Reproductive hormones in relation to white matter hyperintensity volumes among midlife women

Abstract

INTRODUCTION

Although reproductive hormones are implicated in cerebral small vessel disease in women, few studies consider measured hormones in relation to white matter hyperintensity volume (WMHV), a key indicator of cerebral small vessel disease. Even fewer studies consider estrone (E1), the primary postmenopausal estrogen, or follicle‐stimulating hormone (FSH), an indicator of ovarian age. We tested associations of estradiol (E2), E1, and FSH to WMHV among women.

METHODS

Two hundred twenty‐two women (mean age = 59) underwent hormone assays (E1, E2, FSH) and 3T brain magnetic resonance imaging. Associations of hormones to WMHV were tested with linear regression.

RESULTS

Higher E2 (B[standard error (SE)] = –0.17[0.06], P = 0.008) and E1 (B[SE] = –0.26[0.10], P = 0.007) were associated with lower whole‐brain WMHV, and higher FSH (B[SE] = 0.26[0.07], P = 0.0005) with greater WMHV (covariates age, race, education). When additionally controlling for cardiovascular disease risk factors, associations of E1 and FSH to WMHV remained.

DISCUSSION

Reproductive hormones, particularly E1 and FSH, are important to women's cerebrovascular health.

Highlights

• Despite widespread belief that sex hormones are important to women's brain health, little work has considered how these hormones in women relate to white matter hyperintensities (WMH), a major indicator of cerebral small vessel disease.

• We considered relations of estradiol (E2), estrone (E1), and follicle‐stimulating hormone (FSH) to WMH in midlife women.

• Higher E2 and E1 were associated with lower whole‐brain WMH volume (WMHV), and higher FSH with higher whole‐brain WMHV.

• Associations of E1 and FSH, but not E2, to WMHV persisted with adjustment for cardiovascular disease risk factors.

• Findings underscore the importance of E2 and FSH to women's cerebrovascular health.

1. BACKGROUND

Menopause, a universal midlife transition for women, is a time of marked change in reproductive hormones, particularly declines in estradiol (E2) and increases in follicle‐stimulating hormone (FSH). ¹ These hormonal changes are understood to have implications for the brain. This research has its roots in early discoveries of the key influence of ovarian estrogen E2 on the structure and function of the hippocampus. ² , ³ , ⁴ Subsequent work has underscored the key role of reproductive hormones in multiple aspects of brain structure and function with relevance to brain aging and Alzheimer's disease. ⁵

Despite widespread recognition of the importance of reproductive hormones to the brain, there has been limited attention to the cerebrovasculature. This oversight stands in contrast to the recognized importance of reproductive hormones to the peripheral vasculature ⁶ and to both stroke and dementia risk ⁵ in women. ⁷ , ⁸ Research links an earlier natural menopause or younger/premenopausal oophorectomy with elevated stroke risk ⁹ , ¹⁰ , ¹¹ and a longer reproductive lifespan with lower stroke risk. ¹² , ¹³ These findings have been interpreted to indicate a salubrious impact of endogenous E2 on the cerebrovasculature. However, these studies have inferred endogenous E2 exposure indirectly via surgery or events such as pregnancy, menarche, or menopause. Thus, the relationship of endogenous E2 to women's cerebrovascular health remains poorly understood.

Studies of hormones and the brain typically focus on E2, the main ovarian estrogen. However, hallmark changes of menopause include not only decreases in E2 but also increases in the pituitary hormone FSH. ¹ , ¹⁴ Moreover, when E2 stabilizes at low levels postmenopausally, estrone (E1), an estrogen produced through aromatization of adrenal androgens in peripheral tissues (largely adipose tissue), emerges as the predominant postmenopausal estrogen. ¹⁵ , ¹⁶ Consideration of hormones beyond E2 is needed to understand the implications of reproductive hormones for brain health.

