Protein Biomarkers of Early Menopause and Incident Cardiovascular Disease

Abstract

Background

Premature and early menopause are independently associated with greater risk of cardiovascular disease (CVD). However, mechanisms linking age of menopause with CVD remain poorly characterized.

Methods and Results

We measured 71 circulating CVD protein biomarkers in 1565 postmenopausal women enrolled in the FHS (Framingham Heart Study). We examined the association of early menopause with biomarkers and tested whether early menopause modified the association of biomarkers with incident cardiovascular outcomes (heart failure, major CVD, and all‐cause death) using multivariable‐adjusted linear regression and Cox models, respectively. Among 1565 postmenopausal women included (mean age 62 years), 395 (25%) had a history of early menopause. Of 71 biomarkers examined, we identified 7 biomarkers that were significantly associated with early menopause, of which 5 were higher in women with early menopause including adrenomedullin and resistin, and 2 were higher in women without early menopause including insulin growth factor‐1 and CNTN1 (contactin‐1) (Benjamini‐Hochberg adjusted P<0.1 for all). Early menopause also modified the association of specific biomarkers with incident cardiovascular outcomes including adrenomedullin (P _int<0.05).

Conclusions

Early menopause is associated with circulating levels of CVD protein biomarkers and appears to modify the association between select biomarkers with incident cardiovascular outcomes. Identified biomarkers reflect several distinct biological pathways, including inflammation, adiposity, and neurohormonal regulation. Further investigation of these pathways may provide mechanistic insights into the pathogenesis, prevention, and treatment of early menopause‐associated CVD.

Nonstandard Abbreviations and Acronyms

MHT

menopausal hormone therapy

RAAS

renin‐angiotensin‐aldosterone system

SABRe

Systems Approach to Biomarker Research

Clinical Perspective.

What Is New?

• In an analysis of 71 cardiovascular disease‐related protein biomarkers in 1565 post‐menopausal women, we demonstrate that early menopause is associated with alterations in biomarkers representing inflammatory, adipokine signaling, and neurohormonal regulation pathways.

• Biomarkers that were higher in women with history of early menopause predicted incident cardiovascular outcomes, whereas biomarkers that were lower in women with early menopause were associated with lower risk of cardiovascular events.

• Early menopause appears to modify the association of specific biomarkers with incident cardiovascular outcomes; notably, higher levels of adrenomedullin were associated with all‐cause death in women with early menopause but not in women without early menopause.

What Are the Clinical Implications?

• Inflammatory, adipokine signaling, and neurohormonal regulation pathways may underlie higher cardiovascular disease susceptibility among women with early menopause beyond estrogen deficiency.

• Better understanding of the mechanisms that increase cardiovascular disease susceptibility in women with early menopause may enhance the prevention and management of cardiovascular disease in women.

Premature age of menopause represents a “risk‐enhancing factor” for atherosclerotic cardiovascular disease (CVD) in society guidelines for primary prevention. ¹ Cardiovascular risk increases in women with menopause occurring before age 50 and progressively rises with earlier age at menopause, independent of conventional cardiovascular risk factors. ² , ³ , ⁴ Historically, menopause‐associated risks were attributed to loss of cardioprotective estrogen effects. However, in randomized trials, postmenopausal hormone replacement failed to demonstrate cardiovascular preventive benefits, emphasizing the need to identify mechanistic pathways beyond estrogen deficiency. ⁵ , ⁶

We sought to identify proteomic biomarkers of early menopause and characterize their associations with incident CVD among postmenopausal women free of CVD at baseline. The SABRe CVD (Systems Approach to Biomarker Research in Cardiovascular Disease) Initiative was established to characterize the biomarker signature of atherosclerosis and CVD risk factors. As part of the SABRe CVD Initiative, circulating biomarkers representing a broad set of biological pathways, including inflammation, fibrosis, and adiposity, have been shown to predict cardiovascular outcomes and all‐cause mortality. We previously demonstrated prominent sex differences across most of these biomarkers that were further modulated by menopause status and use of menopausal hormone therapy (MHT). ⁷ Leveraging this platform, we examined the association of early menopause with protein biomarkers and evaluated whether early menopause modified the association of biomarkers with incident cardiovascular outcomes.

Methods

Data Sharing

All data and materials have been made publicly available at the National Institutes of Health database of genotypes and phenotypes and can be accessed at https://www.ncbi.nlm.nih.gov/gap/.

