Advertisement

Pulse wave velocity, augmentation index, and carotid intima-media thickness are each associated with different inflammatory protein signatures in young healthy adults: The lifestyle, biomarkers and atherosclerosis study

  • Paul Pettersson-Pablo
    Correspondence
    Corresponding author. Department of Laboratory Medicine, Clinical chemistry, Örebro University Hospital, Södra Grevrosengatan 1, 703 62 Örebro, Sweden.
    Affiliations
    Department of Laboratory Medicine, Faculty of Medicine and Health, Örebro University Hospital, Örebro, Sweden

    School of Medicine, Faculty of Medicine and Health, Örebro University, Örebro, Sweden

    Department of Medical Biosciences/Clinical Chemistry, Umeå University, Umeå, Sweden
    Search for articles by this author
  • Yang Cao
    Affiliations
    Clinical Epidemiology and Biostatistics, School of Medical Sciences, Örebro University, Örebro, Sweden
    Search for articles by this author
  • Lars H. Breimer
    Affiliations
    Department of Laboratory Medicine, Faculty of Medicine and Health, Örebro University Hospital, Örebro, Sweden

    School of Medicine, Faculty of Medicine and Health, Örebro University, Örebro, Sweden
    Search for articles by this author
  • Torbjörn K. Nilsson
    Affiliations
    Department of Medical Biosciences/Clinical Chemistry, Umeå University, Umeå, Sweden
    Search for articles by this author
  • Anita Hurtig-Wennlöf
    Affiliations
    School of Health, Faculty of Medicine and Health, Örebro University, Örebro, Sweden
    Search for articles by this author
Open AccessPublished:October 06, 2020DOI:https://doi.org/10.1016/j.atherosclerosis.2020.09.027

      Highlights

      • In healthy young adults, vascular physiology correlates with the expression of inflammatory proteins.
      • Pulse-wave velocity correlates with urokinase-type plasminogen activator (uPA).
      • Augmentation index correlates with interleukin-6 (IL6).
      • Carotid intima media thickness correlates with fractalkine.

      Abstract

      Background and aims

      We aimed to identify plasma protein biomarkers related to inflammation that correlated with physiological measurements of vascular function and structure in healthy individuals.

      Methods

      We used the OLINK proteomics panel, which measures 92 inflammatory proteins, in 834 young, healthy non-smokers (ages 18–26). Principal component analysis (PCA) was employed to identify patterns of proteins. The following measurements were used: pulse-wave velocity (PWV), carotid intima-media thickness (cIMT) and augmentation index (AIX). Established cardiovascular risk factors were included in multivariable models.

      Results

      PCA showed four principal components (PC 1, PC 2, PC 3, PC 4). PC 3, comprising proteins related to hemostasis, was significantly and inversely correlated with PWV. Among the proteins with the highest factor loadings on PC 3, uPA was negatively correlated with PWV in multivariable regression models. AIX was significantly correlated with PC 2, comprising inflammatory cytokines. Among the proteins with the highest factor loadings on PC 2, interleukin-6 was significantly correlated with AIX in the multivariable model. cIMT was significantly correlated with PC 4, comprising proteins related to chemotaxis. Among the proteins with the highest factor loadings on PC 4, fractalkine was significantly correlated with cIMT in the multivariable model.

      Conclusions

      In young, healthy individuals, OLINK inflammatory proteins correlated with measures of vascular status. Each of the three measures PWV, AIX, and cIMT, which target different parts of the vasculature, correlated with its own specific protein signature, indicating that different subsets of inflammatory mediators affect different parts of the vasculature and are detectable already in young healthy adults.

      Graphical abstract

      Keywords

      1. Introduction

      Vascular changes are associated with an increased cardiovascular risk and predict cardiovascular events independently of other established cardiometabolic risk factors, such as blood pressure and hyperlipidemia [
      • Rosenbaum D.
      • Giral P.
      • Chapman J.
      • Rached F.H.
      • Kahn J.F.
      • Bruckert E.
      • et al.
      Radial augmentation index is a surrogate marker of atherosclerotic burden in a primary prevention cohort.
      ,
      • Mahmood S.S.
      • Levy D.
      • Vasan R.S.
      • Wang T.J.
      The Framingham Heart Study and the epidemiology of cardiovascular disease: a historical perspective.
      ]. Increased stiffness is part of atherogenesis, i.e. an initially asymptomatic pathophysiological progression eventually manifesting as cardiovascular disease (CVD) [
      • Bentzon J.F.
      • Otsuka F.
      • Virmani R.
      • Falk E.
      Mechanisms of plaque formation and rupture.
      ]. The progression is typically slow; CVD most often affects the elderly, but the atherosclerotic process may start already in childhood [
      • McMahan C.A.
      • Gidding S.S.
      • Malcom G.T.
      • Tracy R.E.
      • Strong J.P.
      • McGill Jr., H.C.
      • et al.
      Pathobiological determinants of atherosclerosis in youth risk scores are associated with early and advanced atherosclerosis.
      ]. Several surrogate markers are used to evaluate future CVD risk, such as measurements of vascular elastic function and structure. Pulse-wave velocity (PWV) and augmentation index (AIX) are two examples that estimate the arterial stiffness. Carotid intima-media thickness (cIMT) assesses structural changes in the vasculature. Childhood increases in these measurements are associated with an increased future CVD risk [
      • Raitakari O.T.
      • Juonala M.
      • Kahonen M.
      • Taittonen L.
      • Laitinen T.
      • Maki-Torkko N.
      • et al.
      Cardiovascular risk factors in childhood and carotid artery intima-media thickness in adulthood: the Cardiovascular Risk in Young Finns Study.
      ].
      In the Lifestyle, Biomarkers and Atherosclerosis (LBA) study, we aim to identify plasma biomarkers related to the early stages of the atherosclerotic process by measuring physiological markers of vascular health, in young, healthy adults, a population that is seldom targeted for atherosclerosis research. Soluble proteins in serum or plasma, such as C-reactive protein (CRP), correlate with CVD risk and can be useful as predictors of atherosclerotic risk burden [
      • Koenig W.
      High-sensitivity C-reactive protein and atherosclerotic disease: from improved risk prediction to risk-guided therapy.
      ]. At the same time, an increased serum concentration of CRP has been suggested as not only an indicator of an increased CVD risk, but also a mediator in the atherosclerotic mechanism [
      • Jialal I.
      • Devaraj S.
      • Venugopal S.K.
      C-reactive protein: risk marker or mediator in atherothrombosis?.
      ]. We hypothesized that an extended proteomics analysis beyond CRP of inflammatory pathway mediators could reveal additional potentially useful biomarkers of early atherosclerosis, either for diagnosis or as novel targets for intervention or drug development.