Few studies have considered relationships between measured endogenous reproductive hormones to neuroimaging‐based measures of cerebrovascular health. White matter hyperintensities (WMH), which appear as hyperintense areas on T2‐weighted (T₂w), fluid‐attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI) sequences are considered hallmarks of cerebral small vessel damage and indicators of cerebrovascular health linked to future cognitive decline, dementia, and mortality risk. ¹⁷ Studies examining endogenous hormones and WMH are limited. One study that did not find significant associations between endogenous hormones and white matter hyperintensity volume (WMHV) was a small study (N = 44 women) of human immunodeficiency virus–positive individuals. ¹⁸ Another study considered endogenous hormones (E1, E2, FSH, luteinizing hormone) and WMHV among postmenopausal women (N = 78) who underwent MRI and were participating in the hormone therapy (HT) trial KEEPS. ¹⁹ In KEEPS, no significant associations were observed between endogenous hormones and WMHV at baseline (before randomization) or between changes in hormones and changes in WMHV among women not taking HT (N = 31); only among women randomized to HT were decreases in FSH and increases in E1 associated with smaller increases in WMH over time. ¹⁹ Critically, interpretation is limited by exogenous HT use, which itself can impact WMH. ²⁰ Thus, studies of endogenous reproductive hormones and neuroimaging indicators of cerebrovascular health are very few, based on small sample sizes, and have critical interpretive limitations (e.g., HT use). Further, studies have not considered the role of vasomotor symptoms (VMS), which when measured objectively are linked greater WMHV. ²¹ Thus, whether endogenous reproductive hormone concentrations are related to WMHV in women is unknown.

Among 222 midlife women not using HT who underwent assessments of endogenous reproductive hormones and 3T brain MRI, we tested associations between E1, E2, and FSH in relation to WMHV. We measured E1 and E2 using liquid chromatography with tandem mass spectrometry (LC‐MS/MS), a state‐of‐the‐art method critical to quantifying the low levels of E2 that occur postmenopausally. We hypothesized that lower E1, lower E2, and higher FSH would be associated with greater WMHV. We considered both whole‐brain WMHV as well as the spatial distribution of WMHV. We tested associations adjusting for key covariates. We additionally considered the role of VMS in these associations.

2. METHODS

2.1. Sample

The MsBrain study is a study of menopause and brain health conducted among community participants. From 2017 to 2020, a diverse sample of 274 participants were recruited from the Pittsburgh, Pennsylvania, community from local registries, via advertisements, and from an existing study of menopause and cardiovascular health. ²² Exclusion criteria included pregnancy; hysterectomy and/or bilateral oophorectomy; history of stroke/cerebrovascular accident, Parkinson's disease, dementia; seizure disorder; brain tumor or chemotherapy; history of head trauma with loss of consciousness > 60 minutes; active substance abuse (established via urine toxicology screen and interview); current use of systemic estrogen or progesterone, selective estrogen receptor modulators, aromatase inhibitors, gabapentin, selective serotonin reuptake inhibitors, or serotonin norepinephrine reuptake inhibitors; and contraindication to MRI.

Of the 274 women enrolled, 239 underwent neuroimaging. Of these women, nine were excluded due to detection of brain tumor or stroke on neuroradiological review of the MRI or to reported seizure disorder, five due to a reported history of chemotherapy, and three women for not undergoing phlebotomy. One additional E2 observation that was an extreme outlier (> 5 standard deviations from the mean) was excluded from E2 models. Thus, the primary sample for E1 and FSH models was N = 222 and for E2 models N = 221. Sample sizes for models incorporating apolipoprotein E (APOE) carrier status were N = 214 (due to refusal of genetic testing).

2.2. Design and procedures

MsBrain participants underwent screening, physical measurements, a medical history interview, questionnaires, ambulatory VMS monitoring, fasting phlebotomy, and brain MRI. Study procedures were reviewed and approved by the University of Pittsburgh Human Research Protection Office. All participants provided written informed consent.

RESEARCH IN CONTEXT

1. Systematic review: Reproductive hormones have been implicated in women's brain health. However, there has been limited attention to the cerebrovasculature, particularly using neuroimaging‐based markers. Further, studies typically focus on the estrogen estradiol (E2). Other reproductive hormones, including the estrogen estrone (E1) and follicle‐stimulating hormone (FSH), may be important to consider, particularly during the postmenopause when E2 levels are low.

