Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Exploring the relationship between menopausal hot flushes and Alzheimer's disease biomarkers: a cross-sectional analysis in postmenopausal women
(USC Thesis Other)
Exploring the relationship between menopausal hot flushes and Alzheimer's disease biomarkers: a cross-sectional analysis in postmenopausal women
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Copyright 2024 Michelle Hernandez
Exploring the Relationship Between Menopausal Hot Flushes and Alzheimer's Disease
Biomarkers: A Cross-Sectional Analysis in Postmenopausal Women
by
Michelle Hernandez
A Thesis Presented to the
FACULTY OF THE KECK SCHOOL OF MEDICINE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(BIOSTATISTICS)
DECEMBER 2024
ii
TABLE OF CONTENTS
List of Tables ............................................................................................................................ iii
List of Figures ...........................................................................................................................iv
Abstract ......................................................................................................................................v
Chapter One: Introduction .........................................................................................................1
Chapter Two: Materials and Methods ........................................................................................4
Chapter Three: Results .............................................................................................................10
Chapter Four: Discussion .........................................................................................................39
References ................................................................................................................................44
Appendices ...............................................................................................................................47
iii
LIST OF TABLES
Table 1. Demographics ............................................................................................................11
Table 2. Correlation Table of Continuous Variables ................................................................12
Table 3. Univariate Linear Model Results for Biomarkers Composite Diary .........................15
Table 4. Multivariable Linear Model Results for Biomarkers Composite Diary
(Tot-Flush-D) ...........................................................................................................................17
Table 5. Multivariable Linear Model Results for Biomarkers Sum Diary Composite
Score (Tot-Flush-D) .................................................................................................................21
Table 6. Univariate Linear Model Results for Biomarkers Anyflush-Q Model .......................24
Table 7. Multivariable Linear Model Results for Biomarkers Anyflush-Q Model ..................26
iv
LIST OF FIGURES
Figure 1. Composite Diary Model Residuals vs Fitted ............................................................36
Figure 2. Anyflush-Q Model Residuals vs Fitted ....................................................................37
v
Abstract
Background: Hot flushes are a common menopausal symptom, but their relationship with
Alzheimer's disease (AD) risk remains unclear. This study investigated the association between
hot flushes and AD biomarkers in postmenopausal women.
Methods: We analyzed data from 633 postmenopausal women participating in the Early versus
Late Intervention Trial with Estradiol (ELITE). Hot flush frequency and severity were assessed
through questionnaires and personal diaries. Plasma levels of AD biomarkers, including
phosphorylated tau (pTau181), glial fibrillary acidic protein (GFAP), neurofilament light (NfL),
amyloid-β 40 (Aβ40), and amyloid-β 42 (Aβ42), were measured. Multiple linear regression
models were used to examine associations between hot flushes and biomarker levels, adjusting
for age, race, BMI, and reproductive period length.
Results: Mild hot flushes were associated with increased Aβ40 levels (p=0.036), while moderate
hot flushes were moderately correlated with decreased NfL levels (p=0.078). Severe hot flushes
were linked to decreased GFAP levels (p=0.0018). Significant racial differences in biomarker
levels were observed, with Black participants generally showing lower levels compared to White
participants. Age was positively associated with GFAP and NfL levels (p<0.001). Unexpectedly,
longer reproductive periods correlated with higher levels of GFAP, Aβ40, and Aβ42, and lower
Aβ42/40 ratios (p<0.05).
Conclusions: This study reveals complex relationships between hot flushes, demographic
factors, and AD biomarkers in postmenopausal women. While the observed associations were
generally small in magnitude, they provide a foundation for future research into the potential
mechanistic links between menopause and AD risk. These findings emphasize the importance of
vi
considering individual factors in assessing AD risk in postmenopausal women and may inform
strategies for early detection and prevention.
1
Chapter One: Introduction
Hot flushes (HFs) are a common and often distressing symptom experienced by women during
the menopausal transition and postmenopausal periods. They are characterized by sudden
feelings of warmth, usually intense over the face, neck, and chest, often accompanied by
sweating and a subsequent chill [1]. While the exact mechanisms underlying HFs are not fully
understood, they are believed to result from changes in the thermoregulatory centers of the brain
due to fluctuating estrogen levels [2,3,4]. More than 80% of women experience hot flushes
during menopause, with prevalence varying across racial groups [1, 14].
Alzheimer's disease (AD) is a progressive neurodegenerative disorder and the most common
cause of dementia [7]. It is characterized by cognitive decline, memory loss, and changes in
behavior and personality. The characteristic lesions in AD are neuritic plaques and neurofibrillary
tangles in the medial temporal lobe structures and cortical areas of the brain, together with a
degeneration of neurons and synapses [7,10]. Understanding the risk factors for AD in women is
crucial, as women compose two-thirds of those with the disease [8,9]. Recent research has
suggested a possible link between menopausal symptoms, such as hot flushes, and the risk of
Alzheimer's disease [6]. This connection is particularly intriguing given that the loss of hormones
during menopause presents a risk factor for developing AD [13].
Neurological symptoms of menopause, including temperature dysregulation (hot flashes), have
been designated as potential risk factors for AD [5]. For instance, one study found that women
with hot flashes had marginally higher white matter hyperintensities, which are common findings
in AD pathology [22]. Furthermore, Thurston et al. reported that vasomotor symptoms occurring
2
during sleep were significantly associated with lower amyloid β 42/amyloid β 40 ratios in
plasma, suggestive of greater brain amyloid β pathology [6].
Several biomarkers have been identified as potential indicators of AD pathology, including
phosphorylated tau (pTau), glial fibrillary acidic protein (GFAP), amyloid-beta (Aβ40 and
Aβ42), neurofilament light (NfL), and the Aβ42/40 ratio. These biomarkers are traditionally
measured in cerebrospinal fluid (CSF), but recent advancements have made it possible to detect
them in blood plasma [23]. Plasma biomarkers offer a non-invasive approach to assess possible
AD pathology and risk for cognitive impairment and dementia [11,12]. They have shown utility
in the diagnostic evaluation of suspected AD and in identifying AD pathology in asymptomatic
individuals [15].
Amyloid-beta 42 (Aβ42) is a peptide that aggregates to form plaques, one of the hallmark brain
features of Alzheimer's disease. In Alzheimer's, CSF levels of Aβ42 are typically reduced
because the peptide is deposited in plaques in the brain, leading to lower concentrations in the
CSF. Early studies showed inconsistent results, but newer techniques have found that the
Aβ42/Aβ40 ratio in plasma can be a promising indicator of Alzheimer's [15].
Phosphorylated Tau (pTau) is a form of tau protein that has been modified by the addition of
phosphate groups, which is associated with the formation of neurofibrillary tangles in
Alzheimer's. pTau CSF levels are elevated in Alzheimer's and can help differentiate it from other
dementias. [16,17]. Neurofilament Light (NfL) is a structural protein in neurons, and its presence
in CSF and plasma is a marker of neurodegeneration. Elevated NfL levels are associated with
neurodegenerative diseases, including Alzheimer's, and correlate with cognitive deficits [18].
3
GFAP is a protein found in astrocytes that is elevated in AD due to astrogliosis, a process often
observed around amyloid-beta plaques. Studies have shown that blood levels of GFAP are higher
in individuals with AD or mild cognitive impairment compared to healthy controls [19,20,21].
Despite the growing body of research on AD biomarkers and the potential link between
menopausal symptoms and AD risk, the relationship between hot flushes and AD biomarkers
remains understudied. This gap in knowledge is significant, considering the prevalence of hot
flushes in menopausal women and the importance of early AD detection and intervention.
This study aims to explore the association between hot flushes and AD biomarkers in a cohort of
postmenopausal women who were participants in a randomized clinical trial. By investigating
this relationship, we hope to contribute to the understanding of how menopausal symptoms may
relate to AD pathology and potentially identify early indicators of AD risk in this population.
4
Chapter Two: Materials and Methods
Study Design
The Early Versus Late Intervention Trial with Estradiol (ELITE) was a placebo-controlled,
double-blind randomized trial designed to test the hypothesis that the cardiovascular effects of
postmenopausal hormone therapy vary depending on the timing of therapy initiation, specifically
whether it occurs soon after menopause or later.
This cross-sectional study embedded within the ELITE trial, used the pre-randomization data to
investigate the association between the frequency and severity of hot flushes and Alzheimer's
disease biomarkers in postmenopausal women. The study population consisted of 633 women
who were divided into two groups based on their time since menopause (menopausal status):
early postmenopausal (within 6 years of menopause) and late postmenopausal (10 or more years
since menopause).
Data on five known Alzheimer biomarkers were collected at the baseline visit: pTAU181, GFAP,
NfL, Aβ40 and Aβ42.
Participants
Eligible women were recruited from the greater Los Angeles region to participate in ELITE.
Participants were healthy post-menopausal women without diabetes and without evidence of
cardiovascular disease. Menopause was defined as having had no regular menses for at least 6
months (or who had surgically induced menopause) and a serum estradiol level lower than
25pg/mL. Main exclusion criteria included indeterminate time since menopause, history of breast
cancer, and postmenopausal hormone therapy within 1 month. Demographic information is listed
in Table 1.
5
Data Collection
Data were collected through questionnaires, clinical assessments, and blood samples at screening
or randomization visits. The following variables were recorded:
• Demographic and Clinical Data: Age, self-reported race and ethnicity, education level,
body mass index (BMI) calculated from directly measured weight and height, and
reproductive history.
• Hot Flushes: The data on hot flushes and any flushes were derived from multiple sources
within the ELITE study, specifically using Form 26, "The Women’s Health
Questionnaire," and participant daily diaries. These sources captured different aspects of
hot flush experiences:
o Hot flush and Any flush Data from ELITE Form 26:
▪ Hot flush (Hotflush-Q): This variable captures responses to the statement,
"I have hot flashes," where participants were asked how they felt at the
time of the visit and over the past few days. The response options were:
0: "No, not at all"
1: "No, not much"
2: "Yes, sometimes"
3: "Yes, definitely"
▪ Any flush (Anyflush-Q): This binary variable indicates whether the
participant experienced any hot flushes. It was derived from the above
question, with responses categorized as:
0: No hot flushes (category 0 from above)
1: Yes, experienced hot flushes (category 1-3 from above)
o Total Hot Flush Counts from Participant Diaries (Tot-Flush-D): In addition to the
questionnaire data, participants kept diaries in which they recorded daily the total
6
number of mild, moderate, and severe hot flushes experienced. This diary-based
data collection method provided a more granular and time-sensitive source of
information compared to the retrospective questionnaire responses.