      2. Materials and methods

      2.1 Study population

      The LBA dataset consists of 834 Swedish young adults between 18 and 26 years of age. The recruitment process and clinical data collection has been described previously [
      • Fernström M.
      • Fernberg U.
      • Eliason G.
      • Hurtig-Wennlöf A.
      Aerobic fitness is associated with low cardiovascular disease risk: the impact of lifestyle on early risk factors for atherosclerosis in young healthy Swedish individuals - the Lifestyle, Biomarker, and Atherosclerosis study.
      ]. Subjects reporting chronic disease such as diabetes mellitus or Crohn's disease were excluded, as were smokers. One individual reported having quit smoking 2 weeks prior to the data collection and was excluded. Written consent was obtained from each participant included in the study. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Regional Ethics Review Board, Uppsala, ref 2014/224.

      2.2 Vascular examinations

      The cIMT was measured using a high-resolution ultrasound B-mode system, (GE Healthcare, Vivid E9, Chicago, Illinois, US) with a 12 MHz linear array transducer, as previously described [
      • Fernström M.
      • Fernberg U.
      • Eliason G.
      • Hurtig-Wennlöf A.
      Aerobic fitness is associated with low cardiovascular disease risk: the impact of lifestyle on early risk factors for atherosclerosis in young healthy Swedish individuals - the Lifestyle, Biomarker, and Atherosclerosis study.
      ].
      Stiffness measures (PWV and AIX) were registered using applanation tonometry, using SphygmoCor (AtCor Medical Pty Ltd, SphygmoCor, Sydney, Australia) as previously described [
      • Mackenzie I.S.
      • Wilkinson I.B.
      • Cockcroft J.R.
      Assessment of arterial stiffness in clinical practice.
      ,
      • Fernberg U.
      • Fernström M.
      • Hurtig-Wennlöf A.
      Arterial stiffness is associated to cardiorespiratory fitness and body mass index in young Swedish adults: the Lifestyle, Biomarkers, and Atherosclerosis study.
      ,
      • Laurent S.
      • Cockcroft J.
      • Van Bortel L.
      • Boutouyrie P.
      • Giannattasio C.
      • Hayoz D.
      • et al.
      Expert consensus document on arterial stiffness: methodological issues and clinical applications.
      ]. Carotid and femoral pulse waves were recorded with simultaneously electrocardiogram recording, and the PWV (m/s) calculated as PWV = distance between measurement locations (m)/transit time (s) for the pulse wave. For AIX, the radial artery tonometry was performed at the subject's right wrist. The aortic pressure waveform was derived from the radial waveform by a validated transfer function. AIX is calculated from the aortic pressure waveform and adjusted to a heart rate of 75 beats per minute [
      • Pauca A.L.
      • O'Rourke M.F.
      • Kon N.D.
      Prospective evaluation of a method for estimating ascending aortic pressure from the radial artery pressure waveform.
      ].

      2.3 Established serum CVD risk markers

      Serum was left to clot for at least 30 min before centrifugation. The analytes were included for being well-established as biomarkers of increased CVD risk [
      • Verbeek R.
      • Hovingh G.K.
      • Boekholdt S.M.
      Non-high-density lipoprotein cholesterol: current status as cardiovascular marker.
      ,
      • Dongerkery S.P.
      • Schroeder P.R.
      • Shomali M.E.
      Insulin and its cardiovascular effects: what is the current evidence?.
      ,
      • Ryden L.
      • Viveca G.
      • Schnell O.
      • Jaakko T.
      • investigators E.I.
      Oral glucose tolerance testing and cardiovascular disease.
      ]. Because some studies have observed subsets of subjects displaying discordance between LDL and apolipoprotein B (Apo B) concentrations [
      • Mora S.
      • Martin S.S.
      • Virani S.S.
      Cholesterol insights and controversies from the UK biobank study.
      ], we measured both LDL and HDL and Apo B and apolipoprotein A-1 (Apo A-1) to examine whether the choice of lipid biomarkers would lead to any differences in the associations found in the multivariable models. Insulin and glucose were used in the calculation of the homeostasis model assessment of insulin resistance (HOMA-IR) [
      • Matthews D.R.
      • Hosker J.P.
      • Rudenski A.S.
      • Naylor B.A.
      • Treacher D.F.
      • Turner R.C.
      Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man.
      ]. CRP, orosomucoid, and Apo A-1 and Apo B were analyzed on a Siemens ADVIA 1800 Chemistry instrument. CRP had a CV of 5% at 0.74 mg/L with the Siemens High Sensitivity CRP Assay. Orosomucoid had a CV of 4% at 0.47 g/L using the DAKO orosomucoid immunoturbidimetry assay (Agilent, Santa Clara, CA). The Apo A-1 assay had a CV of 4% at 0.9 g/L and the Apo B assay a CV of 5% at 1,5 g/L. HDL and glucose were assayed with Vitros MicroSlide technology (5.1TM FS; Clinical Chemistry Instruments, Raritan, NJ, USA). LDL was assayed with Vitros MicroWell technology. HDL (6% CV at 1.0 mmol/L), TG (CV of 4% at 1.3 g/L), LDL (5% CV at 2.4 mmol/L) and glucose (4% CV at 4.6 mmol/L) were analyzed on a Vitros 5.1 system (Vitros 5.1TM FS, Clinical Chemistry Instruments, Raritan, NJ, USA). Insulin was analyzed with the Abbott Architect Insulin Assay with a CV of 7% at 8 mIU/L on an Architect i2000SR unit (Abbott, Abbot Park, IL, USA).

      2.4 Proteomic analysis by proximity extension assay (PEA)

      Plasma samples were collected after an overnight fast into ethylenediaminetetraacetic acid (EDTA) plasma vacutainer tubes (BD Vacutainer; BD AB, Stockholm, Sweden). The samples were analyzed using the PEA technology (OLINK® Proteomics, Uppsala, Sweden), which is based on the use of antibodies marked with oligonucleotides for simultaneous measurement of several analytes in a sample [
      • Lundberg M.
      • Eriksson A.
      • Tran B.
      • Assarsson E.
      • Fredriksson S.
      Homogeneous antibody-based proximity extension assays provide sensitive and specific detection of low-abundant proteins in human blood.
      ]. As a pair of antibodies bind to the analyte, their complementary oligonucleotides can be measured using real-time polymerase chain reaction, enabling specific identification of the analyte in question with little cross-reactivity. The results of the measurement are obtained as normalized protein expression (NPX). NPX is relative measurement where higher values correspond to higher protein concentrations in the sample, but it not to an absolute quantification. We employed the inflammation panel (article no 95302, www.olink.com), measuring the relative concentrations of 92 inflammatory proteins [
      • Lundberg M.
      • Eriksson A.
      • Tran B.
      • Assarsson E.
      • Fredriksson S.
      Homogeneous antibody-based proximity extension assays provide sensitive and specific detection of low-abundant proteins in human blood.
      ].