2. Interpretation: Higher E1 was associated with fewer whole‐brain, deep, frontal, and temporal white matter hyperintensity volumes (WMHV). Higher FSH was associated with greater whole‐brain, periventricular, and frontal WMHV. These associations persisted adjusting for demographics and cardiovascular disease (CVD) risk factors. Associations between E2 and WMHV were explained by CVD risk factors, particularly adiposity.

3. Future directions: Findings indicate the importance of endogenous reproductive hormones, particularly E1 and FSH, as determinants of midlife women's cerebrovascular health. Future work can consider whether efforts to promote reproductive health support brain health.

2.3. Measures

2.3.1. Hormones

Women underwent phlebotomy after overnight fast. E1 and E2 were measured at the University of Pittsburgh's Small Biomarker Core using LC‐MS/MS, which uses liquid–liquid extraction, derivatization, and detection with a triple quad mass spectrometer. ²³ E1 and E2 assays had a lower limit of quantitation of 1.0 pg/mL, and for E2, intra‐day errors < 8.1% and relative standard deviations (RSD) < 10.4%, inter‐day errors < 5.0% and RSD < 7.4%; and for E1, intra‐day errors < 15.0% and RSD < 11.2%, inter‐day errors < 11.0% and RSD < 10.9%. For E2 values below the sensitivity of the assay (n = 19), a random number between 0 and the lower limit was generated and assigned to these values, consistent with the approach of prior studies. ²⁴ FSH was measured via a commercially available enzyme‐linked immunosorbent assay (ELISA; Cayman Chemical Company) for in vitro quantitative measurement of FSH in human serum, with a sensitivity of 0.6 mIU/mL, inter‐assay coefficient of variations (CVs) from 4.5% (22.4 mIU/mL) to 8.9% (10.4 mIU/mL), and intra‐assay CVs from 1.0% (22.4 mIU/mL) to 4.5% (10.4 mIU/mL). Standards, blanks, calibrators, and control pools were run simultaneously with all samples. 

2.3.2. WMHs

MRI scanning was performed at the MR Research Center of the University of Pittsburgh with a 3T Siemens Prisma MR scanner and a Siemens 64‐channel head coil. Two series of MR images were analyzed for the current study: magnetization‐prepared rapid gradient echo (MPRAGE) T1‐weighted sequence and T₂w FLAIR sequence. MPRAGE images were acquired in the axial plane using the parameters: repetition time (TR) = 2400 ms; echo time (TE) = 2.22 ms; inversion time (TI) = 1000 ms; flip angle = 8°; field of view (FOV) = 256*240 mm; slice thickness = 0.8 mm; voxel size = 0.8 mm*0.8 mm; matrix size = 320*300; and number of slices = 208. FLAIR images were acquired in the axial plane using the parameters: TR = 9690 or 10,000 ms; TE = 91 ms; TI = 2500 ms; flip angle = 135°; FOV = 256 × 256 mm; matrix = 320 × 320; slice thickness = 1.6 mm; voxel size = 0.8 mm*0.8 mm; and number of slices = 104. The small change in TR from 9690 to 10,000 was performed 1 year into the study to meet specific absorption rate human safety guidelines for participants with a higher body mass index (BMI). This change slightly increased the time of acquisition but had minimal effect on image contrast.

An automated pipeline was used to segment WMHs on the T₂w FLAIR images using previously validated methods. ²⁵ For each participant, cerebral and cerebellar white matter were segmented on the T₁w image and mapped into the T₂w FLAIR image space using SPM12 (http://www.fil.ion.ucl.ac.uk/spm/) and FreeSurfer (version 7.1.1, https://surfer.nmr.mgh.harvard.edu/). Manual inspection of each image was performed to rule out cerebellar WMHs. As there were very few lesions in the cerebellum, cerebellar white matter represented normal‐appearing white matter, and its intensity mean and standard deviation were used for Z transformation of the T₂w FLAIR image. A threshold of two was then applied on Z transformed FLAIR images. This method uses individual mean and standard deviation from normal cerebellar white matter to standardize individual FLAIR images, avoiding systematic bias in seed selection between participants with significant cerebral WMHs versus those with few WMHs. Z transformation also reduces intensity variations across individual FLAIR images.

