0 2.5
Visit (Visit 0 is baseline)
0
200
400
600
800
G
FA
P (pg/ml)
0 2.5
Visit (Visit 0 is baseline)
50
75
100
125
Aß40 (pg/ml)

25
Fig. 4. Boxplot of Aβ42 by baseline and 2.5-year follow-up.
Fig. 5. Boxplot of NfL by baseline and 2.5-year follow-up.
0 2.5
Visit (Visit 0 is baseline)
2
4
6
8
10
Aß42 (pg/ml)
0 2.5
Visit (Visit 0 is baseline)
0
50
100
150
NfL (pg/ml)

26
Fig. 6. Boxplot of ratio of Aβ42/40 by baseline and 2.5-year follow-up.
We used the 2.5-year change of AD blood biomarkers as independent variables to test the effect
of AD biomarker changes on cognitive function. The interaction term between time in years and
the biomarker changes estimated the effect of biomarker changes on cognitive functions per year.
The distribution of AD blood biomarkers at baseline and the 2.5-year post-randomization visits
are shown in Fig. 1 to Fig. 6. In the multivariable analysis, the global composite score increased
0.00058 (SE = 0.00029) units per year when the change of GFAP increased 1 pg/ml, which showed
a statistically significant positive association (p = 0.047). The interaction term of time and the 2.5-
year change in the ratio of Aβ42/40 showed a significant positive association (β (SE) = 4.62 (1.38);
p = 0.0008) with global score as well, as shown in Table 9. The executive function score increased
0.00041 (SE = 0.00018) units per year when the 2.5-year change of GFAP increased 1 pg/ml,
which showed a statistically significant positive association (p = 0.02), as shown in Table 11.
Changes in AD biomarkers were not associated with changes in verbal memory.
0 2.5
Visit (Visit 0 is baseline)
0.04
0.06
0.08
0.10
0.12
ratio of Aß42/40

27
Chapter 3: Discussion
In this sample of postmenopausal women, many demographic characteristics showed statistically
significantly associations with AD blood biomarkers and cognitive scores. Among these
demographic and clinical characteristics, most have been described in previous AD studies,
23-33
expect for hormone usage, hysterectomy, years since menopause and type of menopause (natural
or surgical). In our study, previous hormone usage was significantly positively associated with
NfL blood level and negatively associated with the executive function score. Women who
underwent hysterectomy showed significantly reduced Aβ40 blood levels; surgical menopause
was statistically significantly associated with reduced Aβ40 blood levels and increased Aβ42/40
ratio, as well as a lower verbal memory score. Years since menopause (<6 years since menopause
or ≥10 years post menopause) were statistically significantly associated with all cognitive scores
and AD blood biomarkers except Aβ42 levels.
Comparing our study with previous AD studies, taking race into consideration among AD
individuals,
25,26 some study reported a non-significant association with clinical AD diagnosis,
however we showed significant racial differences when using cognitive score as our outcome.
Another study22 reported positive association of AD in Hispanic individuals compared with nonHispanic Whites while we showed the opposite result. One possible reason for differential
associations is that we used a different sample than did this previous study. Our sample included
only postmenopausal females whereas the previous study included both sexes. Sex has an
important effect on AD risk.
On the other hand, other results from our study showed consistency with and supported previous
studies. For example, ApoE4 genotype is a known modifier of AD risk29. In our study, ApoE