      2.5 Statistical analyses

      Plasma protein values below the limit of detection (LOD) of the analyte were replaced by the LOD divided by the square root of two. Proteins where a high percentage of the samples (>50%) had values below LOD were excluded from further analysis. Missing values were replaced by the sample mean. Principal component analysis (PCA) was used to identify components consisting of patterns of correlated variables. The number of components to retain was chosen by assessment of a scree plot. Varimax oblique rotation was used to minimize the number of variables with high factor loadings. The components obtained were examined for their correlation with the vascular variables using Pearson's correlation analyses. The results were visualized in a profile plot. Further multivariable linear regression analyses were performed to assess the association of PCs with vascular health measurements. The analyses were controlled for established risk factors of CVD as well as sex and age. In further multivariable analysis, established CVD risk markers were included to examine independent associations between the principal components (PC) and vascular health and any possible collinearity between the inflammatory proteins and established risk biomarkers. Finally, we entered the proteins with the highest factor loadings on the PCs into multivariable models to examine their relationship with the vascular measurements independently of the PCs. Statistical analyses were performed in SPSS ver 25 (IBM Corp., Armonk, NY, USA).

      3. Results

      3.1 Sample baseline characteristics

      The clinical characteristics of the LBA population are shown in Table 1. The population comprises more women than men (577 vs 257). Among some of the established markers of CVD risk, such as plasma lipids and systolic blood pressure, but not diastolic blood pressure, there is a statistically significant difference in mean concentrations between men and women in the population. Between men and non-estrogen using women, all included lipid biomarkers differed significantly. As for the vascular structure and function measures, PWV differed between men and non-estrogen using women and between non-estrogen using women and estrogen users. AIX differed between men and women, but no difference was seen based on contraceptive use.
      Table 1Baseline characteristics of the studied population sample.
      Men (n = 257)Non-estrogen using women (n = 428)Estrogen using women (n = 149)p value M vs NEUp value M vs EUp value NEU vs EU
      Age22 ± 2.022 ± 2.022 ± 1.60.690.270.58
      Body fat (%)15 ± 5.628 ± 6.827 ± 5.9<0.001<0.0010.42
      LDL (mmol/L)2.3 ± 0.692.2 ± 0.692.5 ± 0.790.36<0.001<0.001
      HDL (mmol/L)1.2 ± 0.281.4 ± 0.351.5 ± 0.41<0.001<0.0010.81
      CHOL (mmol/L)4.0 ± 0,794.3 ± 0.774.4 ± 0.770.0034<0.0010.0042
      Systolic BP (mmHg)122 ± 11109 ± 9.0112 ± 7.5<0.001<0.0010.0034
      Diastolic BP (mmHg)64 ± 6.764 ± 6.065 ± 6.90.0890.750.015
      TG (mmol/L)0.79 ± 0.350.75 ± 0.320.99 ± 0.390.25<0.001<0.001
      Serum insulin (mIE/L)7.5 ± 3.78.1 ± 4.78.2 ± 4.40.320.560.99
      Apo B (g/L)0.77 ± 0.180.78 ± 0.180.83 ± 0.200.97<0.001<0.001
      Apo A-1 (g/L)1.4 ± 0.211.5 ± 0.271.7 ± 0.325<0.001<0.001<0.001
      Apo B/Apo A-1 ratio0.56 ± 0.140.51 ± 0.140.50 ± 0.15<0.0010.0320.75
      CRP (mg/L)1.3 ± 2.71.4 ± 2.84.1 ± 7.10.99<0.001<0.001
      Orosomucoid (g/L)0.72 ± 0.160.69 ± 0.170.60 ± 0.170.30<0.001<0.001
      PWV (m/s)5.6 ± 0.865.2 ± 0.695.3 ± 0.9<0.0010.0920.0045
      AIX (%)−8.4 ± 9.5−5.0 ± 10−4.7 ± 9.1<0.0010.00190.99
      cIMT (mm)0.60 ± 0.0730.49 ± 0.0570.50 ± 0.0550.480.180.61
      Values are presented as mean ± SD (standard deviation).
      p value: comparison between groups by ANOVA with Tukey post hoc comparison.
      M = men. NEU = non-estrogen using women. EU = estrogen using women. LDL = low-density lipoprotein. HDL = high-density lipoprotein. TG = triglycerides. CHOL = total cholesterol. BP = blood pressure. Apo B = apolipoprotein B. Apo A-1 = apolipoprotein A-1. CRP = C-reactive protein. BP = blood pressure. PWV = pulse-wave velocity. AIX = augmentation index. cIMT = carotid intima-media thickness.

      3.2 Protein expression and PCA

      The proteins monocyte chemotactic protein 3, interleukin 1-alpha, interleukin-2, thymic stromal lymphopoietin were excluded from further analysis due to a high number of results below LOD of the PEA analysis. A scree plot showed inflexions justifying extraction of four PCs (Supplementary Fig. 1), altogether explaining 36% of the total variance of the inflammatory biomarkers (Table 2). The factor loadings of each PC after rotation are shown in Supplementary Table 1. The three inflammatory proteins with the highest factor loadings for each PC were for PC 1: axin-1, sirtuin-like protein 2 (SIRT2) and signal transducing adaptor molecule-binding protein (STAMBP), for PC 2: C-X-C motif chemokine 9 (CXCL9), interleukin-6 (IL6) and interleukin-10 (IL10), for PC 3: tumor necrosis factor superfamily, member 12 (TWEAK), leukemia inhibitory factor receptor (LIF-R) and urokinase-type plasminogen activator (uPA), and for PC 4: macrophage colony-stimulating factor 1 (CSF-1), fractalkine (CX3CL1) and Interleukin-10 receptor subunit beta (IL10RB).
      Table 2Squared loadings and total variance explained of the four components.
      ComponentTotal% of varianceCumulative %
      114.09015.48415.484
      26.6897.35022.834
      36.6387.29430.128
      45.0815.58335.711