▪ Participants were instructed to record the following information daily:
- Total number of mild hot flushes
- Total number of moderate hot flushes
- Total number of severe hot flushes
▪ Given the variability in the number of days for which participants
maintained their diaries, which ranged from 5 to 129 days, it was
necessary to standardize the hot flush counts to allow for meaningful
comparisons across participants. To achieve this, we calculated weekly
estimates for each category of hot flushes. This involved adjusting the
total counts based on the number of days each participant recorded in their
diary to get the estimated number of hot flushes per week, thereby
normalizing the data to reflect a consistent weekly time frame.
▪ To ensure consistency in reporting, hot flushes were categorized based on
the following definitions:
- Mild: A warm sensation without perspiration; does not disrupt
activity
- Moderate: A warm sensation accompanied by perspiration; does
not disrupt activity
- Severe: A hot sensation with perspiration that disrupts activity
7
▪ To account for the varying impact of different severity levels, a composite
diary score was constructed using a weighted approach:
Mild hot flushes: 1 point each
Moderate hot flushes: 2 points each
Severe hot flushes: 3 points each
▪ The composite diary score was calculated as follows:
- Composite Score = (1 × Total Mild) + (2 × Total Moderate) + (3 ×
Total Severe)
- This weighted scoring system allows for a more nuanced analysis
that reflects the increasing impact of hot flush severity on
participants' daily lives.
▪ Alzheimer's Disease Biomarkers: Blood samples were analyzed for levels of pTAU181,
GFAP, Aβ40, Aβ42, and NfL. The Aβ42/40 ratio was also calculated. Biomarkers were
quantitatively determined using Simoa® Neuro 4-Plex E Advantage kits and the Simoa®
Human pTau-181 Advantage V2.1 assay kit. Blood samples were diluted 4x, and each
sample was tested twice to ensure consistency. Results were included in the analysis if
the coefficient of variation across replicates was less than 25%.
▪ Reproductive Period: Length of reproductive period is quantified in years. It was
calculated by subtracting age at menarche from age at menopause to obtain the total
number of a woman’s fertile years. While preliminary analysis did not show a significant
association of reproductive period with either hot flush frequencies or intensity, the
inclusion of reproductive period as a variable in our analysis was informed by previous
8
research showing its association with cognitive outcomes. Specifically, Karim et al. used
data from ELITE to show that a longer reproductive period was positively associated
with global cognition (p=0.04) and executive functions (p=0.04) in postmenopausal
women [24]. Their study demonstrated that reproductive life events related to sex
hormones, including the length of reproductive period, were positively related to aspects
of cognition in later life, independent of other reproductive factors. By including
reproductive period in our analysis, we aim to account for its potential influence on
cognitive biomarkers while examining the relationship between hot flushes and AD
biomarkers.
Statistical Analysis
The data were merged into a comprehensive dataset for analysis. Descriptive statistics were used
to summarize the demographic and clinical characteristics of the participants. The association
between hot flushes and AD biomarkers was assessed using various statistical methods,
including non-parametric tests, regression models, and principal component analysis.
▪ Descriptive Statistics: Demographic and clinical characteristics were summarized using
means, medians, and standard deviations for continuous variables, and frequencies and
percentages for categorical variables.
▪ Correlation Analysis: Pearson and Spearman correlation coefficients were computed to
assess the unadjusted relationships between continuous variables such as age, BMI, and
AD biomarkers.
▪ Kruskal-Wallis Test: This non-parametric test was used to compare the levels of AD
biomarkers across different categorical variables, including race and levels of hot flush
9
intensity. Post-hoc analyses were conducted using Dunn's test with Bonferroni correction
to identify specific group differences.
▪ Wilcoxon Rank-Sum Test: This non-parametric test was used to compare the levels of AD
biomarkers between participants who reported any hot flushes and those who did not
from Women’s Health Questionnaire.
▪ Chi-Square and Fisher's Exact Test: These tests were used to examine associations
between categorical variables, such as race and education. Fisher's exact test was applied
when expected cell counts were low.
▪ Linear Regression Models: General Linear Models were employed to model biomarkers
as dependent outcome variables. To improve interpretability and meet the assumptions of
normality, the raw values of certain biomarkers, specifically pTau181, GFAP, and NfL,
were multiplied by 100 and then log-transformed. This transformation facilitated a more
straightforward interpretation of the results. Aβ40 and Aβ42 were normally distributed
and did not require transformation. Model fit was assessed and confirmed to be adequate.
The statistical analyses were performed using R, with packages including dplyr, ggplot2,
dunn.test, car, corrr, ggcorrplot, MASS, and multcomp. The results were considered statistically
significant at a 2-sided p-value < 0.05.
10
Chapter Three: Results
Table 1 presents the sociodemographic characteristics of participants at baseline. The analysis
included 632 women from the ELITE trial who contributed data for this study. While 643 women
were initially randomized, 10 participants did not complete the necessary assessments for
inclusion in this analysis. The table summarizes sociodemographic characteristics based on the
presence of any hot flushes, as reported on ELITE Form 26. Categorical variables are displayed
as counts and percentages, and continuous variables are reported as means and standard
deviations. Fisher's exact test was used for categorical variables, while the Wilcoxon rank-sum
test was applied to continuous variables. The study included a sample size of approximately 632
participants (with slight variations for some measures), 405 reporting hot flushes and 227
reporting no hot flushes. The majority of participants were white (around 68% in both groups),
followed by Hispanic, Black, and Asian participants. The mean (SD) Age for all participants was
60.3 (1.1), suggesting a middle-aged to older adult sample. BMI had a mean of 26.84 (SD =
1.20), indicating a generally overweight population. Those experiencing hot flushes were slightly
younger on average (59.3 years vs 63.0 years) and had a similar BMI to those not experiencing
hot flushes.
11
Table 1. Demographics
Baseline
characteristic
p-value No Hot Flushes Yes Hot Flushes
n= 226 n= 405
Race 0.035
White 153 (67.70) 278 (68.6)
Black 13 (5.75) 46 (11.4)
Hispanic 35 (15.49) 53 (13.1)
Asian 25 (11.06) 28 (6.9)
Education 0.11
8
th Grade 1 (0.44) 0 (0)
Some High School 1 (0.44) 1 (0.3)
High School Grad 11 (4.87) 8 (2.0)
Trade School 6 (2.65) 8 (2.0)
Some College 53 (23.45) 121 (29.7)
Bachelor’s Degree 67 (29.64) 105 (24.9)
Graduate 87 (28.50) 162 (40.0)
Age <0.001**
*
63.0 (7.3) 59.3 (6.3)
BMI 0.65 27.4 (5.50) 27.2 (5.4)
Reproductive
Period a
0.31 37.3 (5.5) 37.0 (5.1)
a
reproductive period in years (age at menopause – age at menarche)
Sum (%) or Mean (SD) by Any Hot Flushes taken from ELITE Form 26 (Yes or No). Sum and %
are given for categorical variables, and the mean and standard deviation are given for
continuous variables.
12
Table 2. Correlation Table of Continuous Variables
Descriptive Statistics and Pearson’s? Correlations for Continuous Variables
Variable n M SD 1 2 3 4 5 6 7 8
1. Log(Age) 63
2
4.10 0.11 —
2. Log(BMI) 63
2
3.29 0.19 0.01 —
3. Log(PTAU1
81)
62
9
2.99 0.32 0.16*
*
-0.02 —
4. Log(GFAP) 63
2
4.57 0.41 0.38*
*
-
0.18**
0.27*
*
—
5. Log(NfL) 63
2
2.47 0.43 0.45*
*
-
0.19**
0.31*
*
0.52** —
6. Aβ40 63
2
80.3
2
16.4
2
0.06 -0.07 0.19*
*
0.36** 0.33*
*
—
7. Aβ42 63
1
5.90 1.36 0 -0.08* 0.12*
*
0.22** 0.29*
* 0.67**
—
8. Aβ42/40 63
1
0.07 0.01 -0.08 -0.03 -0.07 -
0.14**
-0.01 -
0.27**
0.052*
*
—
*p < 0.05. **p < 0.01.
Table 2 presents descriptive statistics and correlations for eight continuous variables of age,
body mass index (BMI), and various AD-related biomarkers. Log-transformed variables were
used to normalize Age, BMI, PTAU181, GFAP, and NfL
Several significant correlations were observed (Table 2):
13
Interestingly, age was not significantly correlated with Aβ40 or Aβ42 levels but showed a
moderate positive correlation with GFAP (r = 0.38, p < 0.01), NfL (r = 0.45, p < 0.01), and
PTAU (r = 0.16, p < 0.01).
BMI showed negative correlations with GFAP (r = -0.18, p < 0.01), NfL (r = -0.19, p < 0.01),
and Aβ42 (r = -0.08, p < 0.05).
PTAU181 showed a strong positive correlation with the following biomarkers:
GFAP (r = 0.27, p < 0.01), NfL ( r = 0.31, p < 0.01), Aβ40 (r = 0.19, p < 0.01), and Aβ42 (r =
0.12, p < 0.01).
GFAP and NfL were strongly positively correlated (r = 0.52, p < 0.01).
GFAP also showed moderate to week positive correlations with
Aβ40 (r = 0.36, p < 0.01), Aβ42 (r = 0.22, p < 0.01), Aβ42/40 (r = -0.14, p < 0.01).
NfL showed significant correlation with Aβ40 (r = 0.33, p < 0.01) & Aβ42 (r = 0.29, p < 0.01)
and a negative correlation with Aβ42/40 (r = -0.27, p < 0.01).
Aβ40 and Aβ42 demonstrated a strong positive correlation (r = 0.67, p < 0.01).
The Aβ42/40 ratio was negatively correlated with Aβ40 (r = -0.27, p < 0.01) and weakly
positively correlated with Aβ42 (r = 0.052, p < 0.01).