Study Population

The FHS (Framingham Heart Study) is a prospective longitudinal community‐based observational cohort study. ⁸ The SABRe CVD protein assay Initiative included all FHS Offspring cohort participants who attended exam 7 (1998–2001; n=3539, 54% women) and Third Generation cohort participants who attended exam 1 (2002–2005; n=4095, 53% women). Our sample included all postmenopausal women who participated in the SABRe CVD Initiative (N=1782). We excluded individuals who did not undergo biomarker ascertainment (n=139), with missing age at menopause (n=6), chemotherapy or radiation cause of menopause (n=13), prevalent heart failure (HF) (n=8), prevalent myocardial infarction (n=28), end‐stage renal disease (n=14), and missing key clinical covariates (n=9), yielding a final sample of n=1565. For analyses examining incident major CVD, participants with prevalent major CVD disease were excluded (n=20), yielding a sample of n=1545. All participants provided informed consent, and study protocols were approved by the Boston University Institutional Review Board. ⁹

Clinical Assessment

All participants underwent comprehensive medical assessment, including reproductive history, anthropometry, physical examination, and phlebotomy for protein biomarker ascertainment. Reproductive history included menopause status, MHT use, and surgical history including previous hysterectomy and oophorectomy. Menopause was self‐reported and defined as cessation of periods for ≥12 months at the time of exam. Natural menopause was defined as spontaneous menopause not resulting from surgery. Surgical menopause included menopause with bilateral oophorectomy with or without hysterectomy or hysterectomy alone without bilateral oophorectomy. For women who underwent hysterectomy without bilateral oophorectomy before natural menopause, we defined age at menopause as age at the time of hysterectomy.

The primary exposure of interest was early menopause, defined as menopause occurring before age 45 years. Women who underwent premature menopause (defined as menopause occurring before age 40 years) were also classified as early menopause. Covariates of interest included age, body mass index, systolic blood pressure, hypertension treatment, diabetes, current smoking, total cholesterol, and high‐density lipoprotein cholesterol.

Measurement of Circulating Biomarkers

Plasma concentrations of 85 circulating protein biomarkers were measured using a proteomic discovery platform as part of the SABRe CVD Initiative. Candidate biomarkers were selected based on prior associations with atherosclerotic CVD using genome‐wide association studies, gene expression analyses, or discovery proteomics. ⁹ A total of 85 protein biomarkers were initially assayed using a modified ELISA sandwich approach multiplexed on a Luminex xMAP platform (Sigma‐Aldrich, St. Louis, MO). ¹⁰ , ¹¹ Of the 85 biomarkers, 14 had >25% of samples below the detection limit and were excluded, leaving the remaining 71 biomarkers for our analysis. We have previously described detailed protocols for assay development and measurement (performance characteristics presented in Table S1). ⁹

Clinical Outcomes

The primary outcomes of interest included HF, major CVD, and all‐cause death. Major CVD included nonfatal myocardial infarction, stroke, HF, coronary insufficiency, and cardiovascular death. All outcomes were adjudicated using previously described established FHS protocols. ⁸ , ⁹

Statistical Analysis

Baseline characteristics for women with and without early menopause were compared using Student's t test or chi‐square test. Due to skewed distributions, biomarkers were rank normalized. Biomarkers were then standardized to a mean of 0 and SD of 1. In our primary analyses, we examined the cross‐sectional association of early menopause with biomarker concentrations using linear regression analyses in an age‐adjusted model and multivariable model adjusted for age, systolic blood pressure, hypertension treatment, body mass index, diabetes, smoking status, and total and high‐density lipoprotein cholesterol. In exploratory analyses, we further adjusted for MHT and prior oral contraceptive pill use. We also examined the association of age at menopause as a continuous variable with biomarker levels.

To understand the association of early natural versus surgical menopause on biomarker levels, we performed exploratory analyses evaluating the association of early menopause as a 3‐level categorical exposure variable (natural early menopause, surgical early menopause, and menopause after age 45 years). To further explore surgical menopause subtypes, we examined the association of early menopause as a 4‐level categorical exposure variable (natural early menopause, bilateral oophorectomy, hysterectomy without bilateral hysterectomy, and menopause after 45 years) with biomarker concentrations. To address potential misclassification of women with history of hysterectomy without bilateral oophorectomy, we performed sensitivity analyses excluding those with hysterectomy without bilateral oophorectomy.

For protein biomarkers with both significant and suggestive associations with early menopause, we examined their associations with cardiovascular outcomes (HF, major CVD, and all‐cause death) using multivariable‐adjusted Cox proportional hazards models. We next examined whether early menopause modified the association of biomarkers with outcomes. Specifically, we examined early menopause×biomarker interaction terms using multivariable Cox regression models. If early menopause was determined to be an effect modifier, we performed analyses stratified by early menopause status.