      3.3 PCs and PWV, AIX and cIMT

      The correlations between the PCs and each of the vascular variables are visualized in a profile plot (Fig. 1). In multivariable linear regression, the four retained PCs were analyzed for their association with each of the vascular measurements in models including established risk factors of CVD (Table 3). Regardless of the vascular measurement used and despite the narrow range of ages in the population, age was a significant adverse predictor in all analyses. Replacing the apolipoproteins with LDL and HDL did not significantly impact any of the correlations found, having only a minor impact on the β-coefficients and p values in Table 3 (not shown). Similarly, entering insulin and/or glucose instead of HOMA-IR had no major impact on the results (not shown). While the mean PWV, AIX and cIMT in some cases differed significantly between groups of men, non-estrogen using women and estrogen using women in the population (Table 1), sex was not significantly associated with any of these measurements in the multivariable models (Table 3).
      Fig. 1
      Fig. 1Component pattern profile showing the correlation between the principal components and PWV, AIX and cIMT.
      PWV = pulse-wave velocity. AIX = augmentation index. cIMT = carotid intima-media thickness.
      Table 3Multivariable linear regression: vascular measurements as a function of principal components and established risk factors of cardiovascular disease.
      PWV (R2: 0.057)AIX (R2: 0.038)cIMT (R2: 0.019)
      β (95% CI)pβ (95% CI)pβ (95% CI)p
      PC10.031 (−0.037; 0.099)0.380.056 (−0.015; 0.13)0.12−0.028 (−0.10; 0.043)0.44
      PC20.036 (−0.033; 0.11)0.300.10 (0.032; 0.18)0.0046*0.0041 (−0.077; 0.069)0.91
      PC3−0.082 (−0.15;-0.0028)0.045*0.011 (−0.064; 0.086)0.780.034 (−0.042; 0.0.11)0.38
      PC40.026 (−0.045; 0.097)0.45−0.050 (−0.12; 0.024)0.180.083 (0.0087; 0.16)0.029*
      Age0.080 (0.047; 0.11)<0.001*0.065 (0.031; 0.10)<0.001*0.045 (0.011; 0.080)0.011*
      Sex0.052 (−0.10; 0.21)0.51−0.058 (−0.22; 0.10)0.480.015 (−0.15; 0.18)0.86
      Body fat%0.052 (−0.018; 0.12)0.150.051 (−0.020; 0.12)0.160.058 (−0.014; 0.13)0.11
      ApoB/ApoA-10.023 (−0.044; 0.090)0.50−0.0051 (−0.74; 0.064)0.890.018 (−0.052; 0.088)0.61
      HOMA-IR0.042 (−0.11; 0.022)0.200.040 (−0.11; 0.027)0.24−0.011 (−0.079; 0.056)0.74
      Variables are separated on sex and z-score transformed before inclusion in the multivariable model.
      PWV = pulse-wave velocity. AIX = augmentation index. cIMT = carotid intima-media thickness. β = beta coefficient. PC = principal component. ApoB/ApoA-1 = apolipoprotein B/apolipoprotein A-1 ratio. HOMA-IR = homeostatic model assessment insulin resistance.
      * Indicates a p value < 0.05.
      PC 3 was significantly negatively associated with PWV. PC 2 was significantly associated with AIX, and PC 4 was significantly associated with cIMT. In models with PCs included as variables together with the established risk factors, none of the latter was significantly associated with the vascular measurements (Table 3). The proteins with the highest factor loadings on the PCs were entered into further multivariable regression models to examine their relationship with the vascular variable independently of PC (Table 4). In these models, uPA was significantly and negatively correlated with PWV, IL-6 was significantly correlated with AIX, and fractalkine (CX3CL1) was significantly correlated with cIMT.
      Table 4Multivariable linear regression. Vascular variables PWV, AIX and cIMT as functions of established risk factors of CVD and the principal component proteins with the highest factor loadings.
      R2β (95% CI)p value
      PWV
       Model 1 + uPA0.060−0.19 (−0.34;-0.029)0.020
       Model 1+TWEAK0.058−0.17 (−0.37; 0.022)0.083
       Model 1+LIF-R0.056−0.070 (−0.37; 0.022)0.61
      AIX
       Model 1+IL-60.0480.18 (0.078; 0.28)<0.001
       Model 1+CXCL90.0320.031 (−0.069; 0.13)0.55
       Model 1+IL-100.0310.026 (−0.099; 0.15)0.68
      cIMT
       Model 1 + CX3CL10.0140.30 (0.038; 0.56)0.025
       Model 1+CSF-10.00830.16 (−0.15; 0.45)0.32
       Model 1 + IL10RB0.00690.055 (0.026; 0.15)0.60
      Model 1 includes age, sex, body fat%, ApoB/ApoA-1 ratio and HOMA-IR.
      PWV = pulse-wave velocity. AIX = augmentation index. cIMT = carotid intima-media thickness. TWEAK = tumor necrosis factor superfamily, member 12. LIF-R = leukemia inhibitory factor receptor. uPA = urokinase-type plasminogen activator. CXCL9 = C-X-C motif chemokine 9. Il-6 = interleukin-6. Il-10 = interleukin-10. CSF-1 = macrophage colony-stimulating factor 1. CX3CL1 = C-X3-C motif chemokine ligand 1 AKA fractalkine. IL10RB = interleukin-10 receptor subunit beta.
      In subgroup analysis of the women, estrogen contraceptive use was associated with a significantly lower value of PC 3 (p < 0.001) and a significantly higher value of PC 4 (p < 0.001).

      4. Discussion

      In this exploratory study, we found that certain patterns of inflammatory proteins correlate with physiological measurements of vascular structure and function in young and otherwise healthy adults. We used PCA to achieve manageable data reduction and take into account the multiple collinearity that is expected when employing a proteomics panel that simultaneously measures several proteins that are all related to certain pre-defined pathways of the inflammatory responses [
      • Ibrahim G.M.
      • Morgan B.R.
      • Macdonald R.L.
      Patient phenotypes associated with outcomes after aneurysmal subarachnoid hemorrhage: a principal component analysis.
      ]. Among the four PCs obtained, PC 3 was inversely associated with PWV, PC 2 was associated with AIX and PC 4 was associated with cIMT. These findings concord with the idea of atherosclerosis having an inflammatory component that may be present already in adulthood [
      • Hansson G.K.
      Inflammation, atherosclerosis, and coronary artery disease.
      ], and adds new data on the specific protein signatures involved in the changes of the respective vascular variables.

