Advertisement

Plasma autoantibodies against apolipoprotein B-100 peptide 210 in subclinical atherosclerosis

      Highlights

      • We examined the role of apoB-100 autoantibodies in early atherosclerosis.
      • Autoantibody levels differed across centers according to latitude.
      • ApoB-100 autoantibodies proved to be implicated in subclinical atherosclerosis.
      • Results provide indications of a causal protective role for IgM-p210MDA.
      • No strong relationship was observed between IgG-p210nat and early atherosclerosis.

      Abstract

      Objective

      Experimental studies have suggested that autoimmunity is involved in atherosclerosis and provided evidence that both protective and pro-atherogenic immune responses exist. This concept has received support from small clinical studies implicating autoantibodies directed against apolipoprotein B-100 (apoB-100) in human atherosclerosis. We examined circulating autoantibodies directed against native and malondialdehyde (MDA)-modified epitope p210 of apoB-100 (IgG-p210nat and IgM-p210MDA) in relation to early atherosclerosis in a large, European longitudinal cohort study of healthy high-risk individuals.

      Approach and results

      IgG-p210nat and IgM-p210MDA were quantified in baseline plasma samples of 3430 participants in the IMPROVE study and related to composite and segment-specific measures of severity and rate of progression of carotid intima-media thickness (cIMT) determined at baseline and after 30 months. IgM-p210MDA autoantibody levels were independently related to several cIMT measures both in the common carotid artery and in the carotid bulb, including measures of cIMT progression, higher levels being associated with lower cIMT or slower cIMT progression. Consistent inverse relationships were also found between plasma levels of IgG-p210nat and baseline composite measures of cIMT. These associations disappeared when adjusting for established and emerging risk factors, and there were no associations with rate of cIMT progression besides in certain secondary stratified analyses.

      Conclusions

      The present study provides further evidence of involvement of autoantibodies against native and MDA-modified apoB-100 peptide 210 in cardiovascular disease in humans and demonstrates that these associations are present already at a subclinical stage of the disease.

      Keywords

      Abbreviations

      apoB-100
      apolipoprotein B100
      BIC
      Bayesian information criterion
      Bif-IMT
      bifurcation of the carotid artery
      CC-IMT
      common carotid intima-media thickness
      cIMT
      carotid intima-media thickness
      CVD
      cardiovascular disease
      HDL
      high-density lipoprotein
      ICA
      internal carotid artery
      IgG-p210nat
      immunoglobulin G directed against native p210
      IgM-p210MDA
      immunoglobulin M directed against malondealdehyde-modified p210
      IMPROVE
      study of carotid IMT and IMT progression as predictors of vascular events in a high risk population
      IMT
      intima-media thickness
      MDA
      malondealdehyde
      MDS
      multidimensional scaling
      oxLDL
      oxidized low-density lipoprotein
      SD
      standard deviation
      SNP
      single nucleotide polymorphism

