Menopausal Vasomotor Symptoms and Plasma Alzheimer’s Disease Biomarkers

Abstract

Background:

Identifying risk factors for Alzheimer’s disease (AD) in women is important as women comprise two thirds of individuals with AD. Prior work links vasomotor symptoms (VMS), the cardinal menopausal symptom, with poor memory performance and with alterations in brain structure, function, and connectivity. These association are evident when VMS are monitored objectively with ambulatory skin conductance monitors.

Objective:

To determine whether VMS are associated with AD biomarkers.

Study Design:

Between 2017 and 2020 the MsBrain study enrolled 274 community-dwelling women aged 45–67 who had a uterus and at least one ovary and were late perimenopausal or postmenopausal. Key exclusion criteria included neurological disorder, surgical menopause, and recent use of hormonal or non-hormonal VMS treatment. Women underwent 24 hours of ambulatory skin conductance monitoring to assess VMS. Plasma concentrations of AD biomarkers including amyloid β (Aβ) 42/40 ratio, phosphorylated tau (p-tau 181 and 231), glial fibrillary acidic protein (GFAP), and neurofilament light (NfL) were measured using single molecule array (Simoa) technology. Associations between VMS and AD biomarkers were assessed via linear regression models adjusted for age, race/ethnicity, education, body mass index, apolipoprotein E4 status, and in additional models, estradiol and sleep.

Results:

A total of 248 (mean age=59.06 years, 81% white, 99% postmenopausal) of enrolled MsBrain participants contributed data. Objectively-assessed VMS occurring during sleep were associated with significantly lower Aβ42/Aβ40, [B(SE)=−.0010 (.0004), p=.018, multivariable], suggestive of greater brain Aβ pathology. Findings remained significant after additional adjustments for estradiol and sleep.

Conclusions:

Nighttime VMS may be a marker of women at risk of AD. It is yet unknown if these associations are causal.

Introduction

Understanding risk factors for Alzheimer’s disease (AD) in women is critical, as women comprise two-thirds of patients with the disease.^1,2 Although women’s longer life expectancy is a key factor contributing to this sex difference in AD risk, other factors in this sex difference have been identified, including increased risk of the disease among carriers of the apolipoprotein E ɛ4 allele (APOE ɛ4), the strongest genetic risk factor for the disease.^2,3 Sex hormones, including changes in estradiol at menopause, may also play a role in AD through effects on the brain, cardiovascular system, and metabolism.⁴

Recent work highlights the association of the cardinal symptom of menopause, vasomotor symptoms (VMS), or hot flashes and night sweats, with white matter hyperintensities (WMH),^5,6 memory performance,^7–9 carotid atherosclerosis,^10,11 endothelial dysfunction,¹² resting state functional connectivity¹³ and alterations in brain function at rest and during memory tasks.¹⁴ Notably, these associations were established using a validated method for objectively assessing VMS, ambulatory skin conductance monitoring, which does not rely on subjective recall of VMS. Monitoring is particularly useful for assessing VMS during sleep. These VMS studies point to a potentially modifiable risk factor for cognitive impairment and dementia, one that is experienced by up to 80% of women, 33% of whom continue to experience VMS into their seventh decade of life.^15,16 Here we examined the association between VMS and plasma-based biomarkers of AD, including plasma amyloid β (Aβ) 42/40 ratio (a marker of amyloid plaques), phosphorylated tau (p-tau 181 and 231; markers of tangles in the neuronal cytoskeleton), glial fibrillary acidic protein (GFAP; a marker of astroglia activation); and neurofilament light (NfL; a marker of neuronal death) in a sample of midlife women.