In FreeSurfer, white matter was parcellated according to its nearest cortical region with the Desikan–Killiany atlas, which was used to generate the cortical white matter masks for frontal, temporal, parietal, and occipital lobes for the localization of WMHs. White matter parcellations corresponding to frontal cortex regions in the Desikan–Killiany atlas were combined to create a frontal cortical white matter mask to localize frontal WMHs. Cortical white matter masks were generated for temporal, parietal, and occipital lobes. These lobular cortical white matter masks were non‐overlapping and were combined to create an overall cortical/deep white matter mask. White matter surrounding the ventricles that is not part of the cortical/deep white matter mask comprised the periventricular white matter mask. These non‐overlapping lobular cortical masks and periventricular white matter allowed us to investigate the association of hormones to regional WMHs. To account for individual differences in brain size, the total and regional WMHV (in cubic centimeters) were normalized by intracranial volumes (ICV; nWMHs = WMHs/ICV).

2.3.3. Additional measures

Participants underwent physical measurements, interviews, questionnaires, and ambulatory VMS monitoring. Height was measured via a fixed stadiometer and weight via a balance beam scale. BMI was calculated as weight(kg)/height(m²). Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were the average of three seated measurements (Dinamap v100). Demographics; medical, reproductive, and psychiatric history; medication use; and health behaviors were assessed by questionnaires/interview. Race/ethnicity was self‐reported (categorized for analysis as White vs. Black, Asian, or mixed race/ethnicity due to small cell sizes). Education was reported as years of completed education. Participants underwent 24 hours of physiologic VMS monitoring with the wearable VMS monitor (VU‐AMS 5fs, VU University Amsterdam), a validated measure of physiologic VMS quantified via sternal skin conductance. ²⁶ After monitoring, VMS data were downloaded and scored via UFI software (DPS) according to validated methods. ²¹ Only physiologic VMS files with ≥ 70% of usable data were included. Wake and sleep VMS were categorized based upon actigraphy‐defined wake and sleep times, with VMS rates calculated as VMS number/monitoring time, standardized to a 24‐hour day (17 hours wake, 7 hours sleep) to account for between‐participant variations in monitoring duration.

Women underwent phlebotomy after an overnight fast. Glucose, total cholesterol, high‐density lipoprotein (HDL), and triglycerides were determined using enzymatic assays and insulin via immunoturbidimetric assay (Alfa Wassermann). Low‐density lipoprotein (LDL) was calculated using the Friedewald equation. ²⁷ Homeostatic model assessment (HOMA) for insulin resistance was calculated ([insulin*glucose]/22.5). Genotypes for two APOE polymorphisms, rs429358 (APOE ɛ4) and rs7412 (APOE ɛ2), were determined using TaqMan genotyping assays. ²⁸ Because of the strong linkage disequilibrium between the two sites, genotype was also treated as a three‐allele APOE polymorphism: APOE ɛ2, APOE ɛ3, and APOE ɛ4, yielding six genotypes (ɛ2/ ɛ2, ɛ2/ɛ3, ɛ2/ɛ4, ɛ3/ɛ3, ɛ3/ɛ4, ɛ4/ɛ4). Participants were classified as APOE ɛ4 positive (ɛ2/ɛ4, ɛ3/ɛ4, ɛ4/ɛ4) or negative (ɛ2/ɛ2, ɛ2/ɛ3, ɛ3/ɛ3); we also considered models excluding APOE ɛ2–positive women. As findings were comparable, findings with the full sample are reported here.