      1. Introduction

      Atherosclerosis is a chronic inflammatory disease, in which adaptive immunity is considered to play an important role. In particular, immune responses directed against autoantigens in oxidized low-density lipoprotein (oxLDL) represent a key process in atherogenesis [
      • Nilsson J.
      • Hansson G.K.
      Autoimmunity in atherosclerosis: a protective response losing control?.
      ,
      • Hansson G.K.
      • Hermansson A.
      The immune system in atherosclerosis.
      ,
      • Ketelhuth D.F.
      • Tonini G.C.
      • Carvalho M.D.
      • Ramos R.F.
      • Boschcov P.
      • Gidlund M.
      Autoantibody response to chromatographic fractions from oxidized LDL in unstable angina patients and healthy controls.
      ,
      • Binder C.J.
      Natural IgM antibodies against oxidation-specific epitopes.
      ]. Thus, T-cells cloned from human lesions respond to oxLDL in a major histocompatibility complex class II-dependent manner, atherosclerotic lesions contain high amounts of antibodies recognizing various epitopes of oxLDL and hypercholesterolaemic animals with genetic deficiencies introduced in the adaptive immune system develop less atherosclerosis [
      • Stemme S.
      • Faber B.
      • Holm J.
      • Wiklund O.
      • Witztum J.L.
      • Hansson G.K.
      T lymphocytes from human atherosclerotic plaques recognize oxidized low density lipoprotein.
      ,
      • Fredrikson G.N.
      • Hedblad B.
      • Berglund G.
      • et al.
      Identification of immune responses against aldehyde-modified peptide sequences in apoB associated with cardiovascular disease.
      ,
      • Nilsson J.
      • Hansson G.K.
      • Shah P.K.
      Immunomodulation of atherosclerosis: implications for vaccine development.
      ,
      • Hansson G.K.
      • Libby P.
      The immune response in atherosclerosis: a double-edged sword.
      ,
      • Goncalves I.
      • Nitulescu M.
      • Ares M.P.
      • et al.
      Identification of the target for therapeutic recombinant anti-apoB-100 peptide antibodies in human atherosclerotic lesions.
      ,
      • Hansson G.K.
      • Jonasson L.
      The discovery of cellular immunity in the atherosclerotic plaque.
      ]. However, anti-atherogenic effects of immune responses against oxLDL have also been reported from studies where experimental animals were immunized with oxLDL particles [
      • Palinski W.
      • Miller E.
      • Witztum J.L.
      Immunization of low density lipoprotein (LDL) receptor-deficient rabbits with homologous malondialdehyde-modified LDL reduces atherogenesis.
      ,
      • Ameli S.
      • Hultgardh-Nilsson A.
      • Regnstrom J.
      • et al.
      Effect of immunization with homologous LDL and oxidized LDL on early atherosclerosis in hypercholesterolemic rabbits.
      ,
      • Zhou X.
      • Caligiuri G.
      • Hamsten A.
      • Lefvert A.K.
      • Hansson G.K.
      LDL immunization induces T-cell-dependent antibody formation and protection against atherosclerosis.
      ] or specific components of oxLDL such as modified phospholipids or fragments of apolipoprotein B-100 (apoB-100) [
      • Binder C.J.
      • Horkko S.
      • Dewan A.
      • et al.
      Pneumococcal vaccination decreases atherosclerotic lesion formation: molecular mimicry between Streptococcus pneumoniae and oxidized LDL.
      ,
      • Sun J.
      • Hartvigsen K.
      • Chou M.Y.
      • et al.
      Deficiency of antigen-presenting cell invariant chain reduces atherosclerosis in mice.
      ,
      • Wigren M.
      • Kolbus D.
      • Duner P.
      • et al.
      Evidence for a role of regulatory T cells in mediating the atheroprotective effect of apolipoprotein B peptide vaccine.
      ]. Among different factors implicated in diverse inflammatory pathogenic mechanisms of atherosclerosis, certain patterns have been described [
      • Gasparyan A.Y.
      • Stavropoulos-Kalinoglou A.
      • Mikhailidis D.P.
      • Toms T.E.
      • Douglas K.M.
      • Kitas G.D.
      The rationale for comparative studies of accelerated atherosclerosis in rheumatic diseases.
      ]. Overproduction of antibodies against oxLDL has been associated with accelerated atherosclerosis and increased cardiovascular risk in inflammatory conditions such as systemic lupus erythematosus and primary antiphospholipid syndrome, but not in rheumatoid arthritis [