Biomarker Models
Our study investigated the associations between menopausal symptoms, particularly hot flushes,
and Alzheimer’s neurological biomarkers. We analyzed the relationships between these
symptoms and NfL, GFAP, PTAU, Aβ40, Aβ42, and the Aβ42/40 ratio, while controlling for age,
14
race, length of reproductive period, and BMI. We tested separate models controlling for the same
variables but controlling for age at menopause instead of age. While we tested the inclusion of
various reproductive history variables such as having ever been pregnant, number of full-term
pregnancies, length of contraception use, age at menopause, etc., these variables were not found
to be significant or to significantly contribute to the model, with age confounding many of these
associations. Sensitivity analyses were conducted with minimal adjustments, controlling only for
age and race. These analyses did not yield substantial differences in estimates or significance
levels. Furthermore, likelihood ratio tests indicated that the more comprehensive model, which
included age, race, BMI, and length of the reproductive period, provided a significantly better fit
for the data across all biomarkers (p < 0.001).
Model Types:
▪ Anyflush Questionnaire Model (Anyflush-Q):Refers to models using the binary
"Anyflush" variable from the ELITE Form 26 questionnaire, indicating whether
participants experienced any hot flushes.
▪ Diary-Based Total Flush Model (Tot-Flush-D): Refers to models using the composite
total counts of mild, moderate, and severe hot flushes recorded in participant diaries over
a 30-day period.
15
Table 3. Univariate Linear Model Results for Biomarkers Composite Diary
This table presents the results of Multivariable analyses using linear models to explore the
correlation between 6 biomarkers & hot flushes. To enhance the interpretability of the biomarker
outcomes, the raw values for NFL, GFAP, and PTAU181 were multiplied by 100 prior to
applying a logarithmic transformation. Each parameter's estimate, exponentiated estimate,
standard error, confidence interval, and p-value are reported to assess the strength and
significance of its association.
Model Parameter Estimate SE Exp(Estimate
)
Exp(95% CI) p-value
LL UL
Univariate log(NfL)
Model
Weighted total # of
mild hot flushes a
0.10 0.20 1.11 0.74 1.65 0.62
Weighted total # of
moderate hot flushes b
-0.14 0.091 0.87 0.73 2.04 0.14
Weighted total # of
severe hot flushes c
0.067 0.060 1.07 0.95 1.20 0.27
Univariate log(GFAP)
Model
Estimate SE Exp(Estimate
)
Exp(95% CI) p-value
Weighted total # of
mild hot flushes a
4.78e-04 0.20 1.00 0.68 1.48 0.998
Weighted total # of
moderate hot flushes b
0.039 0.088 1.04 0.88 1.24 0.65
Weighted total # of
severe hot flushes c
-0.19 0.058 0.82 0.74 0.92 <0.001***
Univariate
log(PTAU181) Model
Estimate SE Exp(Estimate
)
Exp(95% CI) p-value
Weighted total # of
mild hot flushes a
-0.0036 0.16 0.996 0.73 1.36 0.97
Weighted total # of
moderate hot flushes b
0.025 0.070 1.03 0.89 1.18 0.72
Weighted total # of
severe hot flushes c
0.0097 0.046 1.01 0.92 1.11 0.83
Univariate Aβ40
Model
Estimate SE 95% CI p-value
Weighted total # of
mild hot flushes a
0.016 0.079 - 0.0094 0.32 0.037*
Weighted total # of
moderate hot flushes b
-0.015 0.035 - -0.084 0.054 0.67
16
Weighted total # of
severe hot flushes c
0.0078 0.023 - -0.038 0.053 0.74
Univariate Aβ42
Model
Estimate SE 95% CI p-value
Weighted total # of
mild hot flushes a
0.010 0.0065 - -0.0025 0.023 0.12
Weighted total # of
moderate hot flushes b
-0.0020 0.0029 - -0.0076 0.0037 0.51
Weighted total # of
severe hot flushes c
0.0014 0.0019 - -0.0023 0.0051 0.46
Univariate Aβ42/40
Ratio Model
Estimate SE 95% CI p-value
Weighted total # of
mild hot flushes a
-1.72e-05 6.25e-05 - -1.40e-04 1.06e-04 0.78
Weighted total # of
moderate hot flushes b
-1.64e-05 2.77e-05 - -7.08e-05 3.81e-05 0.56
Weighted total # of
severe hot flushes c
1.17e-05 1.83e-05 - -2.43e-05 4.76e-05 0.52
Note. CI = confidence interval; LL = lower limit; UL = upper limit.
a 1 * total # of mild hot flushes
b 2 * total # of moderate hot flushes
c 3 * total # of severe hot flushes
* p < 0.05, ** p < 0.01, *** p < 0.001
Table 3 summarizes the results of univariate linear models analyzing the relationship between six
biomarkers and the frequency of hot flushes, categorized as mild, moderate, and severe. For NfL, GFAP,
and PTAU181, raw values were scaled and log-transformed to improve interpretability. Notably, the
log(GFAP) model shows a significant negative association with severe hot flushes (p < 0.001), while the
Aβ40 model indicates a significant positive association with mild hot flushes (p < 0.05). Other
associations were not statistically significant, as indicated by their p-values.
17
Table 4. Multivariable Linear Model Results for Biomarkers Composite Diary (Tot-FlushD)
This table presents the results of Multivariable analyses using linear models to explore the
correlation between 6 biomarkers & hot flushes as self-reported in personal diaries. To enhance
the interpretability of the biomarker outcomes, the raw values for NfL, GFAP, and PTAU181
were multiplied by 100 prior to applying a logarithmic transformation. Each parameter's
estimate, exponentiated estimate, standard error, confidence interval, and p-value are reported to
assess the strength and significance of its association.
Model Parameter Estimate SE Exp(Estimate) Exp(95% CI) p-value
LL UL
Multivariable log(NfL)
Model
Weighted total # of
mild hot flushes a
0.26 0.18 1.00 0.91 1.86 0.15
Weighted total # of
moderate hot flushes b
-0.082 0.082 1.00 0.78 1.08 0.32
Weighted total # of
severe hot flushes c
0.094 0.053 1.00 0.99 1.22 0.078
log(Age) 172.61 14.51 5.48 3.91e+62 2.18e+87 <0.001***
Race
White (ref) - - -
Black -20.34 5.68 0.82 2.10e-14 1.03e-04 <0.001***
Hispanic -8.24 4.73 0.92 2.45e-08 2.83 0.08
Asian -6.65 5.60 0.93 2.16e-08 77.64 0.24
log(BMI) -37.42 8.28 0.69 4.84e-24 6.49e-10 <0.001***
Reproductive Period -0.038 0.30 1.00 0.54 1.72 0.90
Multivariable
log(GFAP) Model
Estimate SE Exp(Estimate) Exp(95% CI) p-value
Weighted total # of
mild hot flushes a
0.15 0.18 1.16 0.81 1.65 0.42
Weighted total # of
moderate hot flushes b
0.075 0.082 1.08 0.92 1.27 0.36
Weighted total # of
severe hot flushes c
-0.17 0.053 0.85 0.76 0.94 <0.01**
log(Age) 133.81 14.44 1.30e+58 6.30e+45 2.68e+70 <0.001***
Race
White (ref) - - - - -
Black -0.051 0.057 7.27e-03 0.85 1.061 0.38
18
Hispanic -0.027 0.047 0.076 0.89 1.068 0.58
Asian 0.041 0.056 73.47 0.93 1.16 0.44
log(BMI) -0.36 0.082 7.29e-16 0.59 0.82 <0.001***
Reproductive Period 0.0078 0.0030 2.28 1.00 1.01 <0.01**
Multivariable
log(PTAU181) Model
Estimate SE Exp(Estimate) Exp(95% CI) p-value
Weighted total # of
mild hot flushes a
-0.0029 0.16 0.997 0.73 1.35 0.99
Weighted total # of
moderate hot flushes b
0.057 0.071 1.06 0.92 1.22 0.42
Weighted total # of
severe hot flushes c
0.0026 0.046 1.003 0.92 1.10 0.95
log(Age) 43.10 12.49 5.22e+18 1.16e+08 2.34e+29 <0.001***
Race
White (ref) - - - - -
Black -8.65 4.88 1.75e-04 1.20e-08 2.57 0.077
Hispanic -17.15 4.07 3.55e-08 1.20e-11 1.05e-04 <0.001***
Asian -2.90 4.82 0.055 4.30e-06 705.27 0.55
log(BMI) 5.82 7.15 3.37e+02 2.67e-04 4.25e+08 0.42
Reproductive Period -0.10 0.26 0.90 0.55 1.49 0.69
Multivariable Aβ40
Model
Estimate SE Exp(Estimate) 95% CI p-value
Weighted total # of
mild hot flushes a
0.11 0.075 - -0.034 0.26 0.049*
Weighted total # of
moderate hot flushes b
-0.0012 0.034 - -0.068 0.066 0.13
Weighted total # of
severe hot flushes c
0.0046 0.022 - -0.039 0.048 0.97
log(Age) 3.86 6.03 - -7.99 15.71 0.84
Race
White (ref) - -
Black -10.00 2.36 - -14.64 -5.36 <0.001***
Hispanic -4.72 1.97 - -8.58 -0.86 0.017*
Asian -5.90 2.33 - -10.48 -1.32 0.012*
log(BMI) -3.85 3.44 - -10.61 2.91 0.26
Reproductive Period 0.70 0.12 - 0.46 0.95 <0.001***
Multivariable Aβ42
Model
Estimate SE Exp(Estimate) 95% CI p-value
Weighted total # of
mild hot flushes a
0.0078 0.0065 - -0.0049 0.020 0.23
19
Weighted total # of
moderate hot flushes b
-4.53e-04 0.0029 - -0.0062 0.0053 0.88
Weighted total # of
severe hot flushes c
0.0012 0.0019 - -0.0025 0.0049 0.51
log(Age) 0.035 0.52 - -0.98 1.05 0.95
Race
White (ref) - - - - -
Black -0.53 0.20 - -0.93 -0.13 0.0092**
Hispanic -0.38 0.17 - -0.71 -0.051 0.024*
Asian -0.23 0.20 - -0.62 0.16 0.24
log(BMI) -0.34 0.29 - -0.91 0.24 0.26
Reproductive Period 0.022 0.011 - 8.95e-04 0.042 0.041*
Multivariable Aβ42/40
Ratio Model
Estimate SE Exp(Estimate) 95% CI p-value
Weighted total # of
mild hot flushes a
-3.89e-06 6.26e-05 - -1.27e-04 1.19e-04 0.95
Weighted total # of
moderate hot flushes b
-1.17e-05 2.84e-05 - -6.75e-05 4.42e-05 0.68
Weighted total # of
severe hot flushes c
1.25e-05 1.83e-05 - -2.35e-05 4.85e-05 0.50
log(Age) -0.0042 0.0050 - -0.014 0.0056 0.40
Race
White (ref) - - - - -
Black 2.47e-03 1.96e-03 - -1.39e-03 6.32e-03 0.21
Hispanic -5.09e-04 1.63e-03 - -3.72e-03 2.70e-03 0.76
Asian 2.24e-03 1.94e-03 - -1.57e-03 6.04e-03 0.25
log(BMI) -5.82e-04 2.86e-03 - -6.20e-03 5.04e-03 0.84
Reproductive Period -3.68e-04 1.03e-04 - -5.69e-04 -1.66e-04 <0.001***
Note. CI = confidence interval; LL = lower limit; UL = upper limit.