28
genotype was significantly associated with both Aβ42 blood level and the Aβ42/40 ratio, although
genotype was not associated with cognitive scores, which is consist with other AD studies25-27.
According to the results of the cross-sectional univariable analysis in our study, blood Aβ40 levels
had a positive statistically significantly association with both global cognitive score and executive
function score. On the cross-sectional multivariable analysis controlling for age, race, smoking
status and BMI and ApoE genotype, GFAP had a statistically significant negative association with
verbal memory. Many other studies have reported similar results concerning AD blood biomarkers
with different outcomes, but direction of associations vary.
25,26,32-35
A similar longitudinal analysis to our study using a different sample and different cognitive score
test assessments (MMSE or CERAD-TS) was performed by researchers in Korea (participants
were nationwide Koreans aged 58 years or older with both genders; and as for their results, NfL
showed a statistically significantly negative association with MMSE).
30 Another study, derived
from participants aged 55 to 90 years across the United States and Canada with both genders,
showed that the increase in plasma NfL was associated with changes in other established measures
of neurodegeneration in AD (significantly negatively associated with MMSE, significantly
positively associated with ADAS-Cog score and CDR-SB score).
31 Our longitudinal analysis
results are unique from other studies since we derived them from our specific cohort of
postmenopausal women.
One of the primary limitations in our study was that the sample was derived from a healthy
postmenopausal female population. Most of a healthy population will not develop cognitive
decline over the study period. The 2.5-year duration of the cognitive function follow-up was not
sufficient to detect cognitive decline. However, for the same reason, the trends observed in our

29
study may have broader applicability and could exhibit greater sensitivity in future studies
involving populations with clinical manifestations. Furthermore, since we studied a healthy
population, we were unable to evaluate pathological or clinical Alzheimer’s disease (AD)
diagnoses as the primary outcomes. As a result, our findings may be influenced by the correlations
of AD biomarkers with cognitive scores rather than AD diagnosis.
A main issue in clinical Alzheimer's disease is early diagnosis before symptom onset. Blood
biomarkers are potential indicators that could be useful for early diagnosis of AD. Blood tests for
these biomarkers are more clinically accessible than standard diagnostics such as evaluation of
cerebrospinal fluid biomarkers (invasive) and PET scanning (expensive and limited). Our results
indicated that there was only one statistically significantly negative association detected in crosssectional multivariable analysis between GFAP and verbal memory in the early postmenopausal
group, and the results of simple linear regression and mixed effect longitudinal analysis showed a
potential tendency for blood biomarkers evaluated at baseline to correlate longitudinal cognitive
function decline in healthy postmenopausal women. Our study provided an intriguing perspective
for future studies to examine whether the associations between AD blood biomarkers and cognitive
functions persist or intensify in particular populations with all genders and age. Noting the current
challenges in achieving a definitive pathological diagnosis of Alzheimer’s disease, these
biomarkers may serve as valuable tools for physicians, offering insights and guidance as part of
the clinical criteria for evaluating cognitive functions. While a complete cure for Alzheimer’s
disease remains elusive at present, it is hoped that early interventions hold the potential to
decelerate the process of cognitive function decline.