      • Gasparyan A.Y.
      • Stavropoulos-Kalinoglou A.
      • Mikhailidis D.P.
      • Toms T.E.
      • Douglas K.M.
      • Kitas G.D.
      The rationale for comparative studies of accelerated atherosclerosis in rheumatic diseases.
      ]. Circulating antibodies have been identified in human plasma against a large number of peptide sequences of apoB-100, both native and malondialdehyde (MDA)-modified, some of which (such as peptides 45 and 210) are related to atherosclerosis [
      • Fredrikson G.N.
      • Hedblad B.
      • Berglund G.
      • et al.
      Identification of immune responses against aldehyde-modified peptide sequences in apoB associated with cardiovascular disease.
      ]. These findings prompted immunization studies in apoE knockout mice with some of these native and aldehyde-modified peptides, or the corresponding antibodies, which resulted in up to 70% reduction in atherosclerosis [
      • Klingenberg R.
      • Lebens M.
      • Hermansson A.
      • et al.
      Intranasal immunization with an apolipoprotein B-100 fusion protein induces antigen-specific regulatory T cells and reduces atherosclerosis.
      ]. The corollary of these studies is that activation of adaptive immunity against specific peptides of apoB-100 may be atheroprotective. Accordingly, recent clinical cohort studies of circulating antibodies against native and MDA-modified peptides 45 and 210 have demonstrated associations with both severity of carotid and coronary atherosclerosis as well as with risk of clinical events [
      • Fredrikson G.N.
      • Schiopu A.
      • Berglund G.
      • Alm R.
      • Shah P.K.
      • Nilsson J.
      Autoantibody against the amino acid sequence 661-680 in apo B-100 is associated with decreased carotid stenosis and cardiovascular events.
      ,
      • Fredrikson G.N.
      • Hedblad B.
      • Berglund G.
      • et al.
      Association between IgM against an aldehyde-modified peptide in apolipoprotein B-100 and progression of carotid disease.
      ,
      • Sjogren P.
      • Fredrikson G.N.
      • Samnegard A.
      • et al.
      High plasma concentrations of autoantibodies against native peptide 210 of apoB-100 are related to less coronary atherosclerosis and lower risk of myocardial infarction.
      ,
      • Fredrikson G.N.
      • Anand D.V.
      • Hopkins D.
      • et al.
      Associations between autoantibodies against apolipoprotein B-100 peptides and vascular complications in patients with type 2 diabetes.
      ]. Specifically, antibodies directed against the native peptide 210 of apoB-100 seem to be protective, and this peptide emerges as a potential target for immunization against atherosclerosis in humans.
      So far, clinical studies examining the significance of autoantibodies against apoB-100 for human atherosclerosis and its accompanying potentially fatal complications of myocardial infarction (MI) and stroke remain limited in number and restricted to small, diverse and specific populations. In the present study, we investigated the relationships of plasma levels of IgG antibodies directed against native peptide 210 (IgG-p210nat) respectively IgM antibodies against MDA-modified peptide 210 (IgM-p210MDA) of apoB-100 to severity and rate of progression of carotid intima-media thickness (cIMT) in a large, multicentre, European longitudinal cohort study (acronym: IMPROVE) [
      • Baldassarre D.
      • Nyyssonen K.
      • Rauramaa R.
      • et al.
      Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study.
      ]. Participants were free of clinically overt cardiovascular disease (CVD) at enrollment and had at least 3 established cardiovascular risk factors. This setting allowed us to examine the role of apoB autoantibodies in relation to early atherosclerosis and arterial remodeling in asymptomatic individuals at increased risk of CVD distributed across the established north-to-south cardiovascular risk gradient in Europe [
      • Sans S.
      • Kesteloot H.
      • Kromhout D.
      The burden of cardiovascular diseases mortality in Europe. Task force of the European Society of Cardiology on Cardiovascular Mortality and Morbidity Statistics in Europe.
      ].