Materials and Methods

Sample

Women aged 45–67 were enrolled in the MsBrain Study. Women were recruited between 2017 and 2020 from the Allegheny County, PA community via advertisements in local registries, fliers and brochures placed in community locations, listservs and electronic mailing lists, and from the MsHeart study, a study of non-smoking women previously studied for menopause and cardiovascular health.¹⁷ MsBrain inclusion criteria were being aged 45–67, having a uterus and at least one ovary, and being late perimenopausal (between 2 and 12 months of amenorrhea) or postmenopausal status (12 or more months of amenorrhea).¹⁸ Exclusion criteria included a reported history of stroke or cerebrovascular accident; dementia; seizure disorder; brain tumor; Parkinson’s Disease; a history of head trauma with loss of consciousness for >60 min; metal in the body; active substance use; current chemotherapy; pregnancy; history of hysterectomy and/or bilateral oophorectomy or other procedure that resulted in permanent end of menses; intrauterine device; and current use of select medications impacting VMS, including hormone therapy (oral or transdermal estrogen and/or progesterone), selective estrogen receptor modulators, aromatase inhibitors, selective serotonin reuptake inhibitors (SSRI), or serotonin norepinephrine reuptake inhibitors (SNRI).

Of the 274 women enrolled in MsBrain, 26 women were excluded due to missing all AD biomarker or VMS data, yielding 248 participants in any one of the models. An additional 2–4 women were excluded from select models due to missing or extreme AD biomarker values [>5 standard deviations (SD) from the mean] for final samples of N=246 for NfL; N=245 for p-tau231; and N=244 for p-tau181, Aβ42/Aβ40, and GFAP models.

Design and Procedures

Participant assessments included telephone and in-person screening procedures (including urine toxicology screen), physical measurements, a medical history, questionnaires, fasting phlebotomy, and three days of ambulatory VMS and sleep actigraphy monitoring (24 hours of objective VMS and sleep monitoring. Study procedures were reviewed and approved by the University of Pittsburgh Human Research Protection Office. All participants provided written, informed consent.

Measures

Vasomotor symptoms (VMS)

Participants were equipped with a physiologic VMS monitor and VMS diary to undergo three days of VMS monitoring, the first 24 hours of which included physiologic VMS monitoring. The wearable VMS monitor (VU-AMS 5fs, VU University Amsterdam, Netherlands) quantifies VMS via sternal skin conductance, a validated objective measure of VMS.¹⁹ Participants wore the VU-AMS monitor for 24 hours, after which time they removed it and stored it in a carrying case. Women reported VMS by completing an electronic VMS diary entry and by pressing event mark buttons on the VU-AMS monitor. For all three days, women completed the VMS diary (Moto G5S Plus 4G LTE, Motorola, Chicago, IL; MovisensXS software, movisens GmbH, Karlsruhe, Germany), which provided a date- and time-stamped VMS report. After monitoring, VMS data were downloaded and scored via UFI software (DPS; Morro Bay, CA) according to validated methods that have demonstrated reliability, including in the present laboratory and for this study (ĸ=.89). Only physiologic VMS files with ≥70% of usable data were included. VMS during wake and sleep were categorized based upon actigraphy-defined wake and sleep times. VMS rates were calculated as number of VMS/monitoring time and standardized to a 24-hour day (17 hours wake, 7 hours sleep) to account for variations in monitoring durations across women.

AD biomarkers

Plasma biomarker concentrations were measured using Single molecule array (SIMOA) technology on an HD-X instrument (Quanterix, Billerica, MA, USA). All frozen samples underwent a single thawing cycle. were measured with the Neurology 4-Plex E (#103670) commercial assays from Quanterix (Billerica, MA, USA). Plasma p-tau181 and p-tau231 were measured using validated in-house assays following published protocols.^20–22 For each assay, two or three quality control samples of different concentrations were analyzed in duplicates both at the start and the end of each technical run to estimate reproducibility. The pooled quality control data showed that the within- and between-run variations in signal were <20% (mostly around 10%).