All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, NC). For primary analyses, we set a Benjamini‐Hochberg (SAS PROC MULTTEST) adjusted P value <0.1 as significant and an unadjusted P value <0.05 as suggestive to account for multiple hypothesis testing. ¹² The assumption of positive regression dependency for the Benjamini‐Hochberg method was met. Exploratory analyses relating biomarkers and early menopause×biomarker interactions with outcomes were considered suggestive at unadjusted P<0.05. Inspection of Schoenfeld residuals, Kaplan–Meier curves, and Martingale residuals did not detect major violations of proportional hazards in the Cox regression analyses.

Results

The study sample included 1565 postmenopausal women, of whom 395 (25%) experienced early menopause (Table 1). The overall mean age of the cohort at time of biomarker ascertainment was 62±9 years. The mean age at menopause was 38±5 years among women with early menopause and 50±4 years among women without early menopause. Among women with early menopause, 88 (22%) experienced natural menopause and 307 (78%) experienced surgical menopause (Table 1 and Table S2). Compared with women without early menopause, women with early menopause were more likely to have diabetes (12% versus 8%, P=0.059) and to be current smokers (20% versus 11%, P<0.001). Estimated glomerular filtration rate was higher in early versus without early menopause (73±19 versus 69±17 mL/min per 1.73 m²). MHT use was marginally higher among women with versus without early menopause (38% versus 32%, P=0.033). Prior oral contraceptive pill use and number of live births were similar between women with and without early menopause (Table 1 and Table S3).

Table 1.

Baseline Clinical Characteristics in Women With and Without Early Menopause

	Total (n=1565)	Early menopause (n=395)	Without early menopause (n=1170)	P value
Clinical variables
Age, y	62±9	59±10	62±8	<0.001
Body mass index, kg/m2	27.7±5.8	27.8±6.0	27.6±5.8	0.60
Systolic BP, mm Hg	127±20	124±18	128±20	0.001
Diastolic BP, mm Hg	72±9	72±10	72±9	0.74
Hypertension treatment, n (%)	488 (31)	112 (28)	376 (32)	0.16
Diabetes, n (%)	145 (9)	46 (12)	99 (8)	0.059
Current smoker, n (%)	213 (14)	80 (20)	133 (11)	<0.001
Total cholesterol, mg/dL	209±37	206±38	210±36	0.078
High‐density lipoprotein cholesterol, mg/dL	61±17	60±17	62±17	0.14
Estimated glomerular filtration rate, mL/min per 1.73 m2	70.0±18	73±19	69±17	<0.001
Peripheral artery disease, n (%)	49 (3)	12 (3)	37 (3)	0.90
Menopause‐related variables
Age at menopause, y	47±7	38±5	50±4	<0.001
Natural menopause, n (%)	1091 (70)	88 (22)	1003 (86)	<0.001
Surgical menopause, n (%)	474 (30)	307 (78)	167 (14)	<0.001
Current menopausal hormone therapy use, n (%)	522 (33)	149 (38)	373 (32)	0.033
Prior oral contraceptive pill use, n (%)	730 (47)	196 (50)	534 (45)	0.17

Values are mean±SD or n (%). Between‐group differences were compared using Student's t test or chi‐square test, as appropriate. BP indicates blood pressure.

Associations of Early Menopause With Circulating Levels of CVD Biomarkers

Biomarker profiles differed in women with versus without early menopause (Table 2 and Figure 1, full results are displayed in Table S4). Of the 71 biomarkers examined, we identified 7 biomarkers that significantly differed in women with and without early menopause (Benjamini‐Hochberg adjusted P<0.1) and an additional 6 biomarkers with suggestive associations (P<0.05). Of these 7 biomarkers, 5 were significantly higher in women with early menopause including resistin, adrenomedullin, adipsin, IGFBP1 (insulin‐like growth factor binding protein 1), and APOA1 (apolipoprotein A1). Specifically, early menopause was associated with 0.17‐SD higher level of resistin (multivariable‐adjusted β=0.169, SE 0.057, P=0.003) and a 0.16‐SD greater level of adrenomedullin (β=0.162, SE 0.049, P=0.001). By contrast, 2 of the 7 biomarkers were significantly lower in women with early menopause including insulin‐like growth factor 1 and CNTN1 (contactin‐1). Early menopause was associated with 0.22‐SD lower concentration of insulin growth factor‐1 (β=−0.216, SE 0.057, P<0.001) and 0.16‐SD lower concentration of CNTN1 versus without early menopause (β=−0.156, SE 0.056, P=0.006). In secondary analyses, we examined the association of age at menopause as a continuous predictor of biomarker profiles and confirmed our primary findings (Table S5). Further adjustment for MHT and oral contraceptive pill use in exploratory models yielded similar results to the primary analyses, suggesting that MHT and prior oral contraceptive pill use did not appreciably affect biomarker associations (Tables S6 and S7).