      2. Materials and methods

      2.1 Study population

      The biobank and databases of the IMPROVE study formed the basis of the present study. In brief, IMPROVE is a multicentre, longitudinal, observational cohort study of high-risk individuals, involving seven recruiting centers in five European countries: Finland, Sweden, the Netherlands, France and Italy [
      • Baldassarre D.
      • Nyyssonen K.
      • Rauramaa R.
      • et al.
      Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study.
      ]. A total of 3711 participants, men and women, were enrolled between March 2004 and April 2005, 3430 of whom were included in the present report. Reasons for exclusion or drop-out from the present investigation are provided in Fig. 1. Eligibility criteria included age from 55 to 79 years, presence of at least three cardiovascular risk factors and absence of symptoms of CVD as well as of conditions that might limit longevity or cIMT visualization. A total of 92 participants included in the present report did not complete the follow-up, leaving 3338 subjects for autoantibody analyses in relation to change over time in cIMT.
      Figure thumbnail gr1
      Fig. 1Flow chart illustrating reasons for exclusion of IMPROVE participants from the present study.
      Baseline characteristics of the entire IMPROVE cohort and methods for determination of established cardiovascular risk factors have been described [
      • Baldassarre D.
      • Nyyssonen K.
      • Rauramaa R.
      • et al.
      Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study.
      ]. Characteristics of participants included in the present report are provided in Supplementary Table 2. Single nucleotide polymorphism (SNP) genotypes generated by genotyping with the Illumina CardioMetabo 200K BeadArray were available in all participants. There were no differences between subjects who completed the study and those lost to follow-up with respect to age, sex, body mass index (BMI), diastolic blood pressure (DBP), established and emerging biochemical risk factors, and common carotid intima-media thickness (CC-IMT) at baseline. However, subjects who completed follow-up had a slightly lower baseline systolic blood pressure (SBP) (data not shown).