Covariates

Women completed questionnaires assessing demographics, health behaviors, and medication use. Race/ethnicity was self-reported. Educational attainment was assessed as years of completed education. Although a history of head trauma with loss of consciousness >60 min was exclusionary, other head traumas were recorded and considered. Height was measured via a fixed stadiometer and weight via balance beam scale. Body mass index (BMI) was calculated [weight (kg)/height² (m)]. Genotypes for two APOE polymorphisms, rs429358 (APOE ɛ4) and rs7412 (APOE ɛ2) were determined using TaqMan genotyping assays.²³ Because of the strong linkage disequilibrium between the two sites, this is also treated as a three-allele APOE polymorphism: APOE ɛ2, APOE ɛ3, and APOE ɛ4, resulting in six genotypes (2/2,2/3,2/4,3/3,3/4,4/4). Participants were classified as either APOE ɛ4 positive (2/4, 3/4, 4/4) or negative (2/2, 2/3, 3/3) for analysis. Endogenous estradiol (E2) was measured from serum at the University of Pittsburgh’s Small Molecule Biomarker Core using ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS), the reference method for the measurement of E2 in postmenopausal women.^24,25 The E2 assay had a lower limit of quantitation of 1.0 pg/mL, intra-day errors <8.1% and relative standard deviations (RSD)<10.4%, inter-day errors <5.0% and RSD <7.4%. Glucose and insulin were assessed using an enzymatic assay or immunoturbidimetric assay. Homeostatic model assessment (HOMA) for insulin resistance, was calculated as [(fasting insulin*fasting glucose)/22.5].²⁶ To assess sleep, women wore a Actiwatch 2 wrist actigraph unit on the wrist of the non-dominant hand (Respironics, Inc., Murrysville, PA)²⁷ and completed a sleep diary²⁸ over the three ambulatory monitoring days. Actigraphy data were collected in 1-min epochs and analyzed with Philips Actiware v6.0.0 software, with a wake threshold of 40 and number of epochs of sleep/wake for sleep onset/offset of 10. Wake after sleep onset (WASO; minutes of wakefulness between actigraphy-defined sleep onset time and actigraphy-defined final wake time) and actigraphy-assessed total sleep time were considered.

Statistical analysis

Variables were examined for distributions, outliers, and cell sizes. P-tau181, P-tau231, NfL, GFAP, BMI, WASO, and E2, were log transformed to conform to model assumptions of normality. Aβ42/Aβ40 ratio was calculated. Bivariate relations between study variables and AD biomarkers were examined via Pearson and Spearman correlation coefficients. We tested the relation of VMS (total, sleep, wake) to each AD biomarker separately in linear regression models. Consistent with other work,²⁹ we also classified individuals into homogeneous Aβ42/Aβ40 groups via the unsupervised clustering method K-medoids,³⁰ with the Euclidean distance used to calculate the distance matrix and cluster number fixed at two. We tested the association between VMS and cluster membership via logistic regression. Covariates were age, race/ethnicity, education, BMI, and APOE ɛ4 status. In secondary models, we also considered covariates E2, WASO, total sleep time, a head trauma history, HOMA scores, and diabetes medications as additional covariates. Effect modification of VMS-AD biomarker relationships by APOE ɛ4 status was also tested. In additional models, we considered associations between VMS and AD biomarkers with a false discovery rate correction (for 15 models). All tests were two tailed with an alpha set to 0.05. Analyses were conducted using SAS v9.4 (SAS Institute, Cary, NC). The sample size of 248 provided 80% of power to detect a small to medium effect size (Cohen’s f =0.17) in a linear regression with up to 5 covariates at a two-sided significance level of 0.05.

Results

Participants were on average 59 years old, overweight, and postmenopausal (Table 1). They showed an approximate average of 5 objective VMS per day on ambulatory monitoring.

Table 1.