Table 2.

Protein Biomarkers Significantly Associated With Early Menopause

Biomarkers	β‐coefficient†	SE	P value
Higher in early menopause
Resistin*	0.169	0.057	0.003
Adrenomedullin*	0.162	0.049	0.001
Adipsin*	0.148	0.051	0.004
Insulin‐like growth factor binding protein 1*	0.138	0.051	0.007
Apolipoprotein A1*	0.132	0.048	0.006
Beta‐2‐microglobulin	0.130	0.052	0.012
N‐terminal pro‐B‐type natriuretic peptide	0.126	0.052	0.016
Hemopexin	0.125	0.056	0.025
C‐reactive protein	0.121	0.052	0.020
Lower in early menopause
Butrylcholinesterase	−0.117	0.057	0.039
Dipeptidyl peptidase 4	−0.133	0.059	0.024
Contactin 1*	−0.156	0.056	0.006
Insulin‐like growth factor *	−0.216	0.057	<0.001

^*Biomarkers met Benjamini‐Hochberg adjusted P value threshold of <0.1. Remaining biomarkers demonstrated suggestive associations (P value <0.05). For a full list of the 71 proteins, please refer to Table S2. All biomarkers are rank normalized. Original assay units are pg/mL. Multivariable model adjusts for age, body mass index, systolic blood pressure, hypertension treatment, diabetes, current smoking, total cholesterol, and high‐density lipoprotein cholesterol.

^† ß‐Coefficients show beta‐SD differences in rank normalized biomarker between early menopause and reference menopause. Positive ß‐coefficient represents biomarkers that are higher in women with early menopause, and negative ß‐coefficient represents biomarkers that are higher in reference menopause.

Figure 1. Multivariable‐adjusted associations of single biomarkers with early menopause.

Volcano plot displays relative biomarker concentrations in women with and without history of early menopause. Positive x‐values (red) represent biomarkers that are higher in women with early menopause, and negative x‐values (blue) represent biomarkers that are higher in women without early menopause. APOA1 indicates apolipoprotein A1; CNTN1, contactin 1; FDR, false discovery rate; IGF‐1, insulin‐like growth factor 1; and IGFBP1, insulin‐like growth factor binding protein 1.

In exploratory analyses, we evaluated whether biomarker associations differed in women with natural versus surgical early menopause and observed that the associations of adrenomedullin, adipsin, β2M (beta‐2 microglobulin), and hemopexin with natural early menopause were more pronounced (Table S8). By contrast, for biomarkers with lower observed levels in early menopause, associations persisted in women with early surgical menopause but the absolute effect sizes appeared smaller in women with early natural menopause. Examination of early surgical menopause subtypes (bilateral oophorectomy and hysterectomy without bilateral oophorectomy) and sensitivity analyses excluding women with hysterectomy without bilateral oophorectomy demonstrated overall directionally consistent results with our primary analyses (Tables S9 and S10).

Biomarkers of Early Menopause Predict Incident Cardiovascular Outcomes

We investigated the association of early menopause‐related biomarkers with incident cardiovascular outcomes. During follow‐up (median 15 years), 229 postmenopausal women developed incident cardiovascular events, including 92 HF, 191 major CVD, and 298 all‐cause death. Event rates were similar in women with and without early menopause (Table S11). Of the 9 biomarkers that were positively associated with early menopause, 5 were associated with greater risk of future HF (adrenomedullin, adipsin, β2M, NT‐proBNP [N‐terminal pro‐B‐type natriuretic peptide], and CRP [C‐reactive protein]), 4 with major CVD (adipsin, β2M, NT‐proBNP and CRP), and 6 with all‐cause death (resistin, adrenomedullin, adipsin, β2M, NT‐proBNP, and CRP, P<0.05 for all; Table 3). Among the 4 biomarkers that demonstrated negative associations with early menopause, higher levels of DPP4 (dipeptidyl peptidase 4) were associated with lower risk of incident HF (multivariable‐adjusted hazard ratio [HR], 0.80 [95% CI, 0.65–1.00], P value=0.05) and CNTN1 was associated with lower risk of all‐cause death (HR, 0.86 [95% CI, 0.76–0.98], P value=0.02).

Table 3.