a 1 * total # of mild hot flushes
b 2 * total # of moderate hot flushes
c 3 * total # of severe hot flushes
* p < 0.05, ** p < 0.01, *** p < 0.001
Table 4 presents multivariable linear model results examining the relationship between six
biomarkers and self-reported hot flushes, categorized by severity, in personal diaries. NfL, GFAP,
and PTAU181 values scaled and log-transformed for clarity. Statistically significant findings
include a negative association between severe hot flushes and log(GFAP) (p < 0.01), and various
20
demographic factors such as age, race, and BMI showing significant associations across different
biomarkers. Notably, age and BMI are significant predictors in multiple biomarker models, with
age showing a strong positive association and BMI a negative association. Race also plays a
significant role, with Black and Hispanic individuals showing lower biomarker levels compared
to White individuals, particularly in the Aβ40 and Aβ42 models. The reproductive period is
significantly negatively associated with the Aβ40 and Aβ42/40 ratio models.
21
Table 5. Multivariable Linear Model Results for Biomarkers Sum Diary Composite Score
(Tot-Flush-D)
This table presents the results of multivariable analyses using linear models to explore the
correlation between 6 biomarkers & hot flushes as self-reported in personal diaries. Each
parameter's estimate, exponentiated estimate, standard error, confidence interval, and p-value
are reported to assess the strength and significance of its association.
Model Parameter Estimate SE Exp(Estimate) Exp(95% CI) p-value
LL UL
Multivariable log(NfL)
Model
Diary Composite Score
a 4.08e-04 3.20e-04
1.00
1.00 1.00
0.20
log(Age) 1.72 0.14 5.59 4.21 7.42 <0.001***
Race
White (ref) - - - - -
Black -0.21 0.057 0.81 0.73 0.91 <0.001***
Hispanic -0.086 0.047 0.92 0.84 1.02 0.069 .
Asian -0.066 0.056 0.94 0.84 1.04 0.24
log(BMI) -0.37 0.083 0.69 0.59 0.81 <0.001***
Reproductive Period -1.96e-04 2.96e-03 1.00 0.99 1.01 0.95
Multivariable
log(GFAP) Model
Estimate SE Exp(Estimate) Exp(95% CI) p-value
Diary Composite Score
a
-6.09e-04 3.20e-04
1.00 0.32 3.89 0.058
log(Age) 1.31 0.14 3.69 2.78 4.90 <0.001***
Race
White (ref) - - - - -
Black -0.046 0.057 0.95 0.85 1.07 0.41
Hispanic -0.027 0.047 0.97 0.89 1.07 0.56
Asian 0.035 0.056 1.04 0.93 1.16 0.53
log(BMI) -0.36 0.083 0.70 0.59 0.82 <0.001***
Reproductive Period 8.43e-03 2.96e-03 1.01 1.00 1.01 0.0046**
Multivariable
log(PTAU181) Model
Estimate SE Exp(Estimate) Exp(95% CI) p-value
Diary Composite Score
a 2.18e-04 2.75e-04
1.00
1.00 1.00
0.43
log(Age) 0.43 0.12 1.54 1.20 1.96 <0.001***
Race
White (ref) - - - - -
22
Black -0.085 0.049 0.92 0.83 1.01 0.080
Hispanic -0.17 0.041 0.84 0.78 0.91 <0.001***
Asian -0.030 0.048 0.97 0.88 1.07 0.54
log(BMI) 0.056 0.071 1.06 0.92 1.22 0.43
Reproductive Period -1.04e-03 2.54e-03 1.00 0.99 1.00 0.68
Multivariable Aβ40
Model
Estimate SE Exp(Estimate) 95% CI p-value
Diary Composite Score
a 9.31e-03 0.013
- -0.017 0.035
0.48
log(Age) 3.09 6.01 - -8.71 14.88 0.61
Race
White (ref) - -
Black -10.07 2.36 - -14.70 -5.43 <0.001***
Hispanic -4.86 1.96 - -8.71 -1.00 0.014*
Asian -6.05 2.33 - -10.62 -1.48 0.010*
log(BMI) -3.92 3.44 - -10.68 2.83 0.25
Reproductive Period 0.71 0.12 - 0.47 0.96 <0.001***
Multivariable Aβ42
Model
Estimate SE Exp(Estimate) 95% CI p-value
Diary Composite Score
a 1.03e-03 1.14e0-3
-
-1.20e-03 3.27e-03
0.36
log(Age) -0.0066 0.51 - -1.02 1.00 0.99
Race
White (ref) - - - - -
Black -0.54 0.20 - -0.93 -0.14 0.0082**
Hispanic -0.39 0.17 - -0.72 -0.061 0.020*
Asian -0.24 0.20 - -0.63 0.15 0.23
log(BMI) -0.33 0.29 - -0.91 0.24 0.26
Reproductive Period 0.022 0.011 - 0.0015 0.043 0.035 *
Multivariable Aβ42/40
Ratio Model
Estimate SE Exp(Estimate) 95% CI p-value
Diary Composite Score
a
2.76e-06 1.10e-05 -
-1.89e-06 2.44e-05 0.80
log(Age) -4.02e-03 4.98e-03 - -1.38e-02 5.77e-03 0.42
Race
White (ref) - - - - -
Black 2.43e-03 1.96e-03 - -1.41e-03 6.28e-03 0.21
Hispanic -5.11e-04 1.63e-03 - -3.71e-03 2.69e-03 0.75
Asian 2.29e-03 1.93e-03 - -1.50e-03 6.08e-03 0.24
log(BMI) -4.85e-04 2.85e-03 - -6.09e-03 5.12e-03 0.87
23
Reproductive Period -3.68e-04 1.02e-04 - -5.69e-04 -1.68e04
<0.001***
Note. CI = confidence interval; LL = lower limit; UL = upper limit.
a The composite score is the sum of the weighted values (1* # of weekly mild hot flushes + 2 * # of weekly
moderate hot flushes + 3 * # of weekly severe hot flushes)
* p < 0.05, ** p < 0.01, *** p < 0.001
Table 5 presents multivariable linear model results examining the relationship between six
biomarkers and a composite score of hot flushes, derived from personal diaries. The composite
score is calculated by summing weighted values of mild, moderate, and severe hot flushes.
Significant findings include a strong positive association between age and biomarker levels
across multiple models, with log(Age) showing significant effects in the NfL, GFAP, and
PTAU181 models (p < 0.001). Race also significantly influences biomarker levels, with Black
individuals showing lower levels compared to White individuals in several models, particularly
in the Aβ40 and Aβ42 models. BMI is negatively associated with biomarker levels in the NfL
and GFAP models (p < 0.001). The reproductive period shows a significant positive association
in the GFAP and Aβ40 models, and a negative association in the Aβ42/40 ratio model (p <
0.001). All models had composite diary score estimates near zero or exponentiated estimates
near one, indicating non-significant associations (p > 0.05). The multivariable log(GFAP) model
was the only one with a negative estimate (-6.09e-04, exp(estimate) = 1.00), while the others had
positive estimates, suggesting positive but non-significant associations. Overall, age, race, and
BMI are key factors influencing biomarker levels in relation to hot flushes while the diary
composite score does not show significant associations with biomarker levels in any of the
models.
24
Table 6. Univariate Linear Model Results for Biomarkers Anyflush-Q Model
This table presents the results of Multivariable analyses using linear models to explore the
correlation between 6 biomarkers & hot flushes. Each parameter's estimate, exponentiated
estimate, standard error, confidence interval, and p-value are reported to assess the strength and
significance of its association.
Model Parameter Estimate SE Exp(Estimate
)
Exp(95% CI) p-value
LL UL
Univariate log(NfL)
Model
Any Flush No (ref) - - - - -
Any Flush Yes -0.11 0.035 0.89 0.83 0.96 0.0015**
Univariate log(GFAP)
Model
Estimate SE Exp(Estimate
)
Exp(95% CI) p-value
Any Flush No (ref) - - - - -
Any Flush Yes -0.062 0.034 0.94 0.88 1.01 0.070
Univariate
log(PTAU181) Model
Estimate SE Exp(Estimate
)
Exp (95% CI) p-value
Any Flush No (ref) - - - - -
Any Flush Yes -0.030 0.027 0.97 0.92 1.02 0.26
Univariate Aβ40
Model
Estimate SE Exp(Estimate
)
95% CI p-value
Any Flush No (ref) - - - - -
Any Flush Yes -0.46 1.37 - -3.14 2.22 0.74
Univariate Aβ42
Model
Estimate SE Exp(Estimate
)
95% CI p-value
Any Flush No (ref) - - - - -
Any Flush Yes -0.11 0.11 - -0.33 0.12 0.35
Univariate Aβ42/40
Ratio Model
Estimate SE Exp(Estimate
)
95% CI p-value
Any Flush No (ref) - - - - -
Any Flush Yes -8.02e-04 1.07e-03 - -2.91e-03 1.31e-03 0.46
Note. CI = confidence interval; LL = lower limit; UL = upper limit.
* p < 0.05, ** p < 0.01, *** p < 0.001
Table 6 summarizes univariate linear model results examining the association between six
biomarkers and the presence of any hot flushes, categorized as "Any Flush Yes" or "Any Flush
25
No" (reference group) using data from the ELITE Form 26. Significant findings include a
negative association between the presence of any flushes and log(NfL) levels (p = 0.0015),
indicating that individuals experiencing flushes tend to have lower NfL levels. The log(GFAP)
model shows a marginally non-significant negative association (p = 0.070). Other biomarkers,
including log(PTAU181), Aβ40, Aβ42, and the Aβ42/40 ratio, do not show statistically
significant associations with the presence of flushes, as indicated by their higher p-values.