      4.1 PWV and PC 3

      The inverse relationship between PC 3 and PWV suggests that PC 3 proteins have possible atheroprotective properties. The proteins with the highest factor loadings TWEAK, uPA and LIF-R have been implicated in regulatory pathways relating to hemostasis [
      • Meisel S.R.
      • Shimon I.
      • Edgington T.S.
      • Melmed S.
      • Cercek B.
      • Shah P.K.
      Leukaemia inhibitory factor enhances tissue factor expression in human monocyte-derived macrophages: a gp130-mediated mechanism.
      ,
      • Simone T.M.
      • Higgins S.P.
      • Archambeault J.
      • Higgins C.E.
      • Ginnan R.G.
      • Singer H.
      • et al.
      A small molecule PAI-1 functional inhibitor attenuates neointimal hyperplasia and vascular smooth muscle cell survival by promoting PAI-1 cleavage.
      ]. They might therefore exert a beneficial effect on the endothelium by means of an anti-thrombotic effect. The PCA takes into account multiple variables simultaneously by identifying groups of proteins that covary. This reveals patterns of proteins that constitute parts of a signaling pathway, even when the individual proteins do not vary in a manner distinct enough to show a strong correlation with a certain outcome [
      • Alonso-Gutierrez J.
      • Kim E.M.
      • Batth T.S.
      • Cho N.
      • Hu Q.
      • Chan L.J.G.
      • et al.
      Principal component analysis of proteomics (PCAP) as a tool to direct metabolic engineering.
      ]. In this study, upon examining the PC 3 proteins independently in multivariable regression models, only uPA showed an independent, significant correlation with PWV. The results may be indicative of the intricate but subtle interplay in the relationship between vascular phenotype and plasma proteins in a population composed of healthy individuals whose vascular changes are minor and correlate only weakly with potential risk factors among young, healthy subjects such as those in our study. An increase in plasma uPA concentration was associated with the presence of CVD in hemodialysis patients [
      • Pawlak K.
      • Mysliwiec M.
      • Pawlak D.
      Haemostatic system, biochemical profiles, kynurenines and the prevalence of cardiovascular disease in peritoneally dialyzed patients.
      ]. The authors hypothesized that an activated uPA system may be instrumental in atherogenesis and responsible for the increase in CVD risk seen in hemodialysis patients. This seems to contrast with our findings, in which an increase uPA was associated with a lower arterial stiffness. The difference between the populations may explain the discrepancy in the results, with our study examining young and healthy individuals and the aforementioned one hemodialysis patients with a mean age of 62 years. Increased as well as decreased fibrinolytic activity has been claimed to confer increased CVD risk [
      • Morange P.E.
      • Bickel C.
      • Nicaud V.
      • Schnabel R.
      • Rupprecht H.J.
      • Peetz D.
      • et al.
      Haemostatic factors and the risk of cardiovascular death in patients with coronary artery disease: the AtheroGene study.
      ,
      • Sakkinen P.A.
      • Cushman M.
      • Psaty B.M.
      • Rodriguez B.
      • Boineau R.
      • Kuller L.H.
      • et al.
      Relationship of plasmin generation to cardiovascular disease risk factors in elderly men and women.
      ]. The main inhibitor of uPA is plasminogen activator inhibitor-1 (PAI-1) [
      • Lijnen H.R.
      • Collen D.
      Mechanisms of physiological fibrinolysis.
      ]. Upon exposure of vascular smooth muscle cell to the cleaved form of PAI-1, TWEAK signaling is augmented which provides an anti-stenotic effect in injured arteries [
      • Simone T.M.
      • Higgins S.P.
      • Archambeault J.
      • Higgins C.E.
      • Ginnan R.G.
      • Singer H.
      • et al.
      A small molecule PAI-1 functional inhibitor attenuates neointimal hyperplasia and vascular smooth muscle cell survival by promoting PAI-1 cleavage.
      ]. This interplay of regulators of fibrinolysis could provide an explanation for the collinearity of TWEAK and uPA in our population, with both proteins having the highest factor loadings of PC 3. While TWEAK was non-significant in multivariable analysis in our population (Table 4), the association seen between PC 3 and PWV in our population is in line with a previous study in which lower plasma/serum TWEAK was found to be associated with an increase in atherosclerotic burden and atheromatosis progression in elderly patients with kidney disease free of clinical CVD [
      • Fernandez-Laso V.
      • Mendez-Barbero N.
      • Valdivielso J.M.
      • Betriu A.
      • Fernandez E.
      • Egido J.
      • et al.
      Soluble TWEAK and atheromatosis progression in patients with chronic kidney disease.
      ]. The non-significance of the relationship between TWEAK and PWV (Table 4) could also be explained by the existence of an independent link between uPA activity and a lower vascular stiffness that does not include TWEAK. Further longitudinal studies on uPA and a potentially beneficial effect on PWV in young, healthy subjects are warranted.

      4.2 AIX and PC 2

      PC 2 was significantly and independently correlated with arterial stiffness measurement AIX. The proteins with the highest factor loadings of PC 2 were IL-6, IL-10 and CXCL9. In multivariable models examining the relationship between these proteins and AIX separately, only Il-6 was independently and significantly associated with PWV (Table 4). IL-6 and IL-10 are inflammatory cytokines that are involved in various pro- and anti-inflammatory signaling pathways that are known to increase with aging and are implicated in CVD [
      • Hartman J.
      • Frishman W.H.
      Inflammation and atherosclerosis: a review of the role of interleukin-6 in the development of atherosclerosis and the potential for targeted drug therapy.
      ,
      • Torres K.C.L.
      • Rezende V.B.
      • Lima-Silva M.L.
      • Santos L.J.S.
      • Costa C.G.
      • Mambrini J.V.M.
      • et al.
      Immune senescence and biomarkers profile of Bambui aged population-based cohort.
      ]. In this study, aging was a consistent predictor of adverse vascular structure or function in the multivariable models in Table 3, Table 4. These results are indicative of the importance of aging on endothelial function that is observable even in the limited span of 18–26 years of age. In the multivariable models, the independent correlations between age and the vascular measurements suggest that the adversity of time exerts its effect on the vasculature through mechanisms that are in part independent of the established risk factors included in the model, such as body composition, metabolic status and plasma lipids. To our knowledge, ours is the first study to find an independent relationship between IL-6 and AIX in healthy subjects. Previous studies on patients with CVD as well as various inflammatory diseases have resulted in both significant associations and support for the null hypothesis in the relationship between AIX and IL-6 and other inflammatory cytokines [
      • Barbulescu A.L.
      • Vreju F.
      • Cojocaru-Gofita I.R.
      • Musetescu A.E.
      • Ciurea P.L.
      Impaired arterial stiffness in systemic lupus ertythematosus - correlations with inflammation markers.
      ,
      • Tuttolomondo A.
      • Pecoraro R.
      • Di Raimondo D.
      • Di Sciacca R.
      • Canino B.
      • Arnao V.
      • et al.
      Immune-inflammatory markers and arterial stiffness indexes in subjects with acute ischemic stroke with and without metabolic syndrome.
      ,
      • Kim S.K.
      • Kim K.S.
      • Lee Y.S.
      • Park S.H.
      • Choe J.Y.
      Arterial stiffness and proinflammatory cytokines in fibromyalgia syndrome.
      ]. Ultimately, our findings introduce inflammatory cytokines as agents of early vascular dysfunction even in young, healthy subjects.