Associations of Early Menopause‐Related Biomarkers With Cardiovascular Outcomes

	Heart failure (n=1565)		Major CVD* (n=1545)		All‐cause mortality (n=1565)
	HR‡ (95% CI)	P value	HR‡ (95% CI)	P value	HR‡ (95% CI)	P value
Higher in early menopause
Resistin†	1.18 (0.95–1.47)	0.13	1.04 (0.90–1.21)	0.59	1.22 (1.08–1.37)	0.001
Adrenomedullin†	1.40 (1.09–1.82)	0.01	1.12 (0.94–1.33)	0.21	1.18 (1.02–1.35)	0.02
Adipsin†	1.32 (1.05–1.68)	0.02	1.24 (1.05–1.46)	0.01	1.18 (1.04–1.34)	0.01
Insulin‐like growth factor binding protein 1†	1.12 (0.88–1.42)	0.37	1.07 (0.91–1.27)	0.42	1.01 (0.88–1.16)	0.89
Apolipoprotein A1†	1.19 (0.91–1.56)	0.20	1.10 (0.91–1.32)	0.31	1.01 (0.87–1.16)	0.94
Beta‐2‐microglobulin	1.43 (1.12–1.83)	<0.001	1.36 (1.15–1.61)	<0.001	1.33 (1.16–1.53)	<0.001
N‐terminal pro‐B‐type natriuretic peptide	1.54 (1.22–1.94)	<0.001	1.31 (1.12–1.55)	<0.001	1.33 (1.17–1.52)	<0.001
Hemopexin	1.20 (0.97–1.48)	0.10	1.11 (0.96–1.29)	0.15	1.07 (0.95–1.20)	0.27
C‐reactive protein	1.63 (1.29–2.07)	<0.001	1.25 (1.06–1.48)	0.01	1.24 (1.09–1.41)	0.001
Lower in early menopause
Butrylcholinesterase	1.09 (0.87–1.37)	0.47	1.05 (0.90–1.22)	0.56	0.92 (0.81–1.04)	0.18
Dipeptidyl peptidase 4	0.80 (0.65–1.00)	0.05	0.90 (0.78–1.04)	0.15	0.90 (0.80–1.02)	0.09
Contactin 1†	1.04 (0.83–1.30)	0.74	0.92 (0.79–1.07)	0.29	0.86 (0.76–0.98)	0.02
Insulin‐like growth factor†	0.98 (0.80–1.20)	0.86	1.02 (0.88–1.17)	0.83	0.94 (0.84–1.05)	0.28

Biomarkers displayed demonstrate significant or suggestive associations with early menopause in multivariable‐adjusted models. Multivariable model adjusts for age, body mass index, systolic blood pressure, hypertension treatment, diabetes, current smoking, total cholesterol, and high‐density lipoprotein cholesterol. CVD indicates cardiovascular disease; and HR, hazard ratio.

^*n=1545 individuals (n=387 with early menopause, n=1157 without early menopause) for major CVD analyses.

^† Biomarkers that met Benjamini‐Hochberg adjusted P value threshold of <0.1 in primary analyses.

^‡ HR per 1‐SD increase in rank normalized biomarker is shown.

Early Menopause Modifies the Association of Biomarkers With Cardiovascular Outcomes

We examined whether early menopause modified the association of protein biomarkers with cardiovascular outcomes. We found that early menopause significantly modified the associations with incident cardiovascular outcomes for 6 of 71 biomarkers. These biomarkers included osteocalcin and GRN (granulin) for heart failure, CXL16 (chemokine CXC motif ligand 16), and A1M (alpha‐1‐microglobulin) for major CVD, and ceruloplasmin, adrenomedullin, ANGPTL3 (angiopoietin‐like 3), and APOA1 for all‐cause death, (P _interaction values range from 0.005 to 0.05; Figure 2 and Table S12). We conducted stratified analyses if early menopause was found to modify the effect of biomarker on outcomes. Among women with early menopause, adrenomedullin was significantly associated with all‐cause death (HR, 1.71 [95% CI, 1.30–2.26], P<0.001). Among women without early menopause, this was not the case (HR, 1.03 [95% CI, 0.88–1.22], P=0.69). We also found that GRN was significantly associated with incident HF among women without early menopause (HR, 1.73 [95% CI, 1.35–2.23], P<0.001) but not in those with early menopause (HR, 0.91 [95% CI, 0.56–1.50], P=0.72).

Figure 2. Association of biomarkers with cardiovascular outcomes is modified by early menopause.

Biomarkers displayed show early menopause×biomarker interaction P<0.05. All biomarkers are rank normalized. Original assay units are pg/mL. APOA1 indicates apolipoprotein A1; A1M, alpha‐1‐microglobulin; ANGPTL3, angiopoietin‐like 3; CVD, cardiovascular disease; CXL16, chemokine ligand 16; GRN, granulin; HF, heart failure; and HR, hazard ratio. For major cardiovascular disease analyses, n=1545 individuals (n=387 with early menopause, n=1157 without early menopause).