30
References
1. Lane, C. A., Hardy, J., & Schott, J. M. (2018). Alzheimer's disease. European journal of
neurology, 25(1), 59-70.
2. Khan, S., Barve, K. H., & Kumar, M. S. (2020). Recent advancements in pathogenesis,
diagnostics and treatment of Alzheimer’s disease. Current neuropharmacology, 18(11),
1106-1125.
3. Daviglus, M. L., Bell, C. C., Berrettini, W., Bowen, P. E., Connolly Jr, E. S., Cox, N. J., ...
& Trevisan, M. (2010). National Institutes of Health State-of-the-Science Conference
statement: preventing Alzheimer disease and cognitive decline. Annals of internal
medicine, 153(3), 176-181.
4. Barnes, D. E., & Yaffe, K. (2011). The projected effect of risk factor reduction on
Alzheimer's disease prevalence. The Lancet Neurology, 10(9), 819-828.
5. Breijyeh, Z., & Karaman, R. (2020). Comprehensive review on Alzheimer’s disease:
causes and treatment. Molecules, 25(24), 5789.
6. Rajan, K. B., Weuve, J., Barnes, L. L., McAninch, E. A., Wilson, R. S., & Evans, D. A.
(2021). Population estimates of people with clinical Alzheimer's disease and mild cognitive
impairment in the United States (2020–2060). Alzheimer's & dementia, 17(12), 1966-1975.
7. Altuna-Azkargorta, M., & Mendioroz-Iriarte, M. (2021). Biomarcadores sanguíneos en la
enfermedad de Alzheimer. Neurología, 36(9), 704-710.
8. Ausó, E., Gómez-Vicente, V., & Esquiva, G. (2020). Biomarkers for Alzheimer’s disease
early diagnosis. Journal of Personalized Medicine, 10(3), 114.
9. Mankhong, S., Kim, S., Lee, S., Kwak, H. B., Park, D. H., Joa, K. L., & Kang, J. H. (2022).
Development of Alzheimer’s disease biomarkers: From CSF-to blood-based
biomarkers. Biomedicines, 10(4), 850.
10. Blennow, K. (2017). A review of fluid biomarkers for Alzheimer’s disease: moving from
CSF to blood. Neurology and therapy, 6, 15-24.
11. Kim, K. Y., Shin, K. Y., & Chang, K. A. (2023). GFAP as a potential biomarker for
Alzheimer’s disease: a systematic review and meta-analysis. Cells, 12(9), 1309.
12. Sriprasert, I., Mack, W. J., Hodis, H. N., Allayee, H., Brinton, R. D., & Karim, R. (2019).
Effect of ApoE4 genotype on the association between metabolic phenotype and subclinical
atherosclerosis in postmenopausal women. The American journal of cardiology, 124(7),
1031-1037.

31
13. Karim, R., Koc, M., Rettberg, J. R., Hodis, H. N., Henderson, V. W., John, J. A. S., ... &
Mack, W. J. (2019). Apolipoprotein E4 genotype in combination with poor metabolic
profile is associated with reduced cognitive performance in healthy postmenopausal
women: implications for late onset Alzheimer's disease. Menopause, 26(1), 7-15.
14. Karim, R., Xu, W., Kono, N., Sriprasert, I., Li, Y., Yan, M., ... & Hodis, H. N. (2022).
Effect of menopausal hormone therapy on arterial wall echomorphology: results from the
Early versus Late Intervention Trial with Estradiol (ELITE). Maturitas, 162, 15-22.
15. Rissin, D. M., Kan, C. W., Campbell, T. G., Howes, S. C., Fournier, D. R., Song, L., ... &
Duffy, D. C. (2010). Single-molecule enzyme-linked immunosorbent assay detects serum
proteins at subfemtomolar concentrations. Nature biotechnology, 28(6), 595-599.
16. Wilson, D. H., Rissin, D. M., Kan, C. W., Fournier, D. R., Piech, T., Campbell, T. G., ...
& Duffy, D. C. (2016). The Simoa HD-1 analyzer: a novel fully automated digital
immunoassay analyzer with single-molecule sensitivity and multiplexing. Journal of
laboratory automation, 21(4), 533-547.
17. Lin, F., Pa, J., Karim, R., Hodis, H. N., Han, S. D., Henderson, V. W., ... & Mack, W. J.
(2022). Subclinical carotid artery atherosclerosis and cognitive function in older
adults. Alzheimer's Research & Therapy, 14(1), 63.
18. Henderson, V. W., St. John, J. A., Hodis, H. N., McCleary, C. A., Stanczyk, F. Z., Karim,
R., ... & Mack, W. J. (2013). Cognition, mood, and physiological concentrations of sex
hormones in the early and late postmenopause. Proceedings of the National Academy of
Sciences, 110(50), 20290-20295.
19. Henderson, V. W., St. John, J. A., Hodis, H. N., Kono, N., McCleary, C. A., Franke, A. A.,
& Mack, W. J. (2012). Long-term soy isoflavone supplementation and cognition in women:
a randomized, controlled trial. Neurology, 78(23), 1841-1848.
20. Karim, R., Dang, H., Henderson, V. W., Hodis, H. N., St. John, J., Brinton, R. D., & Mack,
W. J. (2016). Effect of reproductive history and exogenous hormone use on cognitive
function in mid‐and late life. Journal of the American Geriatrics Society, 64(12), 2448-
2456.
21. Henderson, V. W., St. John, J. A., Hodis, H. N., McCleary, C. A., Stanczyk, F. Z., Shoupe,
D., ... & Mack, W. J. (2016). Cognitive effects of estradiol after menopause: a randomized
trial of the timing hypothesis. Neurology, 87(7), 699-708.
22. Dhana, K., Beck, T., Desai, P., Wilson, R. S., Evans, D. A., & Rajan, K. B. (2023).
Prevalence of Alzheimer’s dementia in the 50 US states and 3142 counties: A population
estimate using the 2020 bridged‐race postcensal from the National Center for Health
Statistics. Alzheimer's & Dementia, 19, e074430.