      4.3 cIMT and PC 4

      cIMT was independently, positively associated with PC 4 in our population. CSF-1 (aka M-CSF), CX3CL1 and IL10RB were the proteins with the highest factor loadings on this PC. Upon examination of these proteins independently, CX3CL1 aka fractalkine was significantly associated with cIMT. The R2 of the cIMT models were low (Table 3, Table 4) illustrating the fact that cIMT measurements were low and showed little variance (Table 1) in this healthy population that had not yet developed any major vascular remodeling. Nevertheless, despite the low R2, the fairly strong and independent correlation observed between cIMT and fractalkine suggests that fractalkine increased in conjunction with adverse vascular remodeling in early atherosclerosis. Further studies are warranted to examine whether the relationship with cIMT is causal. Genetic polymorphisms in the fractalkine receptor CX3CR1 have been associated with both an increased and decreased CVD risk [
      • Gomez-Diaz R.A.
      • Gutierrez J.
      • Contreras-Rodriguez A.
      • Valladares-Salgado A.
      • Tanus-Hajj J.
      • Mondragon-Gonzalez R.
      • et al.
      Association of V249I and T280M variants of fractalkine receptor CX3CR1 with carotid intima-media thickness in a mexican population with type 2 diabetes.
      ]. While still relatively little studied, the CX3CL1/CXCR1 pathway's proliferative and anti-apoptotic effect on the endothelium, monocyte related functions and its involvement in atherosclerotic pathophysiology are being increasingly recognized [
      • Apostolakis S.
      • Spandidos D.
      Chemokines and atherosclerosis: focus on the CX3CL1/CX3CR1 pathway.
      ]. It plays an important role in proliferation and differentiation of monocytes and macrophages [
      • Stanley E.R.
      • Berg K.L.
      • Einstein D.B.
      • Lee P.S.
      • Pixley F.J.
      • Wang Y.
      • et al.
      Biology and action of colony--stimulating factor-1.
      ], cells that are procoagulant. The role played by macrophages, as well as the CSFs, in atherogenesis is well-established [
      • Lawrence T.
      • Natoli G.
      Transcriptional regulation of macrophage polarization: enabling diversity with identity.
      ]. In this respect, the non-significant relationship between cIMT and CSF-1, unlike fractalkine, was somewhat surprising given how they both had high factor loadings on PC 4. The subtlety of cIMT changes in this healthy population could contribute, as suggested by the low R2 of the model, but both age and fractalkine showed robust relationships with cIMT. The alternative explanation is that the relationship between fractalkine and cIMT is not primarily mediated via macrophage recruitment. The relation between atherosclerosis and CSFs is complex and seems to differ according to the severity of disease. CSF-1 plays a role in driving differentiation towards increasing M2 phenotype macrophages, counteracting an adverse M1 promoting effect of granulocyte-macrophage colony-stimulating factor, GM-CSF [
      • Trus E.
      • Basta S.
      • Gee K.
      Who's in charge here? Macrophage colony stimulating factor and granulocyte macrophage colony stimulating factor: competing factors in macrophage polarization.
      ]. A CSF-1 deficiency led to a 7-fold decreased presence of atherosclerotic lesions in knockout mice [
      • Smith J.D.
      • Trogan E.
      • Ginsberg M.
      • Grigaux C.
      • Tian J.
      • Miyata M.
      Decreased atherosclerosis in mice deficient in both macrophage colony-stimulating factor (op) and apolipoprotein E.
      ]. In conclusion, the positive correlation between fractalkine and cIMT in our population implies that fractalkine increases in conjunction with the early development of an unfavorable vascular structure, as assessed by cIMT in young, healthy individuals.

      4.4 Findings in women

      The subgroup analysis of the women yielded results showing that estrogen contraceptive using women had a lower expression of PC 3 and a higher expression of PC 4. Studies on the effect of contraceptive use on arterial stiffness in pre-menopausal women have been conflicting, with two studies having found an increased stiffness among contraceptive users [
      • Yu A.
      • Giannone T.
      • Scheffler P.
      • Doonan R.J.
      • Egiziano G.
      • Gomez Y.H.
      • et al.
      The effect of oral contraceptive pills and the natural menstrual cYCLe on arterial stiffness and hemodynamICs (CYCLIC).
      ,
      • Hickson S.S.
      • Miles K.L.
      • McDonnell B.J.
      • Yasmin
      • Cockcroft J.R.
      • Wilkinson I.B.
      • et al.
      Use of the oral contraceptive pill is associated with increased large artery stiffness in young women: the ENIGMA study.
      ] and one study having found no difference [
      • Priest S.E.
      • Shenouda N.
      • MacDonald M.J.
      Effect of sex, menstrual cycle phase, and monophasic oral contraceptive pill use on local and central arterial stiffness in young adults.
      ]. In our population, estrogen contraceptive users had a significantly increased PWV, while no difference was seen in AIX or cIMT (Table 1). Hypothetically, the increased risk of venous thrombosis observed in contraceptive users [
      • Jick S.S.
      • Jick H.
      Cerebral venous sinus thrombosis in users of four hormonal contraceptives: levonorgestrel-containing oral contraceptives, norgestimate-containing oral contraceptives, desogestrel-containing oral contraceptives and the contraceptive patch.
      ] may be related to the regulative actions of the PC 3 and PC 4 proteins.

      4.5 Limitations and strengths

      The main limitation of this exploratory study is its cross-sectional design proscribing inferences of causality in the relationship between plasma protein expression and the physiological measurements. Among its strengths are its large cohort that underwent a rigorous vetting process in the recruitment of the subjects, ensuring that what we examined was the vascular status of young, non-smoking, healthy individuals in various preclinical stages of atherosclerosis and the employment of multivariable analyses with established risk factors to ensure the independence of the novel associations found. The non-significant relationship between the established risk factors and our vascular biomarkers could indicate that these play a subordinate role compared to specific groups of inflammatory protein signatures, each associated with a particular vascular property. Further studies of longitudinal design are warranted to examine the possibly causal relationships between the inflammatory proteins implicated here and CVD risk, especially premature clinical CVD events in young and early middle-aged adults. Any proteins implicated in causal mechanisms that are detrimental to endothelial function warrant studies into their clinical, nutritional and pharmacological application in preventing early atherosclerosis.

      4.6 Conclusion

      In a population of healthy, young adults, three different measurements of arterial function and structure were each independently associated with a different set of inflammatory proteins. Our tentative interpretation is that the main pathophysiological mediators involved in the early atherosclerotic progression differ specifically for structural and functional deterioration as assessed by different measures, which target different parts of the vasculature. This study adds novel insights into the inflammatory processes in the early stages of vascular dysfunction in healthy, young, non-smoking adults.