26
Table 7. Multivariable Linear Model Results for Biomarkers Anyflush-Q Model
This table presents the results of multivariable analyses using linear models to explore the
correlation between 6 biomarkers & hot flushes as self-reported in personal diaries. Each
parameter's estimate, exponentiated estimate, standard error, confidence interval, and p-value
are reported to assess the strength and significance of its association.
Model Parameter Estimate SE Exp(Estimate) Exp(95% CI) p-value
LL UL
Multivariable log(NfL)
Model
Any Flush No (ref) - - - - -
Any Flush Yes 0.0063 0.033 1.01 0.94 1.074 0.85
log(Age) 1.74 0.15 5.67 4.25 7.56 <0.001***
Race
White (ref) - - - - -
Black -0.22 0.055 0.80 0.72 0.89 <0.001***
Hispanic -0.088 0.046 0.92 0.84 1.00 0.055 .
Asian -0.063 0.055 0.93 0.84 1.05 0.26
log(BMI) -0.38 0.082 0.68 0.58 0.80 <0.001***
Reproductive Period 2.60e-04 2.95e-03 1.00 0.99 1.01 0.93
Multivariable
log(GFAP) Model
Estimate SE Exp(Estimate) Exp(95% CI) p-value
Any flush No (ref) - - - - -
Any flush Yes 0.036 0.033 1.036 0.97 1.11 0.28
log(Age) 1.39 0.15 4.00 3.00 5.33 <0.001***
Race
White (ref) - - - - -
Black -0.072 0.055 0.93 0.84 1.04 0.19
Hispanic -0.035 0.046 0.97 0.88 1.06 0.45
Asian 0.057 0.056 1.06 0.95 1.18 0.30
log(BMI) -0.34 0.082 0.71 0.61 0.84 <0.001***
Reproductive Period 0.0080 0.0030 1.01 1.00 1.01 0.0068**
Multivariable
log(PTAU181) Model
Estimate SE Exp(Estimate) Exp(95% CI) p-value
Any Flush No (ref) - - - - -
Any Flush Yes -0.0055 0.028 1.036 0.94 1.05 0.85
log(Age) 0.40 0.13 1.54 1.16 1.90 0.0016**
Race
White (ref) - - - - -
27
Black -0.10 0.047 0.90 0.82 0.99 0.027*
Hispanic -0.18 0.039 0.84 0.77 0.90 <0.001***
Asian -0.033 0.047 0.97 0.88 1.06 0.49
log(BMI) 0.041 0.070 1.04 0.91 1.20 0.56
Reproductive Period -7.00e-04 2.52e-03 1.00 0.99 1.00 0.78
Multivariable Aβ40
Model
Estimate SE Exp(Estimate) 95% CI p-value
Any Flush No (ref) - - - - -
Any Flush Yes 0.60 1.38 - -2.11 3.31 0.66
log(Age) 4.67 6.11 - -7.32 16.66 0.45
Race
White (ref) - -
Black -10.26 2.28 - -14.74 -5.78 <0.001***
Hispanic -4.93 1.92 - -8.70 -1.16 0.011*
Asian -5.58 2.31 - -10.13 -1.04 0.016 *
log(BMI) -4.02 3.42 - -10.75 2.70 0.24
Reproductive Period 0.71 0.12 - 0.46 0.95 <0.001***
Multivariable Aβ42
Model
Estimate SE Exp(Estimate) 95% CI p-value
Any flush No (ref) - - - - -
Any flush Yes -0.046 0.12 - -0.28 0.19 0.85
log(Age) -0.031 0.53 - -1.07 1.01 0.95
Race
White (ref) - - - - -
Black -0.55 0.20 - -0.94 -0.16 0.0054**
Hispanic -0.42 0.17 - -0.74 -0.091 0.012 *
Asian -0.22 0.20 - -0.61 0.17 0.27
log(BMI) -0.41 0.30 - -0.99 0.18 0.17
Reproductive Period 0.022 0.011 - 0.0014 0.043 0.036 *
Multivariable Aβ42/40
Ratio Model
Estimate SE Exp(Estimate) 95% CI p-value
Any Flush No (ref) - - - - -
Any Flush Yes -1.10e-03 1.14e-03 - -3.35e-03 1.14e-03 0.33
log(Age) -5.85e-03 5.05e-03 - -1.58e-02 4.07e-03 0.25
Race
White (ref) - - - - -
Black 2.38e-03 1.88e-03 - -1.31e-03 6.08e-03 0.21
Hispanic -9.50e-04 1.59e-03 - -4.06e-03 2.16e-03 0.55
Asian 2.08e-03 1.91e-03 - -1.67e-03 5.83e-03 0.28
log(BMI) -1.28e-03 2.83e-03 - -6.83e-03 4.27e-03 0.65
28
Reproductive Period
-3.65e-04 1.02e-04
-
-5.65e-04
-1.66e04
<0.001***
Note. CI = confidence interval; LL = lower limit; UL = upper limit.
* p < 0.05, ** p < 0.01, *** p < 0.001
Table 7 summarizes multivariable linear model results exploring the relationship between six
biomarkers and the presence of any hot flushes, as reported in the Form 26, "The Women’s
Health Questionnaire”. Significant findings highlight that age is a strong predictor of increased
biomarker levels, particularly in the log(NfL) and log(GFAP) models, with older age correlating
with higher biomarker levels (p < 0.001). Racial differences are evident, with Black individuals
showing significantly lower biomarker levels compared to White individuals in several models,
including log(NfL), log(PTAU181), Aβ40, and Aβ42. Hispanic individuals also exhibit lower
levels in the log(PTAU181), Aβ40, and Aβ42 models. BMI is negatively associated with
biomarker levels in the log(NfL) and log(GFAP) models, indicating that higher BMI corresponds
to lower biomarker levels (p < 0.001). The reproductive period is positively associated with
biomarker levels in the Aβ40 and Aβ42 models, and negatively in the Aβ42/40 ratio model (p <
0.001). The presence of hot flushes does not significantly impact biomarker levels in most
models, suggesting that hot flush occurrence may not be a strong predictor of biomarker changes.
NfL
For the multivariable NfL models adjusted for age, race, BMI, and reproductive period, age has
the strongest association with NfL levels. Each unit increase in log-transformed age is
statistically significantly associated with an increase in NfL levels, with an exponentiated
estimate suggesting an increase of approximately 448% in the Tot-Flush-D model and 467% in
29
the Anyflush-Q model (p < 0.001 for both). This indicates a robust positive relationship between
age and NfL levels across different models.
The association between hot flushes and NfL levels was inconsistent. In the Tot-Flush-D
Multivariable model, there was essentially no change in NfL levels with mild moderate or severe
hot flushes, all exponentiated estimates were ~1.00 but this was not statistically significant (p =
0.15, p = 0.32, p = 0.078). In the Anyflush-Q Multivariable model, the presence of any flushes
was associated with a slight increase in NfL levels, but this was also not significant (p = 0.85).
Race showed significant differences in NfL levels. Black individuals had lower NfL levels
compared to White individuals, with a reduction of approximately 18% in the Tot-Flush-D model
and about 20% in the Anyflush-Q models (p < 0.001 for all). This consistent finding highlights a
significant racial disparity in NfL levels.
Hispanic and Asian individuals also tended to have lower NfL levels compared to White
individuals, but these associations were not statistically significant (p=0.08 for Hispanics, p=0.24
for Asians in the Tot-Flush-D model; p= 0.055 for Hispanics and p=0.26 for Asians in the
Anyflush-Q model).
BMI demonstrated an inverse relationship with NfL levels. Each unit increase in log-transformed
BMI corresponded to an approximate 31% reduction in NfL levels, which was statistically
significant in both the Tot-Flush-D (p < 0.001) and Anyflush-Q models (p < 0.001). This
30
unexpected association suggests that other physical characteristics may be confounding this
relationship, as higher BMI is typically linked to poorer health outcomes.
The estimates for the length of the reproductive period were very close to one and not significant
across all models, with p-values of 0.90 in the Tot-Flush-D model and 0.93 in the Anyflush-Q
model. This indicates that the reproductive period does not have a meaningful impact on NfL
levels in these analyses.
GFAP
In the GFAP multivariate models adjusted for age, race, BMI, and reproductive period, age
demonstrated a strong association with GFAP levels. Specifically, each unit increase in logtransformed age was associated with a significant increase in GFAP levels. The exponentiated
estimates suggest an increase of approximately 300% in the Anyflush-Q model, and 281% in the
Tot-Flush-D model (p < 0.001 for all models). This indicates a robust positive relationship
between age and GFAP levels across different models. Race showed some differences in GFAP
levels. Black (r= -0.072), Hispanic (r=-0.035), and Asian (r=-0.057) subjects tended to have
lower GFAP levels compared to White subjects, but these differences were not statistically
significant in the Anyflush-Q model.
BMI demonstrated an inverse relationship with GFAP levels. Each unit increase in logtransformed BMI corresponded to an approximate 30% reduction in GFAP levels, which was
statistically significant across all models (p < 0.001). This unexpected association suggests that
other physical characteristics may be confounding this relationship, as higher BMI is typically
31
linked to poorer health outcomes. The length of the reproductive period was also found to be
statistically significantly associated with increased GFAP levels in all models. Although the
estimates were small, each unit increase in the reproductive period was associated with
approximately a 0.78% increase in GFAP levels (p = 0.0086) in the Tot-Flush-D model, a 0.81%
increase in the Anyflush-Q model (p = 0.0068).
The relationship between hot flushes and GFAP levels was inconsistent. The only statistically
significant finding was observed with moderate hot flushes, where each unit increase was
associated with approximately a 0.46% decrease in GFAP levels (p = 0.0023). The Anyflush-Q
Multivariable model had a near 0 estimate for yes any hotflushes compared to no hot flushes and
non-signficant association, (p=0.28).
PTAU181
In the PTAU181 multivariate models adjusted for age, race, BMI, and reproductive period, age
showed a strong association with PTAU levels. Specifically, each unit increase in logtransformed age was associated with a significant increase in PTAU levels. The exponentiated
estimates suggest an increase of approximately 54% in the Anyflush-Q model and 54% in the
Tot-Flush-D model (p < 0.001 for both models). This indicates a robust positive relationship
between age and PTAU181 levels across different models.
Consistent with previous research, we observed differences in PTAU181 levels when accounting
for race. Most notably, Black subjects had lower PTAU levels compared to White subjects.