2.4. Statistical analysis

E1, E2, FSH, BMI, HOMA, triglycerides, and WMHV values were log‐transformed for analysis. Hormones (E1, E2, FSH) were each tested separately in relation to whole‐brain WMHV in linear regression models. Covariates included demographics (age, race, education) and CVD risk factors (BMI, SBP, anti‐hypertensive medication, smoking, HDL, triglycerides, HOMA). One blood pressure variable (SBP) was included in models given the collinearity among blood pressure variables. Exploratory models considered hormones in relation to regional WMHV (periventricular, deep, and frontal, temporal, parietal, occipital lobes). Because APOE ɛ4 status was an important moderating factor of other exposures in relation to WMHV in this cohort we considered effect modification of hormone–WMHV relationships by APOE ɛ4 status in secondary models. ²⁹ , ³⁰ Additional models considered the role of sleep VMS in associations, including a test of sleep VMS as a mediator of relationships between hormones and WMHV using the Sobel test. ³¹ We also tested associations between reproductive hormones and regional WMHV via multivariate analysis of covariance (MANCOVA) that allows for simultaneous consideration of multiple outcomes in the same model. Tests were two tailed with α = 0.05. Analyses were conducted in SAS v9.4 (SAS Software).

3. RESULTS

Participants were on average 59 years old (range 45–67), overweight, postmenopausal, and with a favorable CVD risk factor profile (Table 1). Most participants identified as non‐Hispanic White (82%) or Black (13%).

TABLE 1.

Participant characteristics.

N	222
Age, M (SD)	59.20 (4.29)
Race/ethnicity, N (%)
White	182 (81.89)
Black	29 (13.06)
Asian or mixed race/ethnicity	11 (4.95)
Years of education, M (SD)	15.76 (2.38)
BMI, median (IQR)	26.71 (23.80, 31.87)
SBP, mmHg, M (SD)	118.65 (14.69)
DBP, mmHg, M (SD)	68.37 (8.97)
LDL, mg/dL, M (SD)	119.67 (35.81)
HDL, mg/dL, M (SD)	70.80 (20.88)
Triglycerides, mg/dL, median (IQR)	91.00 (69.00, 122.00)
HOMA, median (IQR)	2.75 (1.21, 4.04)
Current smoking, yes, N (%)	3 (1.35)
Menopause stage, N (%)
Perimenopausal	2 (0.90)
Postmenopausal	220 (99.10)
APOE ɛ4 positive, N (%)	48 (22.43)
E1, pg/mL, median (IQR)	30.00 (22.00, 42.00)
E2, pg/mL, median (IQR) a	3.00 (2.00, 5.00)
FSH, mIU/mL, median (IQR)	57.90 (38.90, 86.14)
WMHV b median (IQR)
Whole brain	0.065 (0.043, 0.102)
Deep	0.013 (0.007, 0.026)
Periventricular	0.050 (0.033, 0.078)
Frontal	0.003 (0.002, 0.006)
Parietal	0.001 (0.0001,0.003)
Temporal	0.003 (0.001, 0.006)
Occipital	0.003 (0.001, 0.009)

Abbreviations: APOE, apolipoprotein E; BMI, body mass index; DBP, diastolic blood pressure; E1, estrone; E2, estradiol; FSH, follicle‐stimulating hormone; HDL, high‐density lipoprotein; HOMA, homeostatic model assessment; IQR, interquartile range; LDL, low‐density lipoprotein; SBP, systolic blood pressure; WMHV, white matter hyperintensity volume.

^a N = 221.

^b WMHV expressed as cm³/Intracranial volume.

When considering associations between hormones and whole‐brain WMHV; in age, race, and education‐adjusted models, higher E1 and higher E2 were each associated with lower whole‐brain WMHV, and higher FSH was associated with greater whole‐brain WMHV. Further adjustment for BMI and other CVD risk factors accounted for the associations between E2 and WMHV, while associations of both E1 and FSH to WMHV remained (Table 2, Figure 1).

TABLE 2.

Association of reproductive hormones to whole‐brain WMHV.

	Whole‐brain WMHV
	Model 1	Model 2
	B(SE)	B(SE)
E1	−0.26 (0.10)***	−0.21 (0.10)**
E2	−0.12 (0.05)**	−0.06 (0.05)
FSH	0.26 (0.07)***	0.18 (0.09)**

Note: E1, E2, FSH, BMI, HOMA, triglycerides log transformed; WMHV normalized by ICV and log transformed.