Discussion

In a community‐based sample of postmenopausal women, we characterized CVD‐related protein biomarkers associated with early menopause. First, we observed that early menopause was associated with circulating levels of CVD biomarkers involved in inflammation (CRP and β2M), adipokine signaling (resistin and adipsin), and neurohormonal regulation (NT‐proBNP and adrenomedullin) pathways. Second, we found that biomarkers that were higher in women with early menopause also predicted incident cardiovascular outcomes, including HF, major CVD, and all‐cause death, whereas biomarkers lower in early menopause were associated with lower risk of cardiovascular events. Lastly, early menopause modifies the association of specific biomarkers with incident cardiovascular outcomes. For example, higher levels of adrenomedullin were associated with all‐cause death among women with early menopause but not in women without early menopause. Taken together, our findings demonstrate that protein biomarkers of early menopause may reflect important pathways implicated in CVD development among women with a history of early menopause (Figure 3).

Figure 3. Early menopause is associated with alterations in cardiovascular biomarkers.

Cardiovascular biomarkers differ in women with and without a history of early menopause and can influence cardiovascular outcomes in 2 ways. First, intrinsic differences in circulating cardiovascular disease biomarker levels between women with and without early menopause may contribute to greater cardiovascular risk associated with early menopause. Second, early menopause may modify the effect of biomarkers on cardiovascular outcomes. CVD indicates cardiovascular disease.

Mechanisms underlying the association between early age of menopause and incident CVD are not well characterized. Previous investigations have focused on the early loss of protective effects of endogenous estrogen as its decline has been hypothesized to explain adverse changes in cardiometabolic health and precipitous rise of CVD in women following menopause. ¹³ Longer duration of estrogen deficiency and associated cardiometabolic changes may explain higher cardiovascular risk in women with premature or early menopause. In keeping with this hypothesis, treatment with estrogen in clinical trials improved lipid profiles and insulin sensitivity. ¹⁴ , ¹⁵ However, improvements in subclinical markers of CVD did not translate to clinical outcomes as administration of exogenous estrogen in randomized trials paradoxically increased cardiovascular events. Notably, age at time of MHT initiation may influence the effect of MHT on cardiovascular outcomes as clinical outcome trials of MHT were conducted in older women (mean age >60 years). When examining MHT use in younger women, the findings are more nuanced. Post hoc analyses of 4476 women <50–59 years of age did not find an increase in CVD events with MHT use, and among women with premature or early menopause, MHT use was associated with fewer cardiovascular events in observational studies. ⁴ , ¹⁶ These data underscore the complex relationship between estrogen deficiency, menopause, and cardiovascular risk and also highlight the need to interrogate pathways orthogonal to estrogen deficiency to better understand the relationship between early menopause and CVD. ⁵ , ⁶ Recent investigations have identified an independent association of premature menopause with clonal hematopoiesis of indeterminate potential, a recently recognized risk factor for accelerated atherosclerosis. ¹⁷ Beyond clonal hematopoiesis of indeterminate potential, genetic variants involved in DNA repair and immune pathways have also been shown to associate with age at menopause, highlighting the potential role of accelerated aging and inflammation in mediating the relationship between premature menopause and CVD. ¹⁸ , ¹⁹

In the present study, we sought to advance mechanistic understanding of the association of early menopause with CVD by assessing an expanded panel of CVD‐relevant protein biomarkers in a large, well‐phenotyped cohort of postmenopausal women. Our findings implicate inflammation, adipokine signaling, and neurohormonal regulation as potential biologic pathways that contribute to the development of CVD in women with early menopause. Inflammatory biomarkers have previously been associated with early menopause, but the available data are mixed and limited to select biomarkers. For example, previous studies have demonstrated higher high sensitivity CRP levels among women with early and premature menopause. ²⁰ , ²¹ By contrast, a case–control study examining inflammatory markers in women with and without early menopause observed a significant association of soluble fraction of tumor necrosis factor‐alpha receptor 2 levels with risk of early menopause, but no relationship between CRP and interleukin‐6 and early menopause. Finally, higher levels of inflammatory biomarkers including ceruloplasmin, complement C3, and fibrinogen ⍺ and β have been described in women with primary ovarian insufficiency, an important cause of natural premature menopause. ²² , ²³ We again showed an association between CRP and early menopause, but the association was only suggestive. We did identify a novel suggestive association between inflammatory marker β2M and early menopause. β2M is a component of the major histocompatibility complex I molecule, which regulates immune surveillance and inflammation. ²⁴ A previous investigation using the same SABRe platform found that higher CRP and β2M levels predicted incident HF, CVD‐related death, and all‐cause mortality among ostensibly healthy men and women. ⁹ Our results suggest that inflammation may play an important role in the development of CVD in women with early menopause. We observed that levels of CRP and β2M are both higher and are also independently associated with HF, major CVD, and all‐cause death in women with early menopause. These findings build upon existing literature that has linked CRP and β2M to CVD, particularly in women. ²⁵ , ²⁶ It is unclear to what extent estrogen deficiency contributes to the association between inflammation and early menopause. ²⁷ Estrogen downregulates inflammatory cytokines and enhances nitric oxide production, ²⁷ but whether early menopause promotes systemic inflammation independent of estrogen deficiency is unknown and warrants further investigation.

We separately observed higher adipokines resistin and adipsin levels in women with early menopause. Resistin and adipsin are both adipokines implicated in mediating the relationship between obesity, insulin resistance, diabetes, and inflammation. ²⁸ Higher levels of resistin, adipsin, and other adipokines have been observed in postmenopausal women, particularly those with metabolic syndrome. ²⁹ Notably, the menopause transition is accompanied by changes in adipokine profiles and adipose tissue distribution. ³⁰ In a cohort study of women through the menopause transition, changes in adipose tissue distribution were accompanied by increases in adiponectin, leptin, and CRP. ³¹ Whether similar changes in adipokine profile and adipose tissue distribution occur in women with early menopause has not been examined. Adipokines have previously been associated with adverse cardiovascular outcomes in postmenopausal women. ³² We too demonstrate independent associations of adipsin and resistin with the development of HF, major CVD, and all‐cause death in postmenopausal women. Our findings suggest that adipose regulatory pathways may be upregulated in women with early or premature menopause. Prolonged estrogen deficiency may again play a role as estrogen regulates body fat distribution and adipocyte differentiation. ³³ Alternatively, inflammation itself may promote upregulation of adipokines and related adipose tissue redistribution and vice versa as previous longitudinal data support a significant correlation between inflammatory markers and postmenopausal visceral adiposity. ³¹

We also found that biomarkers involved in neurohormonal regulation were strongly associated with early menopause, with higher levels of NT‐proBNP and adrenomedullin observed in women with early menopause. These findings are consistent with results from a multiethnic cohort that previously demonstrated greater NT‐proBNP levels in women with early menopause. ³⁴ However, another study of 4352 postmenopausal women found no difference in NT‐proBNP levels between women with and without early menopause. ²⁰ In healthy individuals, neurohormonal homeostasis is regulated by opposing effects of the classical and nonclassical renin‐angiotensin‐aldosterone systems (RAAS). Classical RAAS activation promotes vasoconstriction and fluid retention, whereas activation of the nonclassical RAAS inhibits the classical RAAS pathway. ³⁵ Estrogen potentiates nonclassical RAAS activation, decreasing vascular dysfunction and inflammation. ³⁶ However, estrogen deficiency does not appear to explain higher NT‐proBNP levels in women with early menopause as previous studies have demonstrated higher NT‐proBNP concentrations in premenopausal women compared with postmenopausal women and men, thought to be driven in part by higher levels of testosterone in postmenopausal women and men. ³⁶ With respect to outcomes, NT‐proBNP is a well‐known marker of cardiovascular outcomes in the general population and among patients with HF. ³⁷ In this sample of postmenopausal women, we again confirm that NT‐proBNP is associated with incident HF, major CVD, and all‐cause death. Although few studies have examined the prognostic value of NT‐proBNP in women with early menopause, NT‐proBNP was preferentially associated with a greater risk of incident HF with reduced ejection fraction in women with versus without early menopause in the Atherosclerosis Risk in Communities study. ²⁰

Finally, adrenomedullin may represent a biomarker of particular importance in the setting of early menopause. Adrenomedullin is a vasoactive peptide derived by endothelial cells that maintains vascular tone and endothelial barrier integrity, secreted in response to increased vascular permeability and tissue edema. ³⁸ Higher levels of adrenomedullin have been observed in CVD states, including hypertension, CAD, and HF, and are thought to exert compensatory beneficial effects by reducing volume overload and tissue congestion. ³⁹ Recent bidirectional Mendelian randomization analyses indicate that higher genetically determined adrenomedullin levels are causally protective against HF but that adrenomedullin is further increased in response to HF. ⁴⁰ Moreover, adrenomedullin is being investigated as a potential therapeutic target for HF in ongoing clinical trials. ⁴¹ We found that levels of adrenomedullin were higher in women with early menopause and that adrenomedullin was preferentially associated with all‐cause death in women with versus without early menopause. Little is known about the effect of menopause on adrenomedullin regulation, but sex hormones are known to modulate adrenomedullin levels. In experimental studies, testosterone directly increased the percentage of human endothelial cells that secrete adrenomedullin, whereas estrogen inhibited the stimulatory effect of angiotensin‐II on adrenomedullin‐secreting endothelial cells. ⁴² Whether sex hormones explain why early menopause influences both adrenomedullin levels and the association with all‐cause death is unknown, but our findings highlight the potential role of adrenomedullin‐related pathways in the development of adverse clinical outcomes related to early menopause.