32
23. Sims, J. R., Zimmer, J. A., Evans, C. D., Lu, M., Ardayfio, P., Sparks, J., ... & Kaul, S.
(2023). Donanemab in early symptomatic Alzheimer disease: the TRAILBLAZER-ALZ 2
randomized clinical trial. Jama, 330(6), 512-527.
24. Hayden, K. M., Mielke, M. M., Evans, J. K., Neiberg, R., Molina-Henry, D., Culkin, M., ...
& Espeland, M. A. (2024). Association between Modifiable Risk Factors and Levels of
Blood-Based Biomarkers of Alzheimer’s and Related Dementias in the Look AHEAD
Cohort. JAR life, 13, 1.
25. Brickman, A. M., Manly, J. J., Honig, L. S., Sanchez, D., Reyes‐Dumeyer, D., Lantigua,
R. A., ... & Mayeux, R. (2021). Plasma p‐tau181, p‐tau217, and other blood‐based
Alzheimer's disease biomarkers in a multi‐ethnic, community study. Alzheimer's &
Dementia, 17(8), 1353-1364.
26. Hajjar, I., Yang, Z., Okafor, M., Liu, C., Waligorska, T., Goldstein, F. C., & Shaw, L. M.
(2022). Association of plasma and cerebrospinal fluid Alzheimer disease biomarkers with
race and the role of genetic ancestry, vascular comorbidities, and neighborhood
factors. JAMA Network Open, 5(10), e2235068-e2235068.
27. Garrett, S. L., McDaniel, D., Obideen, M., Trammell, A. R., Shaw, L. M., Goldstein, F. C.,
& Hajjar, I. (2019). Racial disparity in cerebrospinal fluid amyloid and tau biomarkers and
associated cutoffs for mild cognitive impairment. JAMA network open, 2(12), e1917363-
e1917363.
28. McDade, E., Cummings, J. L., Dhadda, S., Swanson, C. J., Reyderman, L., Kanekiyo, M., ...
& Bateman, R. J. (2022). Lecanemab in patients with early Alzheimer’s disease: detailed
results on biomarker, cognitive, and clinical effects from the randomized and open-label
extension of the phase 2 proof-of-concept study. Alzheimer's research & therapy, 14(1),
191.
29. Riedel, B. C., Thompson, P. M., & Brinton, R. D. (2016). Age, APOE and sex: triad of risk
of Alzheimer’s disease. The Journal of steroid biochemistry and molecular biology, 160,
134-147.
30. Park, M. K., Ahn, J., Kim, Y. J., Lee, J. W., Lee, J. C., Hwang, S. J., & Kim, K. C. (2024).
Predicting Longitudinal Cognitive Decline and Alzheimer’s Conversion in Mild Cognitive
Impairment Patients Based on Plasma Biomarkers. Cells, 13(13), 1085.
31. Mattsson, N., Cullen, N. C., Andreasson, U., Zetterberg, H., & Blennow, K. (2019).
Association between longitudinal plasma neurofilament light and neurodegeneration in
patients with Alzheimer disease. JAMA neurology, 76(7), 791-799.
32. Xiao, Z., Wu, X., Wu, W., Yi, J., Liang, X., Ding, S., ... & Ding, D. (2021). Plasma
biomarker profiles and the correlation with cognitive function across the clinical spectrum
of Alzheimer’s disease. Alzheimer's Research & Therapy, 13, 1-12.