Participant characteristics

N	248
Age, M (SD)	59.06 (4.34)
Race/ethnicity, N (%)
White	201 (81.05)
Black	38 (15.32)
Asian or mixed race/ethnicities	9 (3.63)
Years of education, Median (IQR)	16.00 (14.00, 18.00)
Body mass index, Median (IQR)	28.72 (23.78, 32.50)
Menopause stage, postmenopausal, N (%)	245 (98.79)
Estradiol, pg/mL, Median (IQR)	3.00 (2.00, 5.00)
APOE ɛ4 carrier, yes, N (%)	61 (24.50)
Wake after sleep onset, minutes, Median (IQR)	35.67 (26.08, 51.00)
Total sleep time, hours, M (SD)	6.43 (1.07)
History of head trauma, N (%)	39 (15.73)
Aβ42/Aβ40, M (SD)	0.08 (0.01)
P-tau181, pg/mL, Median (IQR)	2.88 (2.47, 3.25)
P-tau231, pg/mL, Median (IQR)	2.54 (2.08, 3.20)
NfL, pg/mL, Median (IQR)	2.41 (2.06, 2.75
GFAP, pg/mL, Median (IQR)	4.24 (3.84, 4.60)
Objective-assessed vasomotor symptoms (VMS), M(SD) †
24-hour	4.98 (1.00, 10.59)
Wake	2.75 (0, 7.51)
Sleep	1.49 (0, 3.61)
Subjectively-reported vasomotor symptoms (VMS), M (SD)†
24-hour	1.86 (3.12)
Wake	1.38 (2.39)
Sleep	0.49 (0.97)

^†To account for variations in monitoring time, VMS expressed as number of VMS/monitoring time and standardized to a 24-hour day (17 hours wake, 7 hours sleep)

Objectively-assessed VMS, particularly sleep VMS, were associated with significantly lower Aβ42/Aβ40 in multivariable models (Table 2, Figure 1). Associations between VMS and other AD biomarkers were not statistically significant. When considering Aβ42/Aβ40 clusters, two clusters characterized by low [M(SD)=.06(.01)] and high [M(SD)=.09(.01)] Aβ42/Aβ40 emerged. More frequent sleep VMS were significantly associated with membership in the low Aβ42/Aβ40 cluster [OR (95%CI)=1.18 (1.05, 1.33), p=.006, multivariable].

Table 2.

Associations between objectively-assessed VMS and AD biomarkers

	Aβ42/Aβ40 B(SE)	P value	p-tau181 B(SE)	P value	p-tau231 B(SE)	P value	NfL B(SE)	P value	GFAP B(SE)	P value
24-hour VMS	−.0002 (.0001)	.099	.997 (.006)	.564	.993 (.008)	.404	1.002 (.004)	.720	1.001 (.005)	.840
Sleep VMS	−.0010 (.0004)	.018	.990 (.017)	.560	.979 (.023)	.349	1.014 (.013)	.290	1.015 (.015)	.301
Wake VMS	−.0002 (.0002)	.687	.997 (.008)	.566	.994 (.010)	.566	1.000 (.006)	.998	.998 (.006)	.811

Note: p-tau181, p-tau231, NfL, GFAP, and BMI log transformed for analysis, results back transformed for p-tau181, p-tau231, NfL, GFAP back transformed for display

Covariates: age, race/ethnicity, education, BMI, APOE ɛ4 status

Figure 1.

Scatterplot of associations between objectively-assessed sleep vasomotor symptoms (VMS) and Aβ42/Aβ40

We conducted several secondary analyses. Associations between sleep VMS and Aβ42/Aβ40 remained with additional adjustment for endogenous E2 [B(SE)=−.001 (.0004), p=.008, multivariable with E2], a history of head trauma [B(SE)=−.001 (.0004), p=.014, multivariable with head trauma history], actigraphy assessed total sleep time [B(SE)= −0.001 (SE=.0004), p=.018, multivariable with total sleep time] or WASO [B(SE)=−.001 (SE=.0004), p=.017, multivariable with WASO], HOMA scores [B(SE)=−.001 (.0004), p=.017, multivariable with HOMA] or diabetes medications [B(SE)=−.001 (.0004), p=.019, multivariable with diabetes medications]. There were no significant interactions between APOE ɛ4 status and VMS in relation to the AD biomarkers (interaction p-values=.270-.914). We additionally conducted models applying a false discovery rate correction; in these models none of the associations between VMS and the AD biomarkers were statistically significant [e.g., associations of objective sleep VMS to Aβ42/Aβ40: B(SE) −.0010 (.0004), p=.274, multivariable]. We also considered self-reported (subjective) VMS; Objective and subjective VMS were moderately correlated (24-hour: r=.56, p<.0001; Wake: r=.55, p<.0001; Sleep: r=.40, p<.0001). Self-reported VMS were not associated with AD biomarkers (Supplemental Table 1).