Covariates: age, race/ethnicity, education, BMI, smoking, SBP, anti‐hypertensive medications, HOMA, HDL‐C, triglycerides.

Abbreviations: BMI, body mass index; E1, estrone; E2, estradiol; FSH, follicle‐stimulating hormone; HDL‐C, high‐density lipoprotein‐cholesterol; HOMA, homeostatic model assessment; ICV, intracranial volume; SBP, systolic blood pressure; SE, standard error; VMS, vasomotor symptoms; WMHV, white matter hyperintensity volume.

^** P < 0.05, ***P < 0.01.

FIGURE 1.

(A) Scatterplots of associations between E1 or FSH with WMHV (B) and representative brain images of women with high, medium, or low E1 or FSH. E1, estrone; FSH, follicle‐stimulating hormone; ICV, intracranial volume; WMHV, white matter hyperintensity volume.

We next considered the spatial distribution of WMHV. Higher E1 was strongly associated with lower deep, frontal, and temporal WMHV in multivariable models. Higher FSH was associated with greater periventricular and frontal WMHV. E2 was not significantly associated with regional WMHV in multivariable models (Table 3).

TABLE 3.

Associations of reproductive hormones to regional WMHV.

	WMHV
	Deep	Periventricular	Frontal	Parietal	Temporal	Occipital
	B(SE)	B(SE)	B(SE)	B(SE)	B(SE)	B(SE)
E1	−0.43 (0.16)***	−0.14 (0.10)	−0.66 (0.18)***	−0.49 (0.28)*	−0.62 (0.19)***	−0.02 (0.25)
E2	−0.09 (0.09)	−0.05 (0.05)	−0.18 (0.10)*	−0.01 (0.15)	−0.13 (0.11)	−0.12 (0.14)
FSH	0.18 (0.14)	0.20 (0.09)**	0.32 (0.16)**	0.28 (0.26)	0.04 (0.18)	0.08 (0.22)

Notes: E1, E2, FSH, BMI, HOMA, triglycerides log transformed; WMHV normalized by ICV and log transformed.

Covariates: age, race/ethnicity, education, BMI, smoking, SBP, anti‐hypertensive medications, HOMA, HDL‐C, triglycerides.

Abbreviations: BMI, body mass index; E1, estrone; E2, estradiol; FSH, follicle‐stimulating hormone; HDL‐C, high‐density lipoprotein‐cholesterol; HOMA, homeostatic model assessment; ICV, intracranial volume; SBP, systolic blood pressure; SE, standard error; VMS, vasomotor symptoms; WMHV, white matter hyperintensity volume.

^* p < 0.10, **p < 0.05, ***p < 0.01.

We conducted several secondary analyses. Given our prior work showing that sleep VMS was associated with greater WMHV, ²¹ we considered the role of sleep VMS in associations of E1 or FSH with WMHV. Associations between E1 and WMHV, particularly deep, frontal, and temporal WMHV, persisted adjusting for sleep VMS. For FSH, associations between FSH and WMHV were explained by sleep VMS; additional models suggested that sleep VMS may serve as a mediator of relationships between FSH and WMHV (e.g., for whole‐brain WMHV, indirect effect: B[standard error (SE)] = 0.041 [0.023], P = 0.075, multivariable models; Table S1 for supporting information). Further, we considered interactions between hormones and APOE ε4 status in relation to whole‐brain WMHV; interactions were not statistically significant (Ps > 0.70). We additionally tested the association between reproductive hormones and regional WMHV via MANCOVA; findings were consistent with primary models (Table S2 for supporting information). We further considered associations with E2 excluding the 19 interpolated E2 values; findings were comparable to primary models (data not shown).