33
33. Bolsewig, K., van Unnik, A. A., Blujdea, E. R., Gonzalez, M. C., Ashton, N. J., Aarsland,
D., ... & European-Dementia with Lewy Bodies (E-DLB) Consortium. (2024). Association
of Plasma Amyloid, P-Tau, GFAP, and NfL With CSF, Clinical, and Cognitive Features
in Patients with Dementia with Lewy Bodies. Neurology, 102(12), e209418.
34. Yerramalla, M. S., Darin‐Mattsson, A., Udeh‐Momoh, C. T., Holleman, J., Kåreholt, I.,
Aspö, M., ... & Sindi, S. (2024). Cognitive reserve, cortisol, and Alzheimer's disease
biomarkers: A memory clinic study. Alzheimer's & Dementia.
35. Moussa, C., Hebron, M., Huang, X., Ahn, J., Rissman, R. A., Aisen, P. S., & Turner, R. S.
(2017). Resveratrol regulates neuro-inflammation and induces adaptive immunity in
Alzheimer’s disease. Journal of neuroinflammation, 14, 1-10.
36. Roessmann, U., Velasco, M. E., Sindely, S. D., & Gambetti, P. (1980). Glial fibrillary
acidic protein (GFAP) in ependymal cells during development. An immunocytochemical
study. Brain research, 200(1), 13-21.

Abstract (if available)

Abstract Background: Alzheimer’s disease (AD) and dementia are global public health priorities. Worldwide, over 55 million people are currently living with Alzheimer's disease. In the US about 7 million people are suffering from AD and this number is expected to increase to 14 million by 2060. AD is usually not diagnosed until symptomatic, even though neurodegenerative processes start decades before the symptoms. However, it is not practical to conduct PET diagnostic or CSF analysis to identify patients pre-symptomatically on a large scale. Recent years, research involving blood biomarkers including ptau181, GFAP, Aβ40, Aβ42, and NfL have shown to be associated with AD. However, knowledge is limited on plasma AD biomarkers in postmenopausal women and if and how those biomarkers are associated with cognition.

Methods: We conducted a post hoc analysis in healthy postmenopausal women, who participated in the Early versus Late Intervention Trial with Estradiol (ELITE), a randomized controlled trial evaluating the effect of hormone therapy on carotid artery atherosclerosis and cognition over an average 4.8 years. Cognitive skills were assessed at baseline, 2.5 years, and 5 years of the end of trial using a battery of 14 tests. Plasma levels of Aβ40, Aβ42, ptau181, GFAP, and NFL were measured at baseline and 2.5-year visit using single molecule array (Simoa) technology. Blood biomarker levels at baseline and cognitive scores from 3 domains including global cognition, executive functions, and verbal memory were tested on using linear mixed effects models. Changes in the blood biomarkers over 2.5 and cognitive skills were also evaluated for association using mixed effects models.

Results: Baseline plasma concentrations of GFAP and NfL were significantly higher in late postmenopausal women compared to the early postmenopausal women (p < 0.02), Aβ40 levels were significantly lower in the late group (p = 0.0003). The Aβ42/40 ratio was statistically significantly higher in the late postmenopausal group (p = 0.04).

In longitudinal analysis, baseline GFAP levels were significantly inversely associated with executive memory (β (SE) = -0.0003 (0.00013); p = 0.03). Baseline plasma NfL concentration was inversely associated with executive memory with a borderline statistical significance (β (SE) = - 0.0013 (0.00069); p = 0.07). Baseline plasma ptau181 concentration showed inverse association with global cognition with borderline statistical significance (β (SE) = -0.0013 (0.00069); p = 0.08).