Comment

Principal Findings

We report here for the first time that VMS, and specifically sleep VMS, are associated with an AD risk factor: lower plasma Aβ42/40 ratio. Importantly, this association persisted after controlling for E2 and actigraphy measures of sleep duration and WASO. Amyloid pathology is the earliest biomarker of AD along the continuum from normal cognitive aging to dementia, and a lower Aβ42/40 ratio predicts longitudinal progression to AD.^31,32 We did not find associations of VMS with other AD biomarkers including p-tau 181 and 231, GFAP, and NfL. Overall, these data suggest a link between sleep VMS and amyloid pathology in midlife women.

Results in the Context of What is Known

To date the literature on menopause and AD has focused on estrogen and AD biomarkers, but to our knowledge this is the first to focus on VMS and their association with AD biomarkers independent of hormonal changes. Prior studies suggest a link between changes in E2 and Aβ in midlife women.^33–35 For example, compared with placebo, transdermal E2 lowered Aβ deposition in early postmenopausal women, particularly among women with the APOE ɛ4 genotype, the most common risk for late-life AD.³⁴ Similar findings were reported in an observational study examining change in plasma AD biomarkers in women before and after treatment with menopausal hormone therapy (MHT) in comparison to non-treated controls. MHT appeared to have differential benefits on Aβ (measured by change in Aβ42/40 ratio) among APOE ɛ4 carriers.³³ Further, regardless of APOE ɛ4 genotype, MHT-treated women showed a more favorable outcome in the balance of the two key AD biomarkers (i.e., Aβ and phosphorylated tau as measured by the ratio of Aβ42/40:ptau231).³³ Consistent with this evidence, compared to premenopausal women, peri- and postmenopausal women show greater Aβ deposition in the prefrontal cortex over time.³⁵ Generally, women may have poorer clearance of Aβ in the brain, particularly if they are APOE ɛ4 carriers.³⁶ Collectively, these studies underscore the importance of menopause and hormonal factors to women’s brain health. The present study showed that sleep VMS were associated with AD biomarkers, even after controlling for E2 measured using a highly sensitive technique.

It is well established that VMS disturb sleep,³⁷ and sleep is an important risk factor for AD.^38–40 Nonetheless, in the present study the associations between sleep VMS and a lower Aβ42/40 ratio persisted after controlling for sleep disturbance (WASO, sleep time, sleep efficiency) measured objectively with actigraphy. Additionally, prior work has demonstrated the importance of measuring VMS objectively with ambulatory monitors, as VMS associations with memory, brain structure, brain function and brain connectivity were observed only when VMS were measured objectively and not with self-report.^5–9,13,14 Similarly here the association between VMS and the Aβ42/40 ratio was only observed when VMS were measured objectively, and not with self-report. This pattern of results is consistent with and likely reflects the greater validity of monitored VMS. For example, randomized clinical trials of hormonal and nonhormonal VMS treatments show a placebo effect for reported VMS but not monitored VMS.^41–43 Further, monitored VMS are needed to accurately measure VMS during sleep for the obvious reason that women can only report sleep VMS if they are awake.

Clinical Implications

The findings indicate the potential greater amyloid pathology among midlife women with frequent sleep VMS. These findings indicate that frequent sleep VMS may point to a vulnerable phenotype of women that warrant increased surveillance and aggressive risk factor reduction to preserve brain health with aging. If replicated, the present finding that sleep VMS are associated with indicators of amyloid pathology raises the hypothesis that sleep VMS may be a modifiable risk factor for cognitive decline and possibly dementia.