4. DISCUSSION

This is the first large study to examine associations of endogenous E1, E2, and FSH in relation to WMHV among a well‐characterized community‐dwelling sample of late midlife women. We found that higher E1 was associated with fewer whole‐brain, deep, frontal, and temporal WMHV, particularly notable given that levels of endogenous estrogens are low at this stage in life. Higher FSH was associated with greater whole‐brain, periventricular, and frontal WMHV. These associations persisted with adjustment for a range of demographic factors and CVD risk factors. Higher E2 was associated with fewer WMHV in models adjusted for demographic factors, but these associations were explained by CVD risk factors, particularly BMI. Further, sleep VMS served as a potential mediator of associations between FSH and WMHV. Collectively, these findings underscore the potential importance of endogenous reproductive hormones, particularly E1 and FSH, to women's cerebrovascular health in late midlife, and point to potential mechanisms underlying these associations including CVD risk factors and sleep VMS.

Despite the recognition of the importance of endogenous reproductive hormones to women's brain health at midlife, and the sensitivity of the peripheral vasculature to these hormones, ⁶ , ²² there has been very limited consideration of relationships of reproductive hormones to WMHV. The few existing studies produced largely null findings, were based on small samples, used reproductive history as a proxy for reproductive hormones, or included women using HT; collectively these issues have limited the understanding of the implications of endogenous reproductive hormone concentrations to cerebrovascular health. This study thereby addresses a critical gap in the literature.

E2 is the primary ovarian estrogen, and thus has been the primary focus of research on hormones and brain health in women. Although we found higher E2 associated with fewer WMHV in demographic‐adjusted models, these associations did not persist with adjustment for CVD risk factors; most notably, BMI. Conversely, few studies consider the potential importance of E1 for the brain. E1 is an estrogen produced largely in body fat. Although E1 binds to estrogen receptors with less affinity than does E2, ³² it may become more physiologically relevant during the postmenopause, when E2 is at very low levels. Notably, associations of E1 to WMHV were robust to multivariable adjustment, including to BMI, to other CVD risk factors, and to sleep VMS. These findings are consistent with our prior work underscoring the importance of E1, not E2, for endothelial function in the peripheral vasculature in midlife women. ³³ Given the importance of the vascular endothelium to both cerebral and peripheral vasculature health, ³⁴ , ³⁵ endothelial dysfunction may serve as a critical mechanism by which E1 may be linked to cerebral vascular damage.

We found that higher levels of the gonadotrophin FSH were associated with greater whole‐brain, periventricular, and frontal WMHV. FSH is the pituitary gonadotropin responsible for stimulating maturation of an ovarian follicle and is considered a key indicator of menopause stage; it increases steadily over the menopause transition as negative feedback from E2 decreases. ¹ Studies have generally not considered FSH in relation to WMHV outside of the context of HT use. A limited literature has considered FSH and peripheral vasculature, suggesting that higher FSH may be associated with more subclinical atherosclerosis or adverse vascular remodeling beyond the effects of age. ³⁶ Whereas associations here withstood adjustment for demographic factors and CVD risk factors, associations between FSH and WMHV were explained in part by sleep VMS; in fact, analyses suggested that sleep VMS may be a partial mediator of relationships between FSH and WMHV. A notable strength of the present study is its physiologic assessment of VMS, as it is the physiologic VMS, rather than self‐reported VMS, that are most related to brain health. ²¹ Leveraging these measures, our data indicate the important role that sleep VMS play in FSH–WMHV associations.

A strength of this study was the ability to consider the regional distribution of WMHV. Although of some debate, ¹⁷ the location of the WMH may be important to both their etiology and clinical implications. ³⁷ Our findings indicated that lower E1 was associated with greater deep, frontal, and temporal WMHV, suggesting a sensitivity to the vasculature in these areas to E1. Higher FSH was associated with greater whole‐brain, periventricular, and frontal WMHV. These findings point to a particular sensitivity of frontal WMHV to both E1 and FSH. Notably, frontal WMHV may originate earlier in life than other locations and may be particularly sensitive to CVD risk factors such as hypertension. ³⁷ However, these hormones were related to WMHV across multiple brain areas, pointing to the widespread implications of these hormones for WMH in women.