      2.2 Autoantibody determinations

      Circulating IgG-p210nat and IgM-p210MDA antibodies (directed against peptide p210 containing amino acids 3136–3155 of apoB-100) were quantified by an in-house ELISA in EDTA plasma samples, which had been prepared from blood samples drawn at baseline and stored at −80 °C until analysis (Supplementary Table 1). Procedures for generation of the synthetic p210nat and p210MDA, and details of the ELISA protocol have been described [
      • Fredrikson G.N.
      • Hedblad B.
      • Berglund G.
      • et al.
      Identification of immune responses against aldehyde-modified peptide sequences in apoB associated with cardiovascular disease.
      ] and are also provided as Supplementary Material. Antibody levels for a few samples, 6 for IgG-p210nat and 3 for IgM-p210MDA, were below the detection limit on repeated measurements and therefore replaced with a value confidently below the detection limit (1/100 of the lowest measured value) to allow inclusion of the samples.

      2.3 Carotid ultrasonography

      The carotid ultrasound protocol and precision of the ultrasonographic measurements have been reported [
      • Baldassarre D.
      • Nyyssonen K.
      • Rauramaa R.
      • et al.
      Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study.
      ]. The ultrasonographic variables selected for the present study were the mean and maximum IMT of the common carotid arteries (CC-IMTmean and CC-IMTmax), excluding the first centimeter closest to the bifurcation, the mean and maximum IMT of the bifurcations (Bif-IMTmean and Bif-IMTmax), the mean and maximum IMT of the internal carotid artery (ICA-IMTmean and ICA-IMTmax), the mean and maximum IMT of the whole carotid tree (IMTmean and IMTmax), and the average of maximum IMT values of the whole carotid tree (IMTmean–max). The right and left carotid arteries were also evaluated separately in stratified analyses. To evaluate change of cIMT over time (the former designated cIMT progression), ultrasonographic measurements made at 0 and 30 months were compared. cIMT changes at 30 months, expressed in mm/year, were calculated by linear regression of IMT changes versus time. cIMT measurements were performed in a centralized laboratory (Department of Pharmacological Sciences – University of Milan, Italy) using a dedicated software (M'Ath, Metris SRL France) [
      • Beux F.
      • Carmassi S.
      • Salvetti M.V.
      • et al.
      Automatic evaluation of arterial diameter variation from vascular echographic images.
      ], that allows semi-automatic edge detection of the echogenic lines of the intima-media complex.

      2.4 Ethical considerations

      Ethics committee approvals for the study were obtained by the 7 recruiting centers. All patients gave separate written informed consents for general participation in the study and for genotyping.

      2.5 Statistical analysis

      Variables are reported as number (percentages), mean ± standard deviation (SD) or median (interquartile range). Variables with skewed distributions (e.g. ultrasonographic variables and plasma triglycerides) were log-transformed before they were used in linear regression analysis. Smoking habits were identified as described earlier [
      • Baldassarre D.
      • Nyyssonen K.
      • Rauramaa R.
      • et al.
      Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study.
      ]. Cumulative life-time smoking was expressed as a 5-level categorical variable according to never-smoker status and quartiles of pack-years. Hypertension was defined as a diagnosis of hypertension and/or SBP of at least 140 mmHg and/or DBP of at least 90 mmHg and/or treatment with antihypertensive drugs. Among 3430 subjects included in the current study 900 subjects had type 2 diabetes mellitus. Diabetes was considered to be present in participants who had a diagnosis of diabetes, were on insulin and/or other hypoglycemic drugs, or had a fasting blood glucose ≥7 mmol/L at the baseline examination.
      Differences in autoantibody levels across recruitment centers were analyzed by Kruskal–Wallis non-parametric one-way analysis of variance. The Jonckheere–Terpstra test for ordered alternatives was used to assess trends across centers defined by latitude. Associations between IgG-p210nat and IgM-p210MDA levels and established and emerging cardiovascular risk factors were assessed by Spearman rank correlation. Multivariable analysis of factors associated with autoantibody levels was performed by multiple linear regression analysis. Multiple robust regression, which is less sensitive to influence of outliers, was used to assess associations between autoantibody levels and cIMT measurements [
      • Renauda O.
      • Victoria-Feserb M.-P.
      A robust coefficient of determination for regression.
      ]. Age and sex were always forced into the regression models. To identify variables to be included in the multivariable analysis, a best subset analysis [
      • Tibshirani R.
      • Friedman J.H.
      The elements of statistical learning: data mining, inference, and prediction. Springer series in statistics.
      ] based on the Bayesian information criterion (BIC) was performed. All of the following variables with a p-value <0.20 in a univariable regression were considered for inclusion in the multivariable analysis: latitude, weight, height, waist circumference, BMI, SBP and DBP, hypertension, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, C-reactive protein, diabetes, blood glucose, smoking status, current cigarette consumption, cumulative smoking (pack-years), and physical activity.
      Partial correlation coefficients were calculated for the relationships between plasma autoantibody concentrations and baseline respectively progression measures of cIMT, using robust regression [
      • Huber P.J.
      Robust estimation of a location parameter.
      ] and two sets of covariates. In the first model, age, sex and latitude were used as covariates (plus the IgM-p210MDA-sex interaction term as appropriate). In the second model, clinical and biochemical risk indicators selected by best subset analysis (see above) were added to the basic set of covariates. Stratification by gender or by side of the carotid circulation was made where appropriate.
      SNP genotypes generated with the Illumina CardioMetabo 200K BeadArray were used to control for population substructure. Multidimensional scaling (MDS) coordinates, reflecting genetic distances between individuals (population substructure), were calculated based on the genotype data using PLINK1.07 [
      • Purcell S.
      • Neale B.
      • Todd-Brown K.
      • et al.
      PLINK: a tool set for whole-genome association and population-based linkage analyses.
      ]. Outliers identified by visual inspection of plots of the first three MDS dimensions (MDS1-3) were excluded. Of note, MDS1 and latitude was strongly correlated (r = 0.942, p < 0.001).
      No adjustment for multiple testing was performed since both baseline cIMT measurements and cIMT progression measurements were strongly correlated.
      A two-sided p-value <0.05 was considered significant for all analyses. STATA 11.1 (StataCorp LP) and SAS 9.2 for Windows were used for the statistical analyses.