      Financial support

      This work was supported by AFA Insurance, Sweden [ 130275 ]; Region Örebro County’s Research Committee, Örebro, Sweden [ OLL-780061 ]; Nyckelfonden, Örebro, Sweden [ OLL-787681 ], and Umeå University, Umeå, Sweden [ RV-865861 ]. None of the funding sources had an influence on the design or production of the article.

      CRediT authorship contribution statement

      Paul Pettersson-Pablo: Conceptualization, Methodology, Formal analysis, Writing - original draft, Writing - review & editing, Visualization, Funding acquisition. Yang Cao: Methodology, Formal analysis, Writing - review & editing. Lars H. Breimer: Writing - review & editing, Supervision, Funding acquisition. Torbjörn K. Nilsson: Conceptualization, Methodology, Formal analysis, Writing - review & editing, Supervision, Funding acquisition. Anita Hurtig-Wennlöf: Conceptualization, Methodology, Formal analysis, Investigation, Writing - review & editing, Supervision, Project administration, Funding acquisition.

      Declaration of competing interest

      The authors declared they do not have anything to disclose regarding conflict of interest with respect to this manuscript.

      Appendix A. Supplementary data

      The following are the Supplementary data to this article:

      References

        • Rosenbaum D.
        • Giral P.
        • Chapman J.
        • Rached F.H.
        • Kahn J.F.
        • Bruckert E.
        • et al.
        Radial augmentation index is a surrogate marker of atherosclerotic burden in a primary prevention cohort.
        Atherosclerosis. 2013; 231: 436-441
        • Mahmood S.S.
        • Levy D.
        • Vasan R.S.
        • Wang T.J.
        The Framingham Heart Study and the epidemiology of cardiovascular disease: a historical perspective.
        Lancet. 2014; 383: 999-1008
        • Bentzon J.F.
        • Otsuka F.
        • Virmani R.
        • Falk E.
        Mechanisms of plaque formation and rupture.
        Circ. Res. 2014; 114: 1852-1866
        • McMahan C.A.
        • Gidding S.S.
        • Malcom G.T.
        • Tracy R.E.
        • Strong J.P.
        • McGill Jr., H.C.
        • et al.
        Pathobiological determinants of atherosclerosis in youth risk scores are associated with early and advanced atherosclerosis.
        Pediatrics. 2006; 118: 1447-1455
        • Raitakari O.T.
        • Juonala M.
        • Kahonen M.
        • Taittonen L.
        • Laitinen T.
        • Maki-Torkko N.
        • et al.
        Cardiovascular risk factors in childhood and carotid artery intima-media thickness in adulthood: the Cardiovascular Risk in Young Finns Study.
        J. Am. Med. Assoc. 2003; 290: 2277-2283
        • Koenig W.
        High-sensitivity C-reactive protein and atherosclerotic disease: from improved risk prediction to risk-guided therapy.
        Int. J. Cardiol. 2013; 168: 5126-5134
        • Jialal I.
        • Devaraj S.
        • Venugopal S.K.
        C-reactive protein: risk marker or mediator in atherothrombosis?.
        Hypertension. 2004; 44: 6-11
        • Fernström M.
        • Fernberg U.
        • Eliason G.
        • Hurtig-Wennlöf A.
        Aerobic fitness is associated with low cardiovascular disease risk: the impact of lifestyle on early risk factors for atherosclerosis in young healthy Swedish individuals - the Lifestyle, Biomarker, and Atherosclerosis study.
        Vasc. Health Risk Manag. 2017; 13: 91-99
        • Mackenzie I.S.
        • Wilkinson I.B.
        • Cockcroft J.R.
        Assessment of arterial stiffness in clinical practice.
        QJM. 2002; 95: 67-74
        • Fernberg U.
        • Fernström M.
        • Hurtig-Wennlöf A.
        Arterial stiffness is associated to cardiorespiratory fitness and body mass index in young Swedish adults: the Lifestyle, Biomarkers, and Atherosclerosis study.
        Eur J Prev Cardiol. 2017; 24: 1809-1818
        • Laurent S.
        • Cockcroft J.
        • Van Bortel L.
        • Boutouyrie P.
        • Giannattasio C.
        • Hayoz D.
        • et al.
        Expert consensus document on arterial stiffness: methodological issues and clinical applications.
        Eur. Heart J. 2006; 27: 2588-2605
        • Pauca A.L.
        • O'Rourke M.F.
        • Kon N.D.
        Prospective evaluation of a method for estimating ascending aortic pressure from the radial artery pressure waveform.
        Hypertension. 2001; 38: 932-937
        • Verbeek R.
        • Hovingh G.K.
        • Boekholdt S.M.
        Non-high-density lipoprotein cholesterol: current status as cardiovascular marker.
        Curr. Opin. Lipidol. 2015; 26: 502-510
        • Dongerkery S.P.
        • Schroeder P.R.
        • Shomali M.E.
        Insulin and its cardiovascular effects: what is the current evidence?.
        Curr. Diabetes Rep. 2017; 17: 120
        • Ryden L.
        • Viveca G.
        • Schnell O.
        • Jaakko T.
        • investigators E.I.
        Oral glucose tolerance testing and cardiovascular disease.
        Lancet Diabetes Endocrinol. 2016; 4: 732-733
        • Mora S.
        • Martin S.S.
        • Virani S.S.
        Cholesterol insights and controversies from the UK biobank study.
        Circulation. 2019; 140: 553-555
        • Matthews D.R.
        • Hosker J.P.
        • Rudenski A.S.
        • Naylor B.A.
        • Treacher D.F.
        • Turner R.C.
        Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man.
        Diabetologia. 1985; 28: 412-419
        • Lundberg M.
        • Eriksson A.
        • Tran B.
        • Assarsson E.
        • Fredriksson S.
        Homogeneous antibody-based proximity extension assays provide sensitive and specific detection of low-abundant proteins in human blood.
        Nucleic Acids Res. 2011; 39e102
        • Ibrahim G.M.
        • Morgan B.R.
        • Macdonald R.L.
        Patient phenotypes associated with outcomes after aneurysmal subarachnoid hemorrhage: a principal component analysis.
        Stroke. 2014; 45: 670-676
        • Hansson G.K.
        Inflammation, atherosclerosis, and coronary artery disease.
        N. Engl. J. Med. 2005; 352: 1685-1695
        • Meisel S.R.
        • Shimon I.
        • Edgington T.S.
        • Melmed S.
        • Cercek B.
        • Shah P.K.