The mechanisms that may link these endogenous reproductive hormones to WMHV are likely multiple and specific to the hormone. Associations between E2 and WMHV were explained by other CVD risk factors, and particularly adiposity, indicating their importance to both levels of E2 and the development of WMHV. For FSH, our current findings suggested that FSH may be related to WMHV through sleep VMS; importantly, VMS appear to be particularly sensitive to FSH levels beyond E2 levels. ³⁸ None of the many CVD risk factors considered here accounted for relationships between E1 and WMHV, yet our prior work points to a potential impact of E1 on the vascular endothelium. ³³ Other work underscores the sensitivity to reproductive hormones of angiotensin II and related signaling molecules important to brain cardiovascular regulatory circuits. ³⁹ Additional potential mechanisms, including microglial activation, neuroinflammation, hypoperfusion, and blood–brain barrier changes, warrant consideration in future work. ¹⁷

This study had several limitations. First, reproductive hormone concentrations were assessed once. It is notable that 99% of the women in this study were postmenopausal. Although reproductive hormones fluctuate dramatically over the menopause transition, they typically stabilize in postmenopause, rendering a single observation during this period of the lifespan more suitable compared to earlier in the lifespan. Future work can consider how trajectories of reproductive hormones over the menopause transition relate to cerebrovascular health later in life. Second, although this sample had some racial/ethnic diversity, future work should be conducted with more diverse samples, particularly including larger numbers of Asian and Latina women, so that any racial/ethnic differences in associations can be rigorously considered and to enhance the generalizability of findings. Participants had a relatively favorable CVD risk factor profile, due in part to recruitment of a portion of the study sample from a cohort that excluded women who smoked, had insulin‐dependent diabetes, or had clinical CVD; future work can consider these associations among women with a higher CVD risk factor burden. This sample was conducted among largely postmenopausal late midlife women; how endogenous reproductive hormones relate to WMHV at other points in the female lifespan or how these hormones relate to WMHV among men cannot be inferred from these data.

This study had several strengths. It is the first study to rigorously assess how multiple reproductive hormones, including E1 and FSH, relate to WMHV in a community sample of women not using HT. In contrast to earlier work using assay methods with limited sensitivity, estrogens were measured via LC‐MS/MS, which is able to quantify the low levels of estrogens characteristic of postmenopause. It included a large sample of midlife, community‐dwelling women. Both whole‐brain and regional WMHV were characterized via rigorous methods. Multiple potentially confounding, mechanistic, and moderating factors, including CVD risk factors, APOE ε4 status, and physiologically assessed VMS, were measured and considered here.

This study highlights the importance of endogenous reproductive hormones to cerebrovascular health in late midlife postmenopausal women. Higher E1 was associated with lower whole‐brain, deep, frontal, and temporal WMHV. Higher FSH was associated with greater whole‐brain, periventricular, and frontal WMHV, a relationship largely mediated by sleep VMS. Associations between E2 and WMHV were explained by CVD risk factors, particularly BMI. Future work should consider how endogenous reproductive hormones relate to other aspects of brain health (e.g., brain volume) as well as whether efforts to promote the health of the reproductive axis may support women's cerebrovascular health. Findings indicate the importance of endogenous reproductive hormones, particularly E1 and FSH, as key determinants of midlife women's cerebrovascular health.

CONFLICT OF INTEREST STATEMENT

Dr. Thurston is a consultant/advisor for Astellas, Bayer, and Hello Therapeutics, and past consultant for Happify Health and Vira Health. Dr. Aizenstein is an advisor to Eisai. Dr. Maki is a consultant for Astellas, Bayer, Pfizer, and has equity in Alloy, Estrigenix, and Midi Health. Drs. Wu, Chang, Derby, and Barinas‐Mitchell and Ms. Harrison have no disclosures relevant to the manuscript. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All human subjects provided written informed consent.

Supporting information

Supporting Information

Supporting Information

Thurston RC, Chang Y, Wu M, et al. Reproductive hormones in relation to white matter hyperintensity volumes among midlife women. Alzheimer's Dement. 2024;20:6161–6169. 10.1002/alz.14093