      3. Results

      3.1 Autoantibody levels according to center and in relation to established and emerging risk factors

      Both IgG-p210nat and IgM-p210MDA levels differed significantly across centers (Fig. 2). The Jonckheere–Terpstra (J) test for ordered alternatives showed a significant trend for lower IgG-p210nat levels in the north (J (corrected for ties) = −2.96, p = 0.003) whereas no trend in either direction was found for IgM-p210MDA levels. Differences in established and emerging risk factors and medication across centers did not account for the between-center differences in autoantibody levels (data not shown). Between-individual differences in IgG-p210nat (SD 0.34 U) were greater than in IgM-p210MDA (SD 0.19 U), whereas mean IgM-p210MDA (0.92 U) levels were higher than mean IgG-p210nat (0.88 U) levels (p < 0.001). IgM-p210MDA levels were higher in women (0.93 ± 0.004 U) than in men (0.91 ± 0.005 U, p = 0.013), but not IgG-p210nat (p = 0.999). Subjects with diabetes had significantly lower IgG-p210nat levels (0.85 ± 0.323 vs 0.90 ± 0.346 U, p < 0.001) whereas IgM-p210MDA levels did not differ between those with and without diabetes. Significantly higher levels of IgM-p210MDA (0.92 ± 0.186 vs 0.90 ± 0.186 U, p = 0.001) were noted in hypertensive patients. Patients on lipid-lowering medication had significantly lower titers of IgM-p210MDA (0.91 ± 0.187 U) than those without (0.93 ± 0.183 U, p = 0.003) whereas IgG-p210nat levels did not differ with respect to lipid-lowering medication.
      Figure thumbnail gr2
      Fig. 2Plasma levels of IgG antibodies directed against native peptide 210 (IgG-p210nat) respectively IgM antibodies against MDA-modified peptide 210 (IgM-p210MDA) of apolipoprotein B-100 in subjects grouped according to recruitment center (with latitude indicated in parenthesis for each center). Box plots with darker shade are used for IgG-p210nat, whereas box plots with brighter shade are used for IgM-p210MDA. Differences between centers were explored by using the Kruskal–Wallis test: p < 0.0001 for IgG-p210nat; p < 0.0001 for IgM-p210MDA.
      Overall, the relationships of antibody concentrations to established and emerging risk factors were weak (Table 1). Significant inverse correlations were observed between IgG-p210nat and latitude, BMI, systolic blood pressure, triglycerides, presence of manifest diabetes, blood glucose and cigarette smoking, whereas positive correlations were noted for age and physical activity. In addition, IgG-p210nat levels were inversely correlated with the first multidimensional scaling dimension, MDS1 (r = −0.053, p = 0.002). For IgM-p210MDA, positive correlations were found with female sex, systolic blood pressure, presence of hypertension, C-reactive protein, and smoking status. Contrary to IgG-p210nat, IgM-p210MDA was not associated with MDS1 (r = −0.013, p = 0.455). Gender-specific correlations between autoantibodies and cardiovascular risk factors are provided in Supplementary Table 3.
      Table 1Relationships of plasma levels of autoantibodies against native and MDA-modified peptide 210 of apolipoprotein B-100 to established and emerging risk factors.
      Risk factorIgG-P210natIgM-P210MDA
      Age0.059†−0.016
      Sex (male/female)0.0000.043*
      Latitude−0.051**−0.029
      BMI−0.034*0.016
      Waist circumference−0.0200.014
      Systolic blood pressure−0.035*0.034*
      Diastolic blood pressure−0.0220.013
      Hypertension (no/yes)0.0180.053**
      LDL-cholesterol0.0150.011
      HDL-cholesterol0.0170.005
      Triglycerides−0.065†0.029
      CRP−0.0240.051**
      Diabetes (no/yes)−0.074†−0.007
      Blood glucose−0.055**0.003
      Smoking status (never/ever)−0.083†0.041*
      Pack-years (categories)−0.086†0.033
      Cigarettes per day−0.075†0.019
      High physical activity (no/yes)0.039*−0.021
      n = 3430. Values are Spearman rank correlation coefficients.
      MDA, malondialdehyde; IgG-p210nat, IgG antibodies directed against native apoB peptide 210; IgM-p210MDA, IgM antibodies directed against MDA-modified apoB peptide p210; BMI, body mass index; LDL, low-density lipoprotein; HDL, high-density lipoprotein; CRP, high-sensitivity C-reactive protein.
      *p < 0.05; **p < 0.01; †p < 0.001.
      Table 2 reports the multivariable analyses of the relationships of established and emerging risk factors to autoantibody levels. In all, independently related variables accounted for a total of 1.9% and 1.5% of the variations in IgG-p210nat and IgM-p210MDA, respectively. Adjustment for medication showed that IgG-p210nat levels were completely independent of any type of medication (lipid-lowering, glucose-lowering, anti-hypertensive, platelet-inhibiting and anti-inflammatory), which hence did not influence the risk factor relationships to IgG-p210nat, whereas lipid-lowering and anti-hypertensive medication was significantly related to IgM-p210MDA but added very little (<0.5%) to the variation in IgM-p210MDA accounted for by the model (data not shown). Multiple linear regression analyses stratified by gender are provided in Supplementary Table 4. Age and SBP respectively age and hypertension were common determinants in men and women for IgG-p210nat and IgM-p210MDA, correspondingly.
      Table 2Multiple linear regression analyses of risk indicators related to levels of autoantibodies against native and MDA-modified peptide 210 of apolipoprotein B-100 in human plasma.
      Risk factorIgG-p210natIgM-p210MDA
      Partial rP-valuePartial rP-value
      Age0.0510.003−0.0430.014
      Sex−0.0100.5860.0440.011
      Latitude−0.0410.018
      Weight−0.0270.126
      HDL-cholesterol−0.0230.191
      LDL-cholesterol0.0240.158
      Triglycerides0.0330.059
      Smoking status (never/ever)−0.0590.0010.0520.003
      CRP−0.0310.0780.0270.114
      Systolic blood pressure−0.0570.001
      Diastolic blood pressure0.0310.069
      Hypertension (no/yes)0.0370.0340.069<0.001
      High physical activity (no/yes)0.0550.009−0.0240.172
      Diabetes (no/yes)−0.0510.003
      Model multiple r20.019<0.0010.015<0.001
      Variables included in the multiple linear regression analyses were selected by best subset analysis. Age and sex were forced into the best subset analysis. MDA, malondialdehyde; r, partial correlation coefficient; IgG-p210nat, IgG antibodies directed against native apoB peptide 210; IgM-p210MDA, IgM antibodies directed against MDA-modified apoB peptide p210; HDL, high-density lipoprotein; LDL, low-density lipoprotein; CRP, high-sensitivity C-reactive protein.

      3.2 Relationships of autoantibody levels to baseline cIMT

      Autoantibodies against apoB-100 peptides were further analyzed in relation to baseline cIMT (Table 3). Adjustments for age, sex and latitude (Table 3) revealed statistically significant inverse associations between the IgG-p210nat level and the composite cIMT measures of IMTmean, IMTmax and IMTmean–max. In addition, there was an inverse association between IgG-p210nat and Bif-IMTmax. IgM-p210MDA levels were only significantly inversely related to Bif-IMTmean and Bif-IMTmax. After inclusion of additional covariates (Table 3), the IgG-p210nat relationships to composite cIMT measures and Bif-IMTmax weakened and became statistically insignificant, whereas the IgM-p210MDA relationships to Bif-IMTmean and Bif-IMTmax were largely unaltered. Also, IgM-p210MDA now appeared as being significantly related to IMTmean. After introducing medication into the models, the significant relationship remained between IgM-p210MDA and Bif-IMTmean (Supplementary Table 5).
      Table 3Relationships of levels of autoantibodies against native and MDA-modified peptide p210 of apolipoprotein B-100 to baseline dimension and change over time of carotid artery intima-media thickness.
      Regression model 1Regression model 2
      Baseline dimensionChange over timeBaseline dimensionChange over time
      β ± 2SEpβ ± 2SEpβ ± 2SEpβ ± 2SEp
      IgG-p210natIMTmean−0.0085 ± 0.00840.043−0.0026 ± 0.00300.092−0.0038 ± 0.00820.352−0.0018 ± 0.00320.240
      IMTmax−0.0229 ± 0.01640.005−0.0018 ± 0.01260.775−0.0159 ± 0.01640.0530.0025 ± 0.01280.696
      IMTmean–max−0.0112 ± 0.00900.012−0.0024 ± 0.00480.314−0.0064 ± 0.00880.146−0.0021 ± 0.00480.399
      CC-IMTmean−0.0022 ± 0.00640.496−0.0021 ± 0.00240.0800.0011 ± 0.00640.728−0.0017 ± 0.00240.154
      CC-IMTmax−0.0057 ± 0.00920.218−0.0038 ± 0.00560.177−0.0017 ± 0.00920.708−0.0028 ± 0.00580.327
      Bif-IMTmean−0.0129 ± 0.01380.064−0.0040 ± 0.00640.203−0.0087 ± 0.01380.210−0.0035 ± 0.00640.275
      Bif-IMTmax−0.0191 ± 0.01680.024−0.0086 ± 0.01220.159−0.0143 ± 0.01700.092−0.0080 ± 0.01220.189
      ICA-IMTmean−0.0056 ± 0.01360.405−0.0039 ± 0.00440.078−0.0004 ± 0.01360.948−0.0036 ± 0.00440.107
      ICA-IMTmax−0.0120 ± 0.01880.199−0.0045 ± 0.00960.347−0.0025 ± 0.01880.793−0.0036 ± 0.00980.463
      IgM-p210MDAIMTmean−0.0145 ± 0.02140.175−0.0059 ± 0.00800.141−0.0271 ± 0.02040.008−0.0052 ± 0.00820.199
      IMTmax−0.0236 ± 0.04200.261−0.0031 ± 0.03280.849−0.0318 ± 0.04160.127−0.0043 ± 0.01680.798
      IMTmean–max−0.0110 ± 0.02260.331−0.0146 ± 0.01260.020−0.0178 ± 0.02240.113−0.0132 ± 0.00630.037
      CC-IMTmean0.0021 ± 0.01640.792−0.0070 ± 0.00620.022−0.0025 ± 0.01620.758−0.0046 ± 0.03360.157
      CC-IMTmax0.0108 ± 0.02340.356−0.0141 ± 0.01480.0550.0050 ± 0.02340.671−0.0097 ± 0.01500.195
      Bif-IMTmean−0.0358 ± 0.03520.042−0.0031 ± 0.01640.708−0.0423 ± 0.03520.016−0.0034 ± 0.01680.691
      Bif-IMTmax−0.0424 ± 0.04300.048−0.0055 ± 0.03160.728−0.0459 ± 0.04100.0330.0057 ± 0.03160.718
      ICA-IMTmean−0.0220 ± 0.03420.2000.0010 ± 0.01140.868−0.0273 ± 0.03480.1160.0015 ± 0.01140.796
      ICA-IMTmax−0.0178 ± 0.04760.454−0.0306 ± 0.02500.014−0.0232 ± 0.04760.330−0.0343 ± 0.02520.007
      Model 1: robust regression after adjustment for age, sex and latitude.
      Model 2: robust regression after adjustment for age, sex and latitude, and clinical and biochemical variables selected by best subset analysis.
      For IgM-p210MDA, adjustment was also made for IgM-p210MDA-sex interaction. β, regression coefficient; SE, standard error; IgG-p210nat, IgG antibodies directed against native apoB peptide 210; IgM-p210MDA, IgM antibodies directed against MDA-modified apoB peptide p210; IMTmean, mean IMT of the whole carotid tree; IMTmax, maximum IMT of the whole carotid tree; IMTmean–max, average of maximum IMT values of the whole carotid tree; CC, common carotid artery; Bif, bifurcation (bulb); ICA, internal carotid artery. IMT measurements were log-transformed to make them more symmetrically distributed.
      Stratified analysis by gender is provided in Supplementary Table 6. Separate examinations of the right and left carotid arteries (Supplementary Table 7) indicated that effects appeared to be stronger on the right side for the composite and segment-specific measures that were associated with IgG-p210nat antibody levels in the whole group. However, specific inverse relationships were noted between IgM-p210MDA levels and the left ICA-IMTmean and ICA-IMTmax measures.

      3.3 Relationships of autoantibody levels to cIMT progression

      Autoantibodies against apoB-100 peptides were also analyzed in relation to change in cIMT over time (Table 3). Statistically significant associations were sparse. Inverse relationships that were independent of cardiovascular risk factors were noted for IgM-p210MDA in relation to the measures of IMTmean–max and ICA-IMTmax progression. These relationships remained statistically significant after additional adjustment for any type of medication (Supplementary Table 5). Stratification by gender showed significant inverse associations of IgM-p210MDA, which were independent of cardiovascular risk factors, with changes in ICA-IMTmax in men and IMTmean–max in women (Supplementary Table 6). Further stratified analyses according to presence of at least one plaque (IMTmax > 1.5 mm) revealed significant inverse relationships of IgG-p210nat levels to progression of ICA-IMTmean in subjects with plaques and of IgM-p210MDA levels to progression of IMTmean–max in subject without plaques, after adjustment for age, gender and latitude (data not shown). Separate analyses of the right and left carotid arteries showed significant inverse associations between IgG-p210nat levels and progression of left IMTmean, IMTmean–max and Bif-IMTmean, and a significant inverse association between IgM-p210MDA levels and progression of right ICA-IMTmax.

      4. Discussion

      This study is by far the largest clinical investigation of the role of autoantibodies directed against apoB-100 in CVD. Focus was placed on retrospectively assessing associations of baseline levels of specific autoantibodies with subclinical atherosclerosis as determined by ultrasound determination of baseline cIMT and cIMT progression. The choice of middle-aged to elderly high-risk individuals was considered optimal for exploring autoantibody associations with disease progression. By enrolling equal numbers of male and female participants from different parts of Europe, the study allows for data to be interpreted in a wider context than in previous studies.
      Some findings should be considered as particularly important. First, differences in both IgG-p210nat and IgM-p210MDA levels were found to exist between centers, which were not accounted for by differences in risk factor profiles or treatment, nor were they due to center-specific differences in laboratory performance since all assays were run by the same analyst in a central laboratory (at the Centre for Molecular Medicine, Karolinska Institutet, Stockholm, Sweden). These observations suggest that IgG-p210nat and IgM-p210MDA levels are largely regulated by mechanisms that are independent of established risk factors. One candidate mechanism for between-center difference in IgG-p210nat is genetic heterogeneity, as indicated by the association with the first MDS dimension. Interestingly, levels of the putatively protective IgG-p210nat were lower in the north. However, we could not demonstrate conclusive evidence that this contributes to the previously demonstrated north-to-south gradient in cIMT [
      • Baldassarre D.
      • Nyyssonen K.
      • Rauramaa R.
      • et al.
      Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study.
      ] and risk of CVD [
      • Sans S.
      • Kesteloot H.
      • Kromhout D.
      The burden of cardiovascular diseases mortality in Europe. Task force of the European Society of Cardiology on Cardiovascular Mortality and Morbidity Statistics in Europe.
      ].
      Secondly, consistent, albeit weak inverse associations were found between IgG-p210nat and baseline composite measures of cIMT in the entire cohort. These associations disappeared when adjusting for established and emerging risk factors, and there were no associations with rate of cIMT progression. This would speak against a direct causal protective role of IgG-p210nat in early atherosclerosis and either suggests that these antibodies are innocent bystanders or that they exert their effect through conventional risk factor mechanisms. In the former scenario, some components of the metabolic syndrome, which by themselves promote atherosclerosis, would potentially depress the adaptive immune system leading to lower levels of antibodies against native p210. Conversely, a lower risk factor burden would be associated with less atherosclerosis and a more active adaptive immune system. In