Specifically, being Black was associated with a 9% reduction in PTAU levels compared to White
individuals in the Anyflush-Q model (p = 0.027 for Anyflush-Q). While being Hispanic or Asian
32
also tended to be associated with lower PTAU181 levels compared to White individuals, these
associations were not statistically significant. The exception was among Hispanic subjects in the
Tot-Flush-D model, where being Hispanic was associated with a 16% decrease in PTAU181
levels compared to White subjects (p < 0.001).
BMI and the length of the reproductive period were not found to be statistically significantly
associated with PTAU181 levels, although there was a small trend indicating that higher BMI
might be associated with higher PTAU181 levels (3.5-5.99% increase in PTAU181 levels for
each unit increase in BMI, p > 0.05).
The hot flushes variables showed inconsistent associations with PTAU levels, with some
estimates being positive and others negative. However, all estimates were small (ranging from
0.01% to 4.74%), and none of the associations reached statistical significance (p > 0.05).
Aβ40
Across all Aβ40 multivariate models adjusted for age, race, BMI, and reproductive period, the
reproductive period showed a strong association with Aβ40 levels. In all models, the
reproductive period is positively associated with Aβ40 levels, with an estimated increase of
approximately 0.70 to 0.71 units for each additional unit increase in the reproductive period (p <
0.001 for all models). This indicates a consistent positive relationship between the length of the
reproductive period and Aβ40 levels.
33
Consistent with previous research, our analysis revealed significant differences in Aβ40 levels
when accounting for race. Most notably, Black subjects had lower Aβ40 levels compared to
White individuals. Specifically, being Black was associated with an 18.26% reduction in Aβ40
levels in the Tot-Flush-D data and a 10.02% decrease in the original Aβ40 variable compared to
the reference race category. These associations were statistically significant, with p-values <
0.001.
Similarly, Hispanic and Asian subjects also exhibited trends toward lower Aβ40 levels compared
to White individuals. In the Anyflush-Q model, being Hispanic was linked to a 4.93% decrease
in Aβ40 levels (p = 0.011) and being Asian was linked to a 5.58% decrease (p = 0.016). In the
diary hot flush data, being Hispanic was associated with a 4.85% decrease (p = 0.014) and being
Asian with a 6.04% decrease in Aβ40 levels (p = 0.0097), compared to White subjects.
BMI was found to have a decreasing trend with Aβ40 levels, but this association was not
statistically significant (p=0.26 for Tot-Flush-D model; p=0.24 for Anyflush-Q model). Age
showed a slight positive association with Aβ40 levels across the models. For each unit increase
in log-transformed age, Aβ40 levels tended to increase slightly, with estimates suggesting
increases of approximately 3.09% to 4.67%. However, these increments were small and not
statistically significant in the models (p=0.25 for Tot-Flush-D model; p=0.45 for Anyflush-Q
model).
The multivariable Aβ40 model shows that the associations between hot flushes (mild, moderate,
severe, or any) and Aβ40 levels are generally small and not statistically significant. For mild hot
34
flushes, the estimate is 0.022 with a p-value of 0.14, indicating a slight positive association,
while moderate and severe hot flushes have negligible estimates of -0.000841 (p=0.90) and -
0.00149 (p=0.79), respectively. Similarly, the presence of any flushes is associated with a nonsignificant increase in Aβ40 levels, with an estimate of 0.60 (p=0.66), reflecting wide confidence
intervals and substantial uncertainty in these estimates.
Aβ42
The model results for Aβ42 closely mirrored those for Aβ40. Consistent with previous research,
significant differences in Aβ42 levels were observed by race. Most notably, Black subjects
exhibited lower Aβ42 levels compared to White individuals. Specifically, being Black was
associated with a statistically significant 55% reduction in Aβ42 levels in the Anyflush-Q model
compared to White subjects (p=0.0054), indicating statistical significance. Hispanic subjects also
showed a trend toward lower Aβ42 levels compared to White individuals. Being Hispanic was
associated with a 42% decrease in Aβ42 levels in the Anyflush-Q model (p = 0.012) compared to
White subjects.
BMI and age were not significantly associated with Aβ42 levels, showing only minor decreases
that were not statistically significant. The length of the reproductive period, however, was found
to be significantly associated with Aβ42 levels. Each year increase in the reproductive period
was associated with approximately a 2% increase in Aβ42 levels across all models, this is
statically significant ( p=0.035 in the Tot-Flush-D model; p=0.036 in Anyflush-Q model).
35
The analysis of the multivariate Aβ42 model shows that the presence and severity of hot flushes
have minimal and statistically insignificant effects on Aβ42 levels. Specifically, the estimate for
mild hot flushes is 1.35e-03 (p = 0.28), for moderate hot flushes is -1.75e-04 (p = 0.77), and for
severe hot flushes is -3.29e-05 (p = 0.95). Additionally, the estimate for having any flushes is -
0.046 (p=0.85). All these p-values exceed 0.05, and the confidence intervals include zero,
indicating that these associations are not statistically significantly associated with Aβ42 levels.
Aβ42/40 RATIO
We examined the Aβ42/40 ratio, as current scientific literature recommends assessing this ratio
in conjunction with individual Aβ40 and Aβ42 levels. Our analysis revealed that the reproductive
period was the only variable consistently and statistically significantly associated with the Aβ
ratio across all models. For each year increase in reproductive period, there was a decrease of
approximately 0.000365 in the Aβ ratio (p < 0.001 in all models). Notably, the HF variable in the
Anyflush-Q model did not show statistically significant associations with the Aβ ratio after
controlling for age, BMI, reproductive period, and race. While the association between
reproductive period and Aβ ratio is statistically significant and the effect size per year is small
relative to the overall range of the ratio, the reproductive period spans many years. This means
the cumulative effect could be more meaningful. Therefore, while there may be a relationship
between reproductive period and Aβ ratio, other factors not captured in our models likely play a
more substantial role in determining this ratio.
Model Diagnostics
36
To ensure the validity and reliability of our statistical analyses, we conducted comprehensive
model diagnostics for each of our primary outcome measures. We focused on the composite
diary models and the Anyflush-Q models for each of the AD biomarkers: Log(GFAP),
Aβ42/Aβ40 Ratio, Log(NfL), and Log(PTAU181). Residuals vs Fitted plot are listed.
Composite Diary Model Diagnostics
Figure 1. Composite Diary Model Residuals vs Fitted
The diagnostic plots for NfL, GFAP, PTAU181, Aβ40, Aβ42, and Aβ42/40 Ratio generally
indicate reasonable model fits, with residuals scattered around zero for all variables. The spread
of residuals is fairly consistent across fitted values for most plots, suggesting good
homoscedasticity, particularly for NfL, GFAP, and PTAU181. There are no clear non-linear
patterns, indicating that linear model assumptions are generally met. However, Aβ40 and Aβ42
plots show more spread in residuals and potential outliers, suggesting higher variability in these
37
models' predictions. The Aβ42/40 ratio plot has a narrow range of fitted values but wide spread
of residuals, indicating potential difficulty in accurate prediction. Overall, while the models
perform adequately, there is room for improvement, especially for Aβ40 and Aβ42 models,
where further investigation or refinement could be beneficial to address the higher variability and
potential outliers.
Anyflush-Q Model Diagnostics
Figure 2. Anyflush-Q Model Residuals vs Fitted
The residuals vs. fitted plots for the biomarkers NfL, GFAP, PTAU181, Aβ40, Aβ42, and the
Aβ42/40 ratio generally show a random scatter of points around the horizontal line, indicating
that the models fit the data reasonably well without obvious patterns or trends. However, the plot
38
for Aβ40 shows a slightly wider spread of residuals, suggesting potential heteroscedasticity or
outliers that might affect the model's assumptions. Overall, while most plots suggest a good fit,
the Aβ40 plot may require further investigation to ensure model robustness.
39
Part V. Discussion
This study investigated the associations between menopausal symptoms, particularly hot flushes,
and Alzheimer's disease (AD) biomarkers in a cohort of postmenopausal women. Our findings
reveal complex relationships between hot flushes, demographic factors, and AD biomarkers,
contributing to the growing body of research on the potential links between menopausal
symptoms and neurological health.
In examining the basic presence or absence of hot flushes (Anyflush-Q model), we found that
experiencing hot flushes was not significantly associated with most AD biomarkers after
adjusting for demographic and reproductive factors. The presence of hot flushes showed minimal
associations with NfL (1% increase, p=0.85), GFAP (3.6% increase, p=0.28), PTAU181 (-0.55%
decrease, p=0.85), Aβ40 (0.60 unit increase, p=0.66), and Aβ42 (-0.046 unit decrease, p=0.85).
These findings suggest that the mere occurrence of hot flushes may not substantially impact AD
biomarker levels.
However, when examining the severity-specific associations (Tot-Flush-D model), we observed
more nuanced relationships. Mild hot flushes were significantly associated with increased Aβ40
levels (p < 0.05), while severe hot flushes showed a significant negative association with GFAP
levels (p < 0.001). Moderate hot flushes demonstrated a trend toward decreased NfL levels,
though this association did not reach statistical significance. These differential associations
suggest that the relationship between hot flushes and AD biomarkers may be more complex than
a simple presence/absence effect, with different severities of hot flushes potentially reflecting
distinct underlying physiological processes. However, it's important to note that these
40
associations were not consistently significant across all models, which may be due to our
moderate sample size or the complexity of the underlying biological mechanisms.
Our study also highlighted significant racial differences in biomarker levels, with Black
participants generally showing lower levels of AD biomarkers compared to White participants.
These racial disparities underscore the need for culturally sensitive approaches in both research
and clinical practice related to menopause and cognitive health. The observed differences may be
influenced by genetic, environmental, or socio-economic factors, warranting further investigation
to understand their underlying causes.
Age emerged as a strong predictor of several AD biomarkers, particularly GFAP and NfL, with
older age associated with higher levels of these biomarkers. This finding aligns with previous
research indicating that age is a significant risk factor for AD and related neurodegenerative
processes. However, the association between age and the levels of Aβ40 or Aβ42 is weaker
compared to the association between age and other biomarkers like GFAP and NfL., suggesting
that age may influence different aspects of AD pathology to varying degrees.
Interestingly, we found an inverse relationship between BMI and several AD biomarkers,
including NfL and GFAP. This finding contrasts with some previous studies that have linked
higher BMI to increased AD risk. However, the relationship between BMI and AD is complex,
particularly in older adults. While high BMI in midlife is consistently associated with increased
risk for dementia, including AD, there is an "obesity paradox" in late life where high BMI may
be associated with lower risk of developing dementia [27]. This paradox underscores the need for
41
further research to elucidate the interactions between body composition, menopausal symptoms,
and AD biomarkers across different life stages.