Conclusion: Our study showed that some plasma AD biomarkers were associated with cognitive decline over 2.5 years in cognitively healthy postmenopausal women. These data indicate a potential use of AD risk blood biomarkers in cognitively healthy people for screening individuals at risk of cognitive decline. Further research is warranted to assess the implications of these results for dementia and Alzheimer’s disease in women and men.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The impact of hormone therapy on Alzheimer’s disease biomarkers in early and late postmenopausal women

PDF

Exploring the relationship between menopausal hot flushes and Alzheimer's disease biomarkers: a cross-sectional analysis in postmenopausal women

PDF

Association between endogenous sex hormone levels and cognition: the Women’s Isoflavone Soy Health (WISH) trial

PDF

Association of carotid artery stiffness with measures of cognition in older adults

PDF

Relationship of blood pressure and antihypertensive medications to cognitive change in the BVAIT, WISH, and ELITE clinical trials

PDF

Associations between physical activities with bone mineral density in postmenopausal women

PDF

Associations between inflammatory markers and change in cognitive endpoints

PDF

Long-term blood pressure variability across the clinical and biomarker spectrum of Alzheimer’s disease

PDF

Sex hormones and atherosclerosis in postmenopausal women

PDF

The associations of inflammatory markers with progression of subclinical atherosclerosis in early and late postmenopausal women

PDF

Effect of estradiol on circulating levels of inflammatory cytokines in postmenopausal women

PDF

Association of arterial stiffness progression with subclinical atherosclerosis measurements in postmenopausal women

PDF

Cross-sectional association of blood pressure, antihypertensive medications, MRI volumetric measures and cognitive function scores in an aging population

PDF

Comparison of participant and study partner predictions of cognitive impairment in the Alzheimer's disease neuroimaging initiative 3 study

PDF

The relationship of resistin, leptin, adiponectin, and ghrelin with bone mineral density in healthy postmenopausal women: longitudinal analysis

PDF

Effect of soy isoflavones on anthropometric and metabolic measurements in postmenopausal women

PDF

Associations between isoflavone soy protein (ISP) supplementation and breast cancer in postmenopausal women in the Women’s Isoflavone Soy Health (WISH) clinical trial

PDF

Bone mineral density is associated with carotid atherosclerosis in healthy postmenopausal women: a longitudinal analysis of randomized clinical trial data

PDF

A randomized controlled trial of hormone therapy on postmenopausal women’s quality of life

PDF

Blood-brain barrier pathophysiology in cognitive impairment and injury

Asset Metadata

Creator Jin, Tiansheng (author)

Core Title Association between Blood Biomarkers of Alzheimer’s Disease and Cognition in Postmenopausal Women

Contributor Electronically uploaded by the author (provenance)

School Keck School of Medicine

Degree Master of Science

Degree Program Biostatistics

Degree Conferral Date 2025-05

Publication Date 04/02/2025

Defense Date 04/02/2025

Publisher University of Southern California (original), Los Angeles, California (original), University of Southern California. Libraries (digital)

Tag AD blood biomarkers,Alzheimer's disease,cognitive composite scores,longitudinal analysis,OAI-PMH Harvest,postmenopausal women

Format theses (aat)

Language English

Advisor Karim, Roksana (committee chair), Mack, Wendy (committee member), Hodis, Howard N. (committee member)

Creator Email tjin@usc.edu,jinbohan1992@163.com

Unique identifier UC11399K54E

Identifier etd-JinTianshe-13886.pdf (filename)

Legacy Identifier etd-JinTianshe-13886

Document Type Thesis

Format theses (aat)

Rights Jin, Tiansheng

Internet Media Type application/pdf

Type texts

Source 20250403-usctheses-batch-1248 (batch), University of Southern California Dissertations and Theses (collection), University of Southern California (contributing entity)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu