Plasma Metabolites, Blood-borne Microvesicles and History of Preeclampsia as predictors of Coronary Artery Calcification in Postmenopausal Women

Abstract

Background:

Compared to men, women with similar coronary artery calcification (CAC) scores face higher cardiovascular disease (CVD) mortality.

Objectives:

We posited that circulatory factors, like blood-borne extracellular vesicles (EVs) and metabolites, may be associated with the development of CAC and subsequent CVD in postmenopausal women. Additionally, we hypothesized that a history of preeclampsia (PE), a sex-specific risk factor, might be a contributing factor.

Methods:

Clinical data were obtained from medical records for postmenopausal women (median age 60 years) free of cardiovascular events with (n=29) and without (n=29) CAC. EVs per μL plasma were quantified by digital flow cytometry, and plasma metabolites were measured using gas chromatography-mass spectrometry. CACs were measured by computer tomography and reported as Agatston score.

Results:

Patients with, versus those without, CAC demonstrated i) less favorable cardiovascular and metabolic profiles; ii) elevation in six EVs populations, including those positive for TF (tissue factor), CD3 (T-cells), SM22α (smooth muscle cells), Pref-1 (adipocytes), FABP4 (adipocytes/macrophages) and p16 (senescent cells); iii) significantly higher levels of proline, allothreonine (amino acid metabolism) and ribitol (carbohydrate metabolism), and lower levels of lactic acid (carbohydrate metabolism); and iv) significantly increased risk of developing CVD and chronic kidney disease (CKD) (P<0.05 for all). In the CAC positive group, women with PE vs those with normotensive pregnancy histories, demonstrated 4-times higher risk of developing cardiovascular events or CKD later in life (P=0.028).

Conclusions:

Selected plasma metabolites, EVs and PE history could serve as biomarkers of and potential therapeutic targets for CAC and CVD in postmenopausal women.

Graphical Abstract

INTRODUCTION

Cardiovascular diseases (CVD) are the leading cause of mortality in women worldwide. Coronary artery calcification (CAC), which commonly develops simultaneously with advanced atherosclerosis,¹ is a strong predictor of future major adverse cardiac events and stroke. Compared to men, who tend to have higher CAC scores, women face higher CVD mortality than men despite similar CAC scores.^2,3 The reason(s) for these sex-differences remain largely unknown. On a mechanistic level, CAC was viewed historically as a passive, unregulated, degenerative consequence of atherosclerosis and aging, a premise that is contested by emerging evidence demonstrating the complex nature of coronary calcifications, involving both passive and active (cellular specific) processes (see Figure 1).^4–6

Figure 1.

Risk factors and pathogenesis of coronary artery calcification (CAC) In addition to atherosclerosis, factors initiating and accelerating CAC include chronic kidney disease (CKD), metabolic syndrome, obesity, diabetes, genetic predisposition, hypertension, chronic inflammation, aging, being post menopause and history of preeclampsia (PE).¹¹ The key points in CAC formation are endothelial dysfunction, accumulation of lipids, inflammation, plaque formation, activation of smooth muscle cells, osteogenic differentiation of smooth muscle cells and deposition of calcium phosphate in the form of hydroxyapatite. Vascular cells derived from activated smooth muscle and circulating hematopoietic cells undergo chondrogenic or osteogenic transformation, resulting in mineralization of membranous bone and the creation of endochondral bone.^6,66 EV – extracellular vesicle; PE – preeclampsia; ALP – alkaline phosphatase; *major novel findings from this study; MACE – major adverse cardiovascular events; DM-Diabetes Mellitus.

Calcifications start within the necrotic core of an atherosclerotic plaque in inflamed regions of the vascular wall where local collagen fibers are decreased. Studies to date have implicated blood-borne extracellular vesicles (EVs), small (0.03–1 μm) double membrane-enclosed sacs shed from activated or injured cells, in the process of calcification in atherosclerotic plaques. Notably, a high resolution microscopic and spectroscopic analysis of calcified human plaques demonstrated that the aggregation of calcifying EVs leads to macrocalcifications in human atherosclerotic plaques.⁷ These changes in the local microenvironment of the atheroma are exaggerated by metabolic factors that disrupt local regulatory control pathways of calcium and phosphate concentrations, thus contributing to the formation of calcium phosphate crystals or calcifications.^6,8

The mechanisms and markers of cellular damage and their interactions with a pro-inflammatory and metabolic milieu in the development of CAC are not fully understood, despite significant research efforts in this field. Although CAC scoring is a valuable tool for assessing cardiovascular risk, there are several barriers that limit wide spread use, including limited access, insurance coverage issues, susceptibility to small variations in noise, and concerns about radiation exposure.^9,10

The aim of this study was to identify EVs and metabolite markers that are associated with detectable CAC in asymptomatic postmenopausal women, which can inform personalized screening and management strategies, and may set the stage for future studies of pathophysiological mechanism(s) of CAC. We also sought to investigate if a history of preeclampsia (PE), a pregnancy specific hypertensive disorder that has been associated with future CVD,^11,12 leaves an additional metabolic imprint, thus further increasing CVD risks in the presence of CAC.

METHODS

Study Participants

This study was approved by the Institutional Review Board at Mayo Clinic, Rochester, MN. Age and parity matched women who gave birth between 1976 through 1982 with (n=40) and without (n=40) a history of PE were identified through the Rochester Epidemiology medical records linkage system, as previously described.¹³ Medical records were screened to confirm no prior history of CVD (such as myocardial infarction and congestive heart failure), stroke, dementia, any cancer (except for nonmelanoma skin cancer), autoimmune disease (e.g., systemic lupus erythematosus), and neurological conditions (e.g., multiple sclerosis). A flowchart of inclusion and exclusion criteria, allocation to the groups and timeline of the experiments is shown in Figure 2, while demographic and clinical characteristics of the study subjects at baseline are reported in Table 1. All participants were white and gave written informed consent to enroll in the study between 2011 and 2013, at which point all blood and urine samples were obtained and all imaging was performed. Coronary artery calcification (CAC) images were obtained using a 64-detector computed tomography scanner (Siemens Sensation 64, Siemens Medical Solutions, Forcheim, Germany) with a scan configured to cover the heart. Plasma and urine from a subset of patients included in this study with CAC >0 Agatston units (n = 29, median age = 60 years) and age-matched controls with CAC=0 Agatston units (n = 29, median age = 60 years) were used for metabolomics and blood-borne EV analysis, see Figure 2 and Figure 3. To ensure blinded analyses, all clinical identifiers were removed before sample processing and all the experiments. We also collected information via chart review regarding coronary artery disease (CAD), congestive heart failure (CHF), transient ischemic attack (TIA), stroke, and CKD events that occurred between the time of a woman’s enrollment in the study through 2021. Follow-up was censored at the time of a woman’s last visit to a Rochester Epidemiology Project-affiliated provider or death, whichever came first.

Figure 2.

Flowchart of Inclusion and Exclusion Criteria

Table 1.

Demographic and Clinical Characteristics of Study Participants at Baseline, by CAC Status

Patient characteristic	No CAC (N=29)	CAC (N=29)	P
Age (years)	60 (57, 63)	60 (57, 63)	0.86
BMI (kg/m2)	25.2 (23.3, 33.0)	31.4 (29.2, 33.7)	0.004
Waist circumference (cm)	84 (79, 99)	101 (95, 105)	<0.001
Systolic blood pressure (mm Hg)	133 (124, 140)	132 (119, 139)	0.47
Diastolic blood pressure (mm Hg)	76 (70, 84)	80 (76, 83)	0.25
Current Hypertension, n (%)	12 (41.4%)	15 (51.7%)	0.43
Hyperlipidemia, n (%)	8 (27.6%)	11 (37.9%)	0.40
Total cholesterol (mg/dL)	191 (176, 216)	195 (175, 218)	0.71
LDL cholesterol (mg/dL)	113 (92, 130)	106 (93, 136)	0.93
HDL cholesterol (mg/dL)	65 (47, 76)	52 (42, 64)	0.12
Triglycerides (mg/dL)	97 (77, 120)	106 (90, 150)	0.12
Fasting glucose (mg/dL)	94 (91, 102)	101 (91, 115)	0.079
Insulin (μlU/mL)	4.5 (3.5, 6.7)	7.4 (5.5, 15.4)	0.001
HOMA-IR	1.09 (0.84, 1.56)	1.92 (1.33, 4.60)	0.001
High sensitivity CRP (mg/dL)	0.093 (0.052, 0.226)	0.186 (0.138, 0.258)	0.015
Hemoglobin A1C (%)	5.40 (5.20, 5.60)	5.60 (5.30, 5.90)	0.031
Alkaline phosphatase (U/L)	62 (57, 74)	76 (64, 88)	0.018
Leptin (ng/mL)	10492 (5369, 22971)	20018 (12058, 26731)	0.022
Adiponectin (ng/mL)	149188 (61702, 246603)	74502 (33817, 174232)	0.075
Adiponectin/Leptin ratio	14.0 (5.1, 24.0)	3.8 (1.6, 6.3)	0.003
Plasma uromodulin (ng/ml)	161 (98, 228)	126 (97, 156)	0.12
Urine uromodulin (μg/ml)	14.7 (10.2, 20.4)	14.4 (10.3, 20.5)	0.96*
History of PE, n (%)	11 (37.9%)	19 (65.5%)	0.036

Abbreviations: BMI=body mass index; CAC=coronary artery calcification; HDL=high-density lipoproteins; HOMA-IR=homeostatic model assessment-insulin resistance; LDL=low-density lipoproteins; CRP=C-reactive protein.

Continuous variables were summarized as median (IQR) and categorical variables were summarized as n (%). Comparisons across groups were evaluated using the Wilcoxon rank sum test for continuous variables and the Chi-square test for categorical variables.

^*For urine uromodulin, 18 samples were available for No CAC and 22 for the CAC group.

Figure 3.

Representative scans of postmenopausal women with and without coronary artery calcification (CAC) measured in Agatston units

*NT - normotensive

Chemicals and Reagents

Myristic acid d27 and O-methoxyl-amine were purchased from Sigma- Aldrich, St. Louis, MO. Pyridine and N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA)/1%TMCS were purchased from Thermo, Rockford, IL. All other reagents and solvents are analytical/reagent grade.

Identification of Blood-borne Extracellular Vesicles

Standardized and published methods for EV isolation from protease inhibitor anti-coagulated blood by differential centrifugation, and their detection, characterization, and quantification using digital flow cytometry, have been described in detail previously.^14,15,67 The concentrations of blood-borne EVs are expressed as EV/μL plasma. A set of 21 EV markers was tested based on the rationale that chronic inflammation and/or oxidative stress create a microenvironment arising from activation of and interactions between the blood elements, endothelium and smooth vascular muscle, which subsequently initiate processes resulting in vascular calcification.

Identification of Metabolites by Gas Chromatography-Mass Spectrometry (GC-MS)

Blood was collected into EDTA anticoagulant added tubes after an overnight fast. Blood samples were centrifuged, and the plasma was frozen at −80°C until use. Samples were thawed at 4°C for 5 minutes and 100 μL of each sample was aliquoted into four separate 1.5 mL Eppendorf tubes for GC-MS, according to a protocol adapted from Dunn WB et al, 2011.¹⁶ Briefly, a pooled sample of all participants’ plasma (100 μL per participant) was used as an internal quality control. GC-MS samples were prepared with 1:3 (sample: methanol) and then centrifuged at 13,300g for 15 minutes to precipitate and pellet the protein. The supernatant was transferred to a new 1.5 mL Eppendorf tube to which 20 μL of internal standard solution was added and dried using a centrifugal vacuum evaporator. For analysis, O-methoxyl-amine (20 μL) prepared in a pyridine solution was added to the dried samples and incubated at 30°C for 90 min. Then, 80 μL of MSTFA/1% TMCS solution was added and the samples were incubated for an additional 30 minutes at 37°C. After cooling, each sample was centrifuged at 13,300g for 15 minutes. The supernatant was transferred to a 200 μL insert inside a 2 mL vial with a sealed screw cap and loaded into the GC-MS for analysis. Metabolites were normalized with an internal standard (myristic-d27 acid) and results are expressed as relative amounts.

ELISA tests

Plasma and urine uromodulin levels were quantified using a commercially available ELISA kit (Bio vendor, Brno, Czech Republic) according to manufacturer instructions. Adiponectin and leptin were measured by Milliplex map Human Adipocyte Magnetic Bead Panel (catalog no. HADK1MAG-61K). The adiponectin/leptin ratio (A/L ratio) was calculated as described previously.¹⁷

Statistical Analysis

Demographic, clinical, and laboratory characteristics, as well as blood EVs at baseline, were summarized as median with 25th and 75th percentiles (IQR; interquartile range) for continuous variables, and as n (%) for categorical variables; comparisons of these characteristics across CAC status were evaluated using the Wilcoxon rank sum test and the Chi-square test, as appropriate. Similar analyses were also performed comparing women with and without PE, stratified by CAC status. Due to the non-normality of the metabolite distributions, they were transformed for all subsequent analyses using a log base 2 transformation. Association analysis between CAC status with each metabolite was performed using ANCOVA models adjusted for body mass index (BMI), HOMA-IR, and C-reactive protein (CRP), with mean and standard deviation (SD) reported for each metabolite among patients with and without CAC, and the false discovery rate (FDR) controlled for at the 0.05 level by using the Benjamini–Hochberg procedure.

Cumulative incidence of the composite event of CAD, CHF, TIA, stroke, or CKD was estimated using the Kaplan-Meier method. The association between CAC status and the incidence of the composite event was evaluated using a Cox proportional hazards model. Two patients with CKD at the time of enrollment were excluded from the analysis to evaluate de novo conditions. The cohort was followed from the enrollment date to the date of the condition diagnosis or was censored at the time of the last clinical visit recorded in the Rochester Epidemiology Project or death, whichever came first. Models were fit using age as the time scale, with women entering the risk set at their ages at study enrollment. Similar models also were fit by evaluating the effect of PE on the cumulative incidence of CAD, CHF, TIA, stroke, or CKD stratified by CAC status.

To evaluate the combined effects of plasma metabolite and EV panels on the risk of future major cardiovascular events or CKD, we conducted principal component analyses (PCA) separately on the log base2 -transformed EV panel plasma and metabolite panel to get reduced representations of each of these panels (principal components). We then fit Cox proportional hazards regression models using the 1st principal component as the predictors for each (due to the limited number of events in our sample); we utilized the first principal component because this component explains the most variance from the original data. Results were considered statistically significant at the 0.05 alpha level. All analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Study subjects

There were 58 patients retained in our cohort for analysis (n=29 with CAC greater than 0 and n=29 with CAC equal 0, see Figure 2). Comparisons of demographic and clinical characteristics of the study participants at baseline across CAC status are reported in Table 1. Patients with CAC had significantly higher body mass index (BMI), larger waist circumference, and higher levels of insulin, Homeostasis Model Assessment for Insulin Resistance (HOMA-IR), leptin, CRP, hemoglobin A1C and alkaline phosphatase (ALP), with lower A/L ratios, compared to patients without CAC (P<0.05 for all).

Blood-borne EVs in CAC positive vs. CAC negative subjects

There were six types of EVs detected in plasma that were significantly elevated in women who were CAC >0 Agatston units, compared to those who were negative (CAC=0 Agatston unit). These included tissue factor (TF) positive (median [IQR] 17.89 [10.19, 36.45] vs 10.24 [7.09, 17.25] in those with and without CAC, respectively, P=0.007), T-cell (CD3)- (7.85 [4.29, 16.95] vs 4.97 [3.70, 7.61], P=0.023), smooth muscle cell (SM22α)- (1.69 [1.13, 3.07] vs 0.78 [0.57, 1.49], P<0.001), adipocyte (Pref-1)- (9.78 [5.32, 15.66] vs 6.59 [4.21, 9.44], P=0.031), adipocyte/macrophage (Fatty Acid-Binding Protein 4 (FABP4)- (1.69 [1.13, 3.06] vs 0.78 [0.58, 1.49], P<0.001) and senescent cell (p16)-derived EVs (1.06 [0.69, 2.08] vs 0.67 [0.44, 1.30], P=0.033), as shown in Table 2. Additionally, although not statistically significant, three other marker positive EVs displayed strong trends of elevation in the CAC positive group, including dendritic cell/macrophage (CD11c)-, B-cell (CD19)- and endothelium (CD62E)- EVs, compared to women without CAC.

Table 2:

Characterization of Blood-borne Circulating EVs between CAC Negative and Positive Women

Microvesicles (EV)/ μL plasma	No CAC (N=29)	CAC (N=29)	P value
Procoagulant EV
Phosphatidylserine positive	822.3 (568.9, 1178)	1014 (716.6, 1441)	0.35
Tissue factor positive	10.24 (7.09, 17.25)	17.89 (10.19, 36.45)	0.007
Anticoagulant EVs/ μL plasma
Tissue factor pathway inhibitor positive	1.72 (0.93, 2.58)	1.76 (1.18, 3.80)	0.39
Cellular adhesion / inflammatory molecule positive EVs/ μL plasma
ICAM-1 positive	3.51 (2.25, 4.77)	4.08 (2.90, 7.82)	0.13
VCAM-1 positive	1.43 (0.81, 2.01)	1.76 (0.98, 2.17)	0.31
MCP-1 positive	2.24 (0.98, 3.47)	2.75 (1.57, 3.71)	0.34
Specific cell of origin of EVs/ μL plasma
Erythrocyte (CD235a)-derived	23.69 (13.06, 44.72)	32.97 (19.30, 47.27)	0.42
Leukocyte (CD45)-derived	15.91 (9.35, 19.82)	14.61 (10.7, 22.9)	0.64
Monocyte (CD14)-derived	1.40 (0.23, 2.34)	2.09 (0.60, 3.13)	0.21
Dendritic-cell and macrophage (CD11c)-derived	3.65 (2.19, 7.31)	6.26 (4.03, 8.12)	0.069
T-cell (CD3)-derived	4.97 (3.70, 7.61)	7.85 (4.29, 16.95)	0.023
B-cell (CD19)-derived	2.09 (1.21, 3.12)	2.59 (1.62, 4.92)	0.067
Platelet (CD42a)-derived	661.9 (526.6, 1024)	779.3 (605.3, 1205)	0.23
Endothelium (CD62E)-derived	5.37 (2.57, 7.03)	6.73 (4.23, 11.63)	0.074
Smooth muscle cell (SM22α)-derived	0.78 (0.57, 1.49)	1.69 (1.13, 3.07)	<0.001
Adipocyte (Pref-1)-derived	6.59 (4.21, 9.44)	9.78 (5.32, 15.66)	0.031
Adipocyte/Macrophage (FABP4)-derived	0.78 (0.58, 1.49)	1.69 (1.13, 3.06)	<0.001
Stem/progenitor cells (CD117)-derived	6.03 (3.40, 8.45)	6.23 (3.95, 9.59)	0.51
Senescent cells (p16-set)-derived	0.67 (0.44, 1.30)	1.06 (0.69, 2.08)	0.033
Exosome marker positive EVs/ μL plasma
Tetraspanin (CD9) positive	733.9 (470.1, 1006)	837.1 (578.3, 1150)	0.36
Tetraspanin (CD63) positive	31.68 (20.44, 46.87)	28.21 (18.33, 41.22)	0.52

Variables were summarized as median (IQR). Comparisons across groups were evaluated using the Wilcoxon rank sum test.

Plasma metabolites in CAC positive vs. CAC negative subjects

Summaries of the log-base-2-transformed plasma metabolite levels in patients with and without CAC are reported in Supplementary Table 1, and sorted by statistical significance at the FDR <0.05 level on the basis of the results from the ANCOVA models. Women with CAC exhibited significantly higher plasma levels of proline (amino acid metabolism), allothreonine (amino acid metabolism) and ribitol (carbohydrate metabolism), and lower levels of lactic acid (carbohydrate metabolism), compared to women without CAC after adjustment for BMI, HOMA-IR and CRP (see Supplementary Table 1 and Figure 4). Prior to the FDR correction, significant associations were also found between CAC status and the plasma metabolites, trans 13 octadecenoic acid, alpha ketoglutaric acid, uric acid, glycerol 1-phosphate, tryptophan, succinic acid, fumaric acid and malic acid, (see Supplementary Table 1 and Figure 5).

Figure 4.

Significant adjusted associations between plasma metabolite levels and CAC status

*Metabolites were normalized with an internal standard (myristic-d27 acid) and results are expressed as relative amounts.

Figure 5.

Significant associations between plasma metabolite levels and CAC status prior to FDR correction.

*Metabolites were normalized with an internal standard (myristic-d27 acid) and results are expressed as relative amounts.

The association of CAC, EVs and metabolites with long-term outcomes

Among the 56 women without prevalent CKD at enrollment (2 women with CKD were excluded to account for known CVD risks associated with kidney disease), there was a total of 22 incident composite events (13 with CAD, 1 with CHF, 2 with TIA/stroke, and 6 with CKD) over a mean (SD) follow-up of 6.2 (0.3) years. Women with CAC demonstrated an increased risk of developing one of the events (CAD, CHF, TIA, stroke, or CKD), compared to women without CAC (hazard ratio, 5.87 [95% CI, 1.94–17.7]; P=0.002; see Figure 6). The first principal component derived from the EV panel was not significantly associated with the risk of developing one of these composite events (P=0.12), while the first principal component derived from the metabolite panel was found to be significantly associated with the risk of developing CAD, CHF, TIA, stroke, or CKD (P=0.030).

Figure 6.

Cumulative incidence of time to the composite event of coronary artery disease (CAD), chronic heart failure (CHF), transient ischemic attack (TIA)/ stroke, or chronic kidney disease (CKD), by CAC status over a mean (SD) follow-up of 6.2 (0.3) years.

The effects of a history of PE on CAC and long-term outcomes

Patients with CAC were found to have higher rates of PE compared to patients without CAC (19 (65.5%) vs 11 (37.9%), respectively, P=0.036), as our prior studies have demonstrated.¹¹ Among patients with CAC, women with a history of PE were found to have higher rates of current hypertension, hyperlipidemia, higher insulin and HOMA-IR levels, and lower levels of plasma uromodulin, compared to women with normotensive pregnancies (see Supplementary Table 2A and Figure S1). This unfavorable cardiometabolic profile was highly clinically relevant: among women with CAC (n=29), those with a history of PE (n=8) were at a 4-time higher risk of developing one of the major cardiovascular events or CKD, compared to normotensive women (hazard ratio, 4.21 [95% CI, 1.17-15.13]; P=0.028), although due to the small sample size, we were unable to obtain robust adjusted estimates. Due to similar issues, including the low event rate in women without CAC, models were not estimable in this subgroup.

DISCUSSION

Our study presents several novel findings regarding CAC and CVD in postmenopausal women. First, we report associations between various clinical parameters, blood-borne EVs with direct or indirect calcifying properties and pro-inflammatory/pro-oxidant metabolites and CAC/CVD, thus identifying circulating biomarkers that might be utilized for diagnosis of early CAC. Second, we show that women with CAC had a 6-fold risk for CVD and CKD compared to women without CAC after a mean of 6.2 years of follow-up. Third, the levels of 6 blood borne EVs, were significantly elevated in CAC-positive women, supporting their possible role in the formation and growth of calcified atherosclerotic plaques. Fourth, women with CAC demonstrated changes in the levels of 4 metabolites that may contribute to atherosclerotic plaque burden generation and calcification. In addition to the association with CAC, the combined effect of these metabolites was associated with an elevated risk for CVD and CKD later in life. Fifth, we show that in CAC positive women, a history of PE compared to a history of normotensive pregnancy, resulted in an increased risk for CVD and CKD later in life. This risk is likely mediated by higher rates of known CVD risk factors in women with a history of PE. Sixth, CAC positive women with a history of PE demonstrated lower levels of plasma uromodulin, an anti-inflammatory compound, lower levels of which have been previously associated with CVD and CKD, but not in the context of PE.

Women with CAC in this study, as expected, demonstrated a significantly increased risk of developing the composite event of CAD, CHF, TIA, stroke, or CKD, compared with women without CAC, highlighting the importance of managing this disease process and preventing its progression. Significant differences in clinical parameters between CAC positive and negative women reported in this study have confirmed some of the previously known postulations about the etiology and pathogenesis of CAC. Studies to date have indicated that women with a history of PE not only have higher BMIs before pregnancy but also are at a greater risk for post-pregnancy weight gain and, consequently, higher BMIs compared to their normotensive counterparts. Thus, higher BMI, both before and after PE pregnancies, may represent one of the links between PE and future vascular calcification.^18,19 It has also been shown that an increase in visceral adipose tissue, which is strongly associated with obesity-related metabolic shifts, results in elevated release of pro-inflammatory cytokines contributing to the progression of CAC.^20,21 Higher levels of plasma CRP, a marker of inflammation, in CAC positive women, support the notion that inflammation could play an important and possibly crucial role in the pathophysiology of this process.²² Moreover, adiponectin and leptin, hormones secreted from adipose tissue, also known as adipocytokines, have each been linked to CAC, with opposing influences on its progression. Significantly lower A/L ratios in CAC positive patients are consistent with results from other studies demonstrating that lower adiponectin and A/L ratios are positively correlated with the progression of CAC and CAD.²³ Elevated levels of leptin are associated with more substantial calcification in coronary arteries, with studies showing that leptin could be a better predictor of severity of CAC than both adiponectin and/or the A/L ratio.^24,25 Leptin is also strongly associated with HOMA-IR, which was elevated in our CAC positive patients and is already a well-known risk factor for CAC progression.²⁶ A median HOMA-IR value of 1.92 in women with CAC in our study indicated early insulin resistance in women in this group.²⁷ A large Korean longitudinal study on subjects without CAC at baseline revealed that higher HOMA-IR values at baseline significantly increased the incidence and progression of CAC after 4 years of follow-up.²⁸ In addition, another factor significantly greater in CAC positive women, alkaline phosphatase (ALP), is known to increase mineralization by disintegrating pyrophosphate, a vascular calcification inhibitor.²⁹ Studies have shown that high ALP activity, especially above 66 IU/L (median value in CAC positive women in our study was 76 IU/L), is associated with the existence of CAC in individuals without any prior cardiovascular diseases.³⁰

EVs emerging from activated blood cells or other tissues and organs are involved in intracellular communication, transfer of bioactive molecules between cells including proteins, lipids and nucleic acids, and they play key roles in apoptosis and autophagy. They also have pro-coagulant and pro-inflammatory features and are linked to the pathophysiology of cardiometabolic and vascular diseases.^31,32 Major novel findings of this study were significantly elevated plasma EVs positive for six different markers in women with CAC, including tissue factor (TF), T-cell (CD3), smooth muscle cell (SM22α), adipocyte (Pref-1), adipocyte/macrophage (FABP4), and p16, marker of senescent cells. TF is found in EVs that are specifically released within atherosclerotic plaque by macrophages, vascular smooth muscle and endothelial cells, and are strongly involved in promoting vascular calcification.³³ TF with its pro-coagulant properties is involved in destabilization and rupture of plaque and thrombosis, leading to acute coronary syndromes.^34,35 T-cell derived EVs are one of the most important players in both atherosclerosis and CAC due to their involvement in lipid accumulation, increased inflammation and acting directly as platforms for microcalcifications.^36,37 Previous studies have found that vascular smooth muscle cells play a crucial role in the process of calcification, as they undergo osteogenic transformation. It has been described that higher levels of inorganic phosphate stimulate a change in vascular smooth muscle cell phenotype.^38.39 This is in line with elevated levels of alkaline phosphatase observed in CAC positive women, as well as with elevations in smooth muscle cell-derived EVs. Higher levels of inorganic phosphate are not only changing smooth muscle cell phenotypes, but also are stimulating release of more EVs into the vascular microenvironment, acting as a positive feedback mechanism.³⁸ EVs released from perivascular adipocytes act as regulators of vascular inflammation and thus are involved in calcification.⁴⁰ FABP-4 positive microvesicles, derived from adipocytes and macrophages, play significant roles in the development of insulin resistance and acceleration of atherosclerosis, and have been shown to be indicators of increased cardiovascular risk.^41,42 A noteworthy finding is the elevation of p16 positive EVs in CAC positive women. P16 is a marker of senescence, a process of aging characterized by cell cycle arrest. The role of senescence in the pathogenesis of chronic conditions, such as CAD, is well recognized and, indeed, may play a role in the significantly increased CVD risk in these women.^11,12,17 Treatment with senolytics, Dasatinib and Quercetin, drugs that selectively eliminate senescent cells, reduced markers of intimal calcification and osteogenesis in advanced atheromas of aged and atherosclerotic mice.⁴³ This highlights the critical role of aging and senescence in CAC pathophysiology, and the potential role of senolytic drugs in the management of this process, as well as in the reduction in risk of CVD and related mortality.

Women with CAC exhibited significantly higher levels of the plasma metabolites, proline and allothreonine (amino acid metabolism) and ribitol (carbohydrate metabolism), and lower levels of lactic acid (carbohydrate metabolism), compared to women without CAC, even after adjustment for BMI, HOMA-IR and CRP. Two of these metabolites have important roles in fat and fatty acid metabolism (allothreonine) and nucleotide metabolism (ribitol). The physiological roles of these metabolites and roles in CVD are presented in Table 3. Similar to our results, proline levels were elevated in the plasma of women with systemic lupus erythematosus and higher CAC scores.⁴⁴ On the other hand, this is the first study to link CAC to elevated plasma levels of allothreonine, which may contribute to CAC through several mechanisms, including immune function, metabolism of fatty acids and oxidative stress. In CAC negative women, one thought-provoking finding is that of higher plasma lactic acid levels, which have been shown to demonstrate both accelerating or protective effects against CAC formation and progression, and atherosclerosis. ^45–48 It has been generally accepted that elevated lactate impairs mitochondrial function, increases oxidative stress and autophagy, thus promoting vascular calcification.⁴⁹ Lactic acid as a metabolite has been considered a byproduct of metabolism rather than a bioactive molecule. In recent years, however, this opinion has changed with the discoveries that lactic acid presents as an active signal molecule that can regulate important immunological functions, and is involved in tissue regeneration and wound healing.^50,51 Studies have unveiled that lactic acid plays a very complex role in atherosclerosis and may have compensatory and beneficial effects in this process by reducing inflammation, promoting angiogenesis, and being a source of energy for the heart.^46–48 Extensive recent evidence has shown a close connection between lactic acid metabolism and polarization of macrophages. During inflammation, lactic acid promotes polarization of macrophages towards the anti-inflammatory M2 type, and it can prevent excessive inflammatory responses.⁵² This very complex and interesting topic requires further research. There is little to no evidence about connections between ribitol and cardiovascular diseases. A recent metabolic study reported increased levels of ribitol in the plasma of patients with progressive diabetic retinopathy.⁵³ Our study is the first to show a significant association between increased plasma ribitol levels and CAC in postmenopausal women. Collectively, our findings suggest the possibility that proline, allothreonine, ribitol and lactic acid could serve as early biomarkers of CAC, as well as provide potential therapeutic benefit by targeting their metabolism to attenuate CAC progression.

Table 3.

Plasma metabolites that were significantly different between CAC positive and negative women: physiological roles and roles in CVD

Metabolites	Physiological roles and roles in CVD
Proline	A non-essential amino acid engaged in multiple major metabolic pathways, such as protein synthesis and formation of collagen, maintenance of redox homeostasis and apoptosis. Recent findings highlight that biosynthesis and catabolism of proline are essential processes in different diseases due to their involvement in protein synthesis in pathological processes and response to cellular stress.59 Increased levels of plasma proline have been described in patients with obesity and insulin resistance, which is concordant with the results of this study.60
Allothreonine	A non-essential amino acid in humans and isomer of threonine, has antioxidant properties and plays key roles in fat and fatty acid metabolism, muscle growth and immune function by affecting different metabolic pathways.61 It is also a substrate for the enzyme, serine hydroxymethyltransferase-1, which catalyzes the interconversion between serine and glycine. The role of Allothreonine as a diagnostic indicator of other cardiovascular diseases such as atrial fibrillation, as well as candidacy as a CKD biomarker has been published before.62,63
Lactic acid	It has been generally accepted that elevated lactate impairs mitochondrial function, increases oxidative stress and autophagy, thus promoting vascular calcification. 49 Lactic acid as a metabolite has been considered a byproduct of metabolism rather than a bioactive molecule. In recent years, however, this opinion has changed with the discoveries that lactic acid presents as an active signal molecule that can regulate important immunological functions and is involved in tissue regeneration and wound healing.50,51 Studies have shown that lactic acid plays a very complex role in atherosclerosis and may also have compensatory and beneficial effects in this process by reducing inflammation, promoting angiogenesis, and being the source of energy for the heart.46–48
Ribitol	A sugar alcohol derived from ribose has a different role in carbohydrate and nucleotide metabolism.53 Studies have shown that ribitol has a role in metabolic reprogramming, could potentially have anti-cancer properties, and may be beneficial in treating some types of muscular dystrophy.64,65

Another notable finding of our study is decreased plasma uromodulin concentrations in women with a history of PE compared to women with normotensive pregnancies in CAC positive patients. Uromodulin is a protein produced exclusively by epithelial cells in the kidney tubules and a small amount is released into the bloodstream through the basolateral membrane. It protects the kidney by reducing the risk of urinary tract infections and formation of kidney stones, regulates blood pressure, renal ion transport, water balance and urine concentration, and exerts notable immunomodulatory, anti-inflammatory and antioxidant functions in both plasma and urine.⁵⁴ Emerging evidence has highlighted plasma uromodulin as a superior, robust, consistent, and practical biomarker of kidney diseases and CVD. Lower levels of plasma uromodulin are associated with CKD, hypertension, and increased cardiovascular risk. ^55,56 This corresponds with our other findings that CAC positive women with a history of PE have higher rates of current hypertension, hyperlipidemia, and higher insulin and HOMA-IR levels, thus suggesting that plasma uromodulin may be a marker of CAC, particularly in women with a history of PE. This finding, however, needs to be confirmed in a larger patient sample and sets the stage for future research.

In conclusion, our results clearly identify specific metabolites and populations of EVs that could contribute to the development and progression of CAC, and hence CVD and CKD in postmenopausal women. Among women with CAC, those with a history of PE are at even higher risk of developing one of the major cardiovascular events or CKD, compared to normotensive women. The altered metabolism of proteins, carbohydrates and indirectly, fatty acids and nucleic acid metabolism, as well as increased specific cellular origin of EVs and senescence, may contribute to the accumulation of calcium phosphate in coronary arteries and may identify postmenopausal women with CAC. Adequately powered future studies in larger cohorts may enable early identification of CAC positive postmenopausal women and potentially decrease the future risk of CVD morbidity and mortality.

Study limitations

We recognize the following limitations: i) the study population was small and from a single center; in addition, this study consisted of a secondary analysis using a subset of patients from an existing cohort, thus, this study should be interpreted as generating new hypotheses and considered exploratory. Due to the limited sample size, the precision in our estimates from subgroup analyses was low, as evidenced by wide confidence intervals, and we were also unable to perform robust multivariable analyses in these subgroups. ii) All patients were white, which may introduce a racial bias and limit data generalization. Studies have shown that white women have the highest prevalence of CAC, followed by Asian, black, and Hispanic women. White women also tend to have the highest CAC scores.⁵⁷ On the other hand, black women are at a much higher risk of preeclampsia, followed by Hispanic and white women.⁵⁸ Hence, major novel findings from this study require validation in larger, multi-center, prospective cohort studies. iii) EV analysis was performed during 2015-2016 using the most appropriate, accurate and frequently published method at that time.^14,15,67,68 The field of EV biology has undergone significant transformation and improvement since then, especially regarding the characterization of EVs. Thus, our methodological approach is not reflective of, and consistent with, the current MISEV (Minimal Information for Studies of Extracellular Vesicles) guidelines.⁶⁹

Supplementary Material

https://doi.org/10.6084/m9.figshare.31953429

New & Noteworthy.

Circulating extracellular vesicles, pro-inflammatory/pro-oxidant metabolites and history of preeclampsia may be utilized as biomarkers for diagnosis of early CAC in postmenopausal women

Abbreviations:

CVD

cardiovascular disease

CAC

coronary artery calcifications

EVs

extracellular vesicles

BMI

body mass index

CAD

coronary artery disease

CHF

congestive heart failure

TIA

transient ischemic attack

CKD

chronic kidney disease

A/L ratio

adiponectin/leptin ratio