Our study revealed significant correlations between the duration of the reproductive period and
several key biomarkers, including GFAP, Aβ40, Aβ42, and the Aβ42/40 ratio. Intriguingly, these
findings present a paradox when compared to the existing literature on estrogen exposure and
cognitive health. Contrary to expectations, we observed positive associations between the length
of the reproductive period and levels of GFAP, Aβ40, and Aβ42, alongside a negative association
with the Aβ42/40 ratio. These results seem to challenge the widely accepted notion that longer
reproductive spans, which imply greater cumulative estrogen exposure, correlate with improved
cognitive outcomes and reduced dementia risk [26].
However, our biomarker findings can be viewed alongside the cognitive outcomes previously
reported from the ELITE trial [24]. In that analysis, which examined the same cohort of
postmenopausal women, Karim et al. found that a longer reproductive period was significantly
associated with better global cognition (p=0.04) and executive functions (p=0.04), even after
controlling for age, race/ethnicity, income, and education. This apparent contradiction between
biomarker levels and cognitive performance in relation to reproductive period length is
particularly interesting. While we found that longer reproductive periods correlate with higher
levels of AD-associated biomarkers, the same reproductive exposure appears to benefit cognitive
performance. This suggests that the relationship between reproductive period, hormonal
exposure, and brain health may be more complex than previously understood. The reproductive
period reflects the duration of exposure to premenopausal levels of endogenous sex steroid
42
hormones, and while this exposure appears to influence both biomarker levels and cognitive
function, the mechanisms linking these outcomes may involve different pathways or
compensatory processes that warrant further investigation.
These unexpected results highlight the potential complexity of the relationship between
reproductive period, estrogen exposure, and neurological biomarkers. It's possible that this
relationship involves intricate, non-linear interactions that are not yet fully expounded.
Furthermore, the impact of estrogen on the brain might vary depending on factors such as life
stage, timing of exposure, or other physiological variables not accounted for in this study.
While our study provides valuable insights into the relationships between menopausal symptoms
and AD biomarkers, it's important to note that the observed associations were generally small in
magnitude. This suggests that while menopausal symptoms may contribute to neurological
changes, other factors likely play substantial roles in determining AD biomarker levels and, by
extension, AD risk.
The strengths of our study include its large cohort, comprehensive assessment of multiple AD
biomarkers, and detailed characterization of menopausal symptoms. However, our study also has
limitations. Its cross-sectional nature precludes causal inferences, and the self-reported nature of
hot flush data may introduce some bias. Additionally, while we controlled for several potential
confounders, residual confounding cannot be ruled out. The explanatory power of our models
varied, indicating that other unmeasured factors may significantly influence biomarker levels.
43
Furthermore, our cohort had a majority of white participants, limiting the generalizability of our
findings to more diverse populations.
In conclusion, our findings suggest a complex interplay between menopausal symptoms,
particularly hot flushes, and AD biomarkers. The observed associations, while generally small,
provide a foundation for future research into the potential mechanistic links between menopause
and AD risk. Longitudinal studies are needed to further elucidate these relationships and their
implications for women's cognitive health across the menopausal transition and beyond. Our
results also emphasize the importance of considering individual factors such as race, age, and
BMI in assessing AD risk in postmenopausal women. These findings may inform future
strategies for early detection and prevention of AD in postmenopausal women.
44
References
1. Bansal R, Aggarwal N. Menopausal Hot Flashes: A Concise Review. J Midlife Health.
2019 Jan-Mar;10(1):6-13. doi: 10.4103/jmh.JMH_7_19. PMID: 31001050; PMCID:
PMC6459071.
2. Deecher DC, Dorries K. Understanding the pathophysiology of vasomotor symptoms (hot
flushes and night sweats) that occur in perimenopause, menopause, and postmenopause
life stages. Arch Womens Ment Health. 2007;10(6):247-57. doi: 10.1007/s00737-007-
0209-5. Epub 2007 Dec 12. PMID: 18074100.
3. Thurston RC, Joffe H. Vasomotor symptoms and menopause: findings from the Study of
Women's Health across the Nation. Obstet Gynecol Clin North Am. 2011 Sep;38(3):489-
501. doi: 10.1016/j.ogc.2011.05.006. PMID: 21961716; PMCID: PMC3185243.
4. Freedman, R. R. (2014). Menopausal hot flashes: Mechanisms, endocrinology, treatment.
The Journal of Steroid Biochemistry and Molecular Biology, 142, 115-120.
https://doi.org/10.1016/j.jsbmb.2013.08.010
5. Scheyer, O., Rahman, A., Hristov, H., Berkowitz, C., Isaacson, R. S., Brinton, R. D., &
Mosconi, L. (2018). Female sex and Alzheimer’s risk: The menopause connection.
Journal of Prevention of Alzheimer's Disease, 5(4), 225-230.
https://doi.org/10.14283/jpad.2018.34
6. Thurston, R. C., Maki, P., Chang, Y., Wu, M., Aizenstein, H. J., Derby, C. A., & Karikari,
T. K. (2024). Menopausal vasomotor symptoms and plasma Alzheimer disease
biomarkers. American Journal of Obstetrics & Gynecology, 342(e8).
7. Blennow, K., de Leon, M. J., & Zetterberg, H. (2006). Alzheimer’s disease. The Lancet,
368(9533), 387–403.
8. Rajan KB, Weuve J, Barnes LL, McAninch EA, Wilson RS, Evans DA. Population
estimate of people with clinical Alzheimer’s disease and mild cognitive impairment in the
United States (2020-2060). Alzheimers Dement 2021;17: 1966–75.
9. Ferretti MT, Iulita MF, Cavedo E, et al. Sex differences in Alzheimer disease - the
gateway to precision medicine. Nat Rev Neurol 2018;14: 457–69.
10. Masters CL, Simms G, Weinman NA, Multhaup G, McDonald BL, Beyreuther K.
Amyloid plaque core protein in Alzheimer disease and Down syndrome. Proc Natl Acad
Sci USA 1985; 82: 4245–49.
11. Chong MS, Sahadevan S. Preclinical Alzheimer’s disease: diagnosis and prediction of
progression. Lancet Neurol 2005; 4: 576–79.
45
12. 95 Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K, Minthon L. Association
between CSF biomarkers and incipient Alzheimer’s disease in patients with mild
cognitive impairment: a follow-up study. Lancet Neurol 2006; 5: 228–34.
13. Congdon, E. E. (2018). Sex differences in autophagy contribute to female vulnerability in
Alzheimer’s disease. Frontiers in Neuroscience, 12, Article 372.
https://doi.org/10.3389/fnins.2018.00372
14. Gold, E. B., Colvin, A., Avis, N., Bromberger, J., Greendale, G. A., Powell, L., Sternfeld,
B., & Matthews, K. (2006). Longitudinal analysis of the association between vasomotor
symptoms and race/ethnicity across the menopausal transition: Study of Women’s Health
Across the Nation. American Journal of Public Health, 96(7), 1226-1235.
https://doi.org/10.2105/AJPH.2005.066936
15. Gerards, M., Schild, A.-K., Meiberth, D., Rostamzadeh, A., Vehreschild, J. J., WingenHeimann, S., Johannes, W., Martino Adami, P., Onur, O. A., Ramirez, A., Karikari, T. K.,
Ashton, N. J., Zetterberg, H., Blennow, K., Maier, F., & Jessen, F. (2022). Alzheimer’s
disease plasma biomarkers distinguish clinical diagnostic groups in memory clinic
patients. Dementia and Geriatric Cognitive Disorders, 51(182–192).
https://doi.org/10.1159/000524390
16. Janelidze S, Mattsson N, Palmqvist S, Smith R, Beach TG, Serrano GE, et al. Plasma Ptau181 in Alzheimer’s disease: relationship to other biomarkers, differential diagnosis,
neuropa- thology and longitudinal progression to Alz- heimer’s dementia. Nat Med. 2020
Mar;26(3): 379–86.
17. Karikari TK, Pascoal TA, Ashton NJ, Janelid- ze S, Benedet AL, Rodriguez JL, et al.
Blood phosphorylated tau 181 as a biomarker for Alzheimer’s disease: a diagnostic
perfor- mance and prediction modelling study using data from four prospective cohorts.
Lancet Neurol. 2020 May;19(5):422–33.
18. Mattsson N, Andreasson U, Zetterberg H, Blennow K; Alzheimer’s Disease Neuroimaging Initiative. Association of plasma neuro- filament light with neurodegeneration in patients with Alzheimer disease. JAMA Neurol. 2017 May 1;74(5):557–66.
19. Kim KY, Shin KY, Chang KA. GFAP as a Potential Biomarker for Alzheimer's Disease:
A Systematic Review and Meta-Analysis. Cells. 2023 May 4;12(9):1309. doi:
10.3390/cells12091309. PMID: 37174709; PMCID: PMC10177296.
20. Simpson J.E., Ince P.G., Lace G., Forster G., Shaw P.J., Matthews F., Savva G., Brayne
C., Wharton S.B., Function M.R.C.C., et al. Astrocyte phenotype in relation to
Alzheimer-type pathology in the ageing brain. Neurobiol. Aging. 2010;31:578–590. doi:
10.1016/j.neurobiolaging.2008.05.015.
21. Oeckl P., Halbgebauer S., Anderl-Straub S., Steinacker P., Huss A.M., Neugebauer H.,
von Arnim C.A.F., Diehl-Schmid J., Grimmer T., Kornhuber J., et al. Glial Fibrillary
46
Acidic Protein in Serum is Increased in Alzheimer’s Disease and Correlates with
Cognitive Impairment. J. Alzheimers Dis. 2019;67:481–488. doi: 10.3233/JAD-180325.
22. Thurston RC, Aizenstein HJ, Derby CA, Sejdić E, Maki PM. Menopausal hot flashes and
white matter hyperintensities. Menopause. 2016 Jan;23(1):27-32. doi:
10.1097/GME.0000000000000481. PMID: 26057822; PMCID: PMC4673042.