        Leukaemia inhibitory factor enhances tissue factor expression in human monocyte-derived macrophages: a gp130-mediated mechanism.
        Br. J. Haematol. 1999; 107: 747-755
        • Simone T.M.
        • Higgins S.P.
        • Archambeault J.
        • Higgins C.E.
        • Ginnan R.G.
        • Singer H.
        • et al.
        A small molecule PAI-1 functional inhibitor attenuates neointimal hyperplasia and vascular smooth muscle cell survival by promoting PAI-1 cleavage.
        Cell. Signal. 2015; 27: 923-933
        • Alonso-Gutierrez J.
        • Kim E.M.
        • Batth T.S.
        • Cho N.
        • Hu Q.
        • Chan L.J.G.
        • et al.
        Principal component analysis of proteomics (PCAP) as a tool to direct metabolic engineering.
        Metab. Eng. 2015; 28: 123-133
        • Pawlak K.
        • Mysliwiec M.
        • Pawlak D.
        Haemostatic system, biochemical profiles, kynurenines and the prevalence of cardiovascular disease in peritoneally dialyzed patients.
        Thromb. Res. 2010; 125: e40-e45
        • Morange P.E.
        • Bickel C.
        • Nicaud V.
        • Schnabel R.
        • Rupprecht H.J.
        • Peetz D.
        • et al.
        Haemostatic factors and the risk of cardiovascular death in patients with coronary artery disease: the AtheroGene study.
        Arterioscler. Thromb. Vasc. Biol. 2006; 26: 2793-2799
        • Sakkinen P.A.
        • Cushman M.
        • Psaty B.M.
        • Rodriguez B.
        • Boineau R.
        • Kuller L.H.
        • et al.
        Relationship of plasmin generation to cardiovascular disease risk factors in elderly men and women.
        Arterioscler. Thromb. Vasc. Biol. 1999; 19: 499-504
        • Lijnen H.R.
        • Collen D.
        Mechanisms of physiological fibrinolysis.
        Baillieres Clin Haematol. 1995; 8: 277-290
        • Fernandez-Laso V.
        • Mendez-Barbero N.
        • Valdivielso J.M.
        • Betriu A.
        • Fernandez E.
        • Egido J.
        • et al.
        Soluble TWEAK and atheromatosis progression in patients with chronic kidney disease.
        Atherosclerosis. 2017; 260: 130-137
        • Hartman J.
        • Frishman W.H.
        Inflammation and atherosclerosis: a review of the role of interleukin-6 in the development of atherosclerosis and the potential for targeted drug therapy.
        Cardiol. Rev. 2014; 22: 147-151
        • Torres K.C.L.
        • Rezende V.B.
        • Lima-Silva M.L.
        • Santos L.J.S.
        • Costa C.G.
        • Mambrini J.V.M.
        • et al.
        Immune senescence and biomarkers profile of Bambui aged population-based cohort.
        Exp. Gerontol. 2018; 103: 47-56
        • Barbulescu A.L.
        • Vreju F.
        • Cojocaru-Gofita I.R.
        • Musetescu A.E.
        • Ciurea P.L.
        Impaired arterial stiffness in systemic lupus ertythematosus - correlations with inflammation markers.
        Curr Health Sci J. 2012; 38: 61-65
        • Tuttolomondo A.
        • Pecoraro R.
        • Di Raimondo D.
        • Di Sciacca R.
        • Canino B.
        • Arnao V.
        • et al.
        Immune-inflammatory markers and arterial stiffness indexes in subjects with acute ischemic stroke with and without metabolic syndrome.
        Diabetol. Metab. Syndrome. 2014; 6: 28
        • Kim S.K.
        • Kim K.S.
        • Lee Y.S.
        • Park S.H.
        • Choe J.Y.
        Arterial stiffness and proinflammatory cytokines in fibromyalgia syndrome.
        Clin. Exp. Rheumatol. 2010; 28: S71-S77
        • Gomez-Diaz R.A.
        • Gutierrez J.
        • Contreras-Rodriguez A.
        • Valladares-Salgado A.
        • Tanus-Hajj J.
        • Mondragon-Gonzalez R.
        • et al.
        Association of V249I and T280M variants of fractalkine receptor CX3CR1 with carotid intima-media thickness in a mexican population with type 2 diabetes.
        Gac. Med. Mex. 2017; 153: 49-56
        • Apostolakis S.
        • Spandidos D.
        Chemokines and atherosclerosis: focus on the CX3CL1/CX3CR1 pathway.
        Acta Pharmacol. Sin. 2013; 34: 1251-1256
        • Stanley E.R.
        • Berg K.L.
        • Einstein D.B.
        • Lee P.S.
        • Pixley F.J.
        • Wang Y.
        • et al.
        Biology and action of colony--stimulating factor-1.
        Mol. Reprod. Dev. 1997; 46: 4-10
        • Lawrence T.
        • Natoli G.
        Transcriptional regulation of macrophage polarization: enabling diversity with identity.
        Nat. Rev. Immunol. 2011; 11: 750-761
        • Trus E.
        • Basta S.
        • Gee K.
        Who's in charge here? Macrophage colony stimulating factor and granulocyte macrophage colony stimulating factor: competing factors in macrophage polarization.
        Cytokine. 2019; 127: 154939
        • Smith J.D.
        • Trogan E.
        • Ginsberg M.
        • Grigaux C.
        • Tian J.
        • Miyata M.
        Decreased atherosclerosis in mice deficient in both macrophage colony-stimulating factor (op) and apolipoprotein E.
        Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 8264-8268
        • Yu A.
        • Giannone T.
        • Scheffler P.
        • Doonan R.J.
        • Egiziano G.
        • Gomez Y.H.
        • et al.
        The effect of oral contraceptive pills and the natural menstrual cYCLe on arterial stiffness and hemodynamICs (CYCLIC).
        J. Hypertens. 2014; 32: 100-107
        • Hickson S.S.
        • Miles K.L.
        • McDonnell B.J.
        • Yasmin
        • Cockcroft J.R.
        • Wilkinson I.B.
        • et al.
        Use of the oral contraceptive pill is associated with increased large artery stiffness in young women: the ENIGMA study.
        J. Hypertens. 2011; 29: 1155-1159
        • Priest S.E.
        • Shenouda N.
        • MacDonald M.J.
        Effect of sex, menstrual cycle phase, and monophasic oral contraceptive pill use on local and central arterial stiffness in young adults.
        Am. J. Physiol. Heart Circ. Physiol. 2018; 315: H357-H365
        • Jick S.S.
        • Jick H.
        Cerebral venous sinus thrombosis in users of four hormonal contraceptives: levonorgestrel-containing oral contraceptives, norgestimate-containing oral contraceptives, desogestrel-containing oral contraceptives and the contraceptive patch.
        Contraception. 2006; 74: 290-292