Notably, the prevalence of early menopause was 25% in our sample, substantially higher than the 5% to 10% prevalence reported in previous studies due in part to the inclusion of surgical early menopause in addition to natural menopause. ⁴ In exploratory analyses, we observed heterogeneity of biomarker associations across menopause forms. For example, associations of adipsin, adrenomedullin, and β2M levels with early menopause were more pronounced among women with natural early menopause than surgical menopause. By contrast, associations for biomarkers that were lower in early menopause were no longer evident when analyses were restricted to women with natural early menopause. Collectively, our exploratory findings highlight that early menopause‐associated CVD risk cannot be explained by deficiency of ovarian‐derived estrogen alone and suggest that biologic pathways (eg, inflammatory, adiposity, and neurohormonal regulation) may be differentially activated across menopause forms or represent shared upstream risk factors for both early menopause and subsequent CVD. Previous investigations have proposed that differences in baseline clinical profiles may explain cardiovascular risk in natural versus surgical menopause. ²¹ , ⁴³ , ⁴⁴ Importantly, our exploratory findings were not explained by clinical characteristics, highlighting distinct biological pathways in women with natural versus early surgical menopause. Whether these identified biological pathways contribute to the development of CVD in women with natural early menopause or play a role in the pathogenesis of early menopause itself is unknown.

Limitations

Our study has several limitations. First, age at menopause was self‐reported, which may contribute to misclassification bias. Second, suggestive findings that were not adjusted for multiple hypothesis testing should be interpreted as exploratory and hypothesis generating. Third, the limited sample size precluded our assessment of proteomic profiles of premature menopause, defined as menopause before age 40. Premature menopause more strongly associates with CVD risk than early menopause; examining this association may have yielded additional biomarker differences. Fourth, we included women who underwent hysterectomy but not bilateral oophorectomy in the surgical menopause group. We acknowledge that this approach does not capture the complex interplay between ovarian and uterine function. Notably, the biomarker profile of women who underwent hysterectomy without bilateral oophorectomy closely mirrored the profile of women who underwent bilateral oophorectomy. Fifth, biomarkers were measured at only 1 time point, precluding interrogation of longitudinal changes in biomarkers. Characterizing changes in proteomic profiles through the menopause transition may offer further insights into the unique biologic pathways that contribute to CVD in women in early menopause. Moreover, biomarkers were measured at the time of the SABRe CVD Initiative, not during menopause. Biomarker profiles ascertained contemporaneous with the menopause transition may offer additional insights into the pathogenesis of early menopause and associated CVD. Sixth, our sample was predominantly White, and our findings may not generalize to diverse racial and ethnic populations. Finally, the observational nature of our study limited our ability to draw causal inferences and we cannot exclude the possibility of residual confounding. When modeling age as a spline representation, or including other interaction terms, we found that results remained largely consistent.

Conclusions

In a community‐based sample of postmenopausal women, we found that early menopause was associated with alterations in CVD protein biomarkers, including higher circulating levels of biomarkers representing inflammatory, adiposity, and neurohormonal regulatory pathways, which in turn were associated with adverse cardiovascular outcomes. Further, early menopause appears to modify the effect of specific vasoactive and inflammatory biomarkers on outcomes, including adrenomedullin. Taken together, these findings highlight distinct biological pathways that may underlie greater CVD susceptibility among women with early menopause. Further studies that improve mechanistic understanding of CVD in women with early menopause may guide future preventive and therapeutic strategies to address early menopause‐associated CVD.

Sources of Funding

The Framingham Heart Study is supported by grants from the National Institutes of Health N01‐HC25195 and HHSN268201500001I. Dr Honigberg is supported by grants from the National Heart, Lung, and Blood Institute K08HL166687, and the American Heart Association 940166 and 979465. Dr Ho is supported by grants from the National Institutes of Health R01‐HL134893, R01‐HL140224, R01‐HL160003, and K24‐HL153669. Dr Lau is supported by grants from the National Institutes of Health K23‐HL159243 and the American Heart Association 853922. The views expressed in this article are those of the authors and do not necessarily represent the view of the National Heart, Lung, and Blood Institute; the National Institutes of Health; or the US Department of Health and Human Services.

Disclosures

Dr Honigberg reports consulting fees from CRIPSR Therapeutics, advisory board services for Miga Health, and research support from Genentech, all unrelated to this work. Dr Lau reports past modest honoraria from Roche Diagnostics and advisory board services for Astellas Pharma. The remaining authors have no disclosures to report.

Supporting information

Tables S1–S12