23. Gerards, M., Schild, A.-K., Meiberth, D., Rostamzadeh, A., Vehreschild, J. J., WingenHeimann, S., Johannes, W., Martino Adami, P., Onur, O. A., Ramirez, A., Karikari, T. K.,
Ashton, N. J., Zetterberg, H., Blennow, K., Maier, F., & Jessen, F. (2022). Alzheimer’s
disease plasma biomarkers distinguish clinical diagnostic groups in memory clinic
patients. Dementia and Geriatric Cognitive Disorders, 51(182–192).
https://doi.org/10.1159/000524390
24. Karim, R., Dang, H., Henderson, V. W., Hodis, H. N., St John, J., Brinton, R. D., &
Mack, W. J. (2016). Impact of reproductive history and exogenous hormone use on
cognitive function in midlife and late life. Journal of the American Geriatrics Society,
64(12), 2448–2456. https://doi.org/10.1111/jgs.14658
25. Hodis, H. N., Mack, W. J., Henderson, V. W., Shoupe, D., Budoff, M. J., Hwang-Levine,
J., Li, Y., Feng, M., Dustin, L., Kono, N., Stanczyk, F. Z., Selzer, R. H., & Azen, S. P.
(2016). Vascular effects of early versus late postmenopausal treatment with estradiol. The
New England Journal of Medicine, 374(13), 1221–1231.
https://doi.org/10.1056/NEJMoa1505241
26. Jett, S., Malviya, N., Schelbaum, E., Jang, G., Jahan, E., Clancy, K., Hristov, H.,
Pahlajani, S., Niotis, K., Loeb-Zeitlin, S., Havryliuk, Y., Isaacson, R., Brinton, R. D., &
Mosconi, L. (2022). Endogenous and exogenous estrogen exposures: How women's
reproductive health can drive brain aging and inform Alzheimer's prevention. Frontiers in
Aging Neuroscience, 14, 831807. https://doi.org/10.3389/fnagi.2022.831807
27. Hayes, J. P., Moody, J. N., Guzmán Roca, J., & Hayes, S. M. (2020). Body mass index is
associated with smaller medial temporal lobe volume in those at risk for Alzheimer's
disease. NeuroImage: Clinical, 25, 102156. https://doi.org/10.1016/j.nicl.2019.102156
47
Appendices
Significant Relationships
Race vs. Continuous Variables
Our analysis revealed several significant associations between race and continuous variables.
Age varied significantly across different races (p = 0.012), indicating demographic differences in
age distribution. BMI showed a strong relationship with race (p <0.0001), highlighting racial
disparities in body mass index (graph 1).
Race was significantly associated with all the 5 biomarkers (graphs 2-6), excluding Aβ42/40
ratio. Notably, PTAU181 (p < 0.0001), GFAP (p = 0.007), Aβ40 (p < 0.0001), Aβ42 (p < 0.0001),
and NfL (p < 0.0001) levels were all significantly associated with race, suggesting that these
Alzheimer's disease biomarkers exhibit considerable racial variation (Graphs 2-7). In general,
Black participants showed overall lower levels of each of the biomarkers compared to White
individuals, followed by Hispanic and then Asian individuals.
Education vs. Continuous Variables
Education was not significantly associated with any continuous variables, indicating that
educational attainment does not directly influence biomarker levels. However, education and
race were significantly associated (p = 0.0035), suggesting that educational disparities exist
among different racial groups.
Anyflush vs. Biomarkers
Total ELITE Sample:
48
- NfL: Significant association observed, with a p-value of 0.0039, indicating a strong link
between NfL levels and the presence of any hot flushes. (Graph 8)
- GFAP: Marginally significant with a p-value of 0.06, suggesting a potential relationship
that might need further investigation. (Graph 9)
- Other biomarkers (PTAU181, Aβ40, Aβ42, Aβ_ratio) showed no significant association
with hot flushes in the general analysis.
Specific Racial Categories:
White:
- NfL: Marginal significance observed with a p-value of 0.050, highlighting a potential link
between NfL levels and hot flushes.
Hispanic:
- PTAU181: Approaches significance with a p-value of 0.072, indicating a possible
association.
- Aβ42: Shows a lower p-value of 0.1034, suggesting a weaker potential relationship.
Asian:
- GFAP: Exhibits a relatively low p-value of 0.15, which could hint at an association worth
exploring further.
Hotflush vs. Continuous Variables
General Results Across All Races:
The Kruskal-Wallis test for hot flush (categorical 1-4) demonstrated a significant relationship
with NfL levels (p = 0.037). The Dunn test with Bonferroni correction revealed a statistically
significant difference (p ≤ 0.05) between "No not at all" and "Yes definitely" groups (p =
49
0.0472). Contrary to initial expectations, the positive Z-score (2.42) indicated that the "No not at
all" group has higher NfL levels compared to the "Yes definitely" group. This suggests that
individuals who don't experience hot flushes at all have significantly higher log_nfl levels than
those who experience them. This unexpected finding warrants further investigation, as it
contradicts the hypothesis that higher NfL levels would be associated with more severe hot
flushes.
Results by Racial Group:
White: (n=432)
-Aβ40: Approaches significance, indicating potential variability in Aβ40 levels among
individuals with different hot flush statuses within the White population (p = 0.12).
-Other biomarkers did not show significant differences.
Asian: (n=53)
-Aβ42: Approaches significance, suggesting potential variability in Aβ42 levels among
individuals with different hot flush statuses within the Asian population (p = 0.051).
Black and Hispanic: No significant associations found across all tested biomarkers.
Fisher's Exact Test revealed significant associations between race and education (p = 0.004), race
and any flush (p = 0.04). These findings highlight the interplay between race, education, and
menopausal symptoms, suggesting that racial and educational disparities influence the
experience of hot flushes.
50
Graph 1. Significant Dunn's Test results are shown
51
Graph 2-7. Boxplots of biomarkers by Race. Significant Dunn's Test results are shown.
52
53
Graph 8-9.
Race-specific analyses revealed a near-significant relationship between Aβ42 and hot flush in
Asian individuals (p = 0.051). However, there were no other significant relationships observed
between hot flush and the other biomarkers for the other race groups (White, Black, Hispanic).
54
Graph 10-11.
55
56
The Wilcoxon rank sum test for any flush (0 or 1) indicated significant associations with NfL
(Wilcoxon W = 52110, p = 0.0039) and marginally significant associations with GFAP (p =
0.06).
Race-specific findings showed significant associations for NfL in White individuals (p=0.049)
and marginally significant associations for PTAU181 in Hispanic individuals (p=0.072), while
Black and Asian individuals showed no significant associations.
57
Abstract (if available)
Abstract
Background: Hot flushes are a common menopausal symptom, but their relationship with Alzheimer's disease (AD) risk remains unclear. This study investigated the association between hot flushes and AD biomarkers in postmenopausal women. Methods: We analyzed data from 633 postmenopausal women participating in the Early versus Late Intervention Trial with Estradiol (ELITE). Hot flush frequency and severity were assessed through questionnaires and personal diaries. Plasma levels of AD biomarkers, including phosphorylated tau (pTau181), glial fibrillary acidic protein (GFAP), neurofilament light (NfL), amyloid-β 40 (Aβ40), and amyloid-β 42 (Aβ42), were measured. Multiple linear regression models were used to examine associations between hot flushes and biomarker levels, adjusting for age, race, BMI, and reproductive period length.
Results: Mild hot flushes were associated with increased Aβ40 levels (p=0.036), while moderate hot flushes were moderately correlated with decreased NfL levels (p=0.078). Severe hot flushes were linked to decreased GFAP levels (p=0.0018). Significant racial differences in biomarker levels were observed, with Black participants generally showing lower levels compared to White participants. Age was positively associated with GFAP and NfL levels (p<0.001). Unexpectedly, longer reproductive periods correlated with higher levels of GFAP, Aβ40, and Aβ42, and lower Aβ42/40 ratios (p<0.05).
Conclusions: This study reveals complex relationships between hot flushes, demographic factors, and AD biomarkers in postmenopausal women. While the observed associations were generally small in magnitude, they provide a foundation for future research into the potential mechanistic links between menopause and AD risk. These findings emphasize the importance of considering individual factors in assessing AD risk in postmenopausal women and may inform strategies for early detection and prevention.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
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
Associations between physical activities with bone mineral density in postmenopausal women
PDF
Long-term blood pressure variability across the clinical and biomarker spectrum of Alzheimer’s disease
PDF
Effect of estradiol on circulating levels of inflammatory cytokines in postmenopausal women
PDF
Bone mineral density is associated with carotid atherosclerosis in healthy postmenopausal women: a longitudinal analysis of randomized clinical trial data
PDF
Effects of post-menopausal hormone therapy on arterial stiffness in the ELITE trial
PDF
Associations between inflammatory markers and change in cognitive endpoints
PDF
The relationship of resistin, leptin, adiponectin, and ghrelin with bone mineral density in healthy postmenopausal women: longitudinal analysis
PDF
Relationship of blood pressure and antihypertensive medications to cognitive change in the BVAIT, WISH, and ELITE clinical trials
PDF
Relationship between progression of atherosclerosis and coagulation measures in a randomized-controlled trial
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
A biodemographic approach to understanding sociodemographic disparities in kidney functioning on three dimensions: individual, population, and cross-national
Asset Metadata
Creator
Hernandez, Michelle
(author)
Core Title
Exploring the relationship between menopausal hot flushes and Alzheimer's disease biomarkers: a cross-sectional analysis in postmenopausal women
School
Keck School of Medicine
Degree
Master of Science
Degree Program
Biostatistics
Degree Conferral Date
2024-12
Publication Date
12/06/2024
Defense Date
12/02/2024
Publisher
Los Angeles, California
(original),
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
aging,Alzheimer's disease,amyloid-beta,biomarkers,cognitive health,GFAP,hot flushes,Menopause,neurofilament light,OAI-PMH Harvest,phosphorylated tau,postmenopausal women,racial differences,reproductive period,vasomotor symptoms
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Mack, Wendy (
committee chair
), Hodis, Howard (
committee member
), Karim, Roksana (
committee member
)
Creator Email
mdh01908@usc.edu,mhern118@gmail.com
Unique identifier
UC11399EHF9
Identifier
etd-HernandezM-13674.pdf (filename)
Legacy Identifier
etd-HernandezM-13674
Document Type
Thesis
Format
theses (aat)
Rights
Hernandez, Michelle
Internet Media Type
application/pdf
Type
texts
Source
20241210-usctheses-batch-1227
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
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
Tags
Alzheimer's disease
amyloid-beta
biomarkers
cognitive health
GFAP
hot flushes
neurofilament light
phosphorylated tau
postmenopausal women
racial differences
reproductive period
vasomotor symptoms