Research Implications

Broadly these findings suggest that when examining menopause-related risk factors for AD, it is critical to look not only at hormonal factors but also at menopause symptoms, particularly sleep VMS. All women who live into late life experience a permanent decline in estradiol levels, but the experience of VMS varies widely among women, with many women experiencing persistent VMS a decade or more after the final menstrual period.⁴⁴ Self-reported VMS are not sensitive to these associations.^5–9,13,14 Prior studies have examined sleep in relation to AD biomarkers,^45–47 but none to our knowledge have measured or considered the role of nighttime VMS. Here the association of AD biomarkers with nighttime VMS was evident independent of actigraphy-based measures of sleep. It is thereby possible that some of the prior work linking sleep disruption to AD biomarkers in midlife women may be due to VMS. In future studies examining sleep and AD biomarkers, it may be important to account for VMS symptoms.

Strengths and Limitations

This study has strengths, including it being the first to examine whether VMS are related to AD biomarker concentrations. Further, VMS were assessed via objective monitor, which is particularly important to assessing VMS during sleep, as sleep may decrease the accuracy of symptom reporting. This study was conducted among a well-characterized community-dwelling cohort of women. This study used blood-based biomarkers that are considerably less costly and invasive than other modalities to detect amyloid or tau (e.g., positron emission tomography, cerebrospinal fluid sampling) and show moderate to strong sensitivity and specificity relative to these more invasive measures (e.g., the sensitivity and specificity of the Quanterix Aβ1–42/Aβ1–40 assays for the detection of amyloid ranges, respectively, from 0.45 to 0.97 and from .510 to .89).^48–52 However, there were several limitations. The racial/ethnic diversity of the sample was limited; future research should address this question in more diverse cohorts. The lower rates of VMS reporting may have limited our ability to detect associations for subjective VMS. With application of a correction to account for multiple comparisons, associations between sleep VMS and Aβ42/40 were no longer statistically significant; findings should be replicated in future work. Further, similar to our prior reports of associations between VMS and other brain outcomes,^5–9,13,14 it is unknown if the association between sleep VMS and AD biomarkers is causal. Future work should clarify both the directional and temporal course of these relationships. However, in a proof-of-concept clinical trial of a non-hormonal intervention for VMS, the magnitude of improvement in VMS following the intervention correlated with the magnitude of improvement in memory.⁸

Conclusions

Study findings indicate that sleep VMS are associated with a lower plasma Aβ42/40 ratio, suggesting a link between VMS and amyloid pathology in midlife women. If replicated, the present finding that sleep VMS are associated with indicators of amyloid pathology raises the hypothesis that sleep VMS may be a modifiable risk factor for cognitive decline and possibly dementia. The mechanisms by which sleep VMS may affect amyloid are unknown but may include shifts in bioenergetics, a vulnerable cerebrovasculature, and surges in glucocorticoids, as well as indirect effects of chronic sleep disruptions. At minimum, sleep VMS appear to be a sex-specific marker of vulnerability to amyloid pathology in women.

AJOG at a Glance:

A. Why was this study conducted?

Identifying Alzheimer’s disease (AD) risk factors in women is important as women comprise two thirds of those with AD. Some work links menopausal vasomotor symptoms (VMS) to poorer neurocognitive health, yet it is unknown whether VMS are linked to biomarkers of AD. Among 248 midlife women who underwent VMS monitoring, we tested whether objectively-assessed VMS are associated with plasma biomarkers of AD.

B. What are the key findings?

Objectively-assessed VMS occurring during sleep were associated with significantly lower Aβ42/Aβ40, suggestive of greater brain Aβ pathology. Findings were not explained by covariates age, race/ethnicity, education, body mass index, apolipoprotein E4 status, estradiol, or sleep.

C. What does this study add to what is already known?

Sleep VMS may be a marker of women at risk of AD. Findings point to a potentially important female-specific midlife risk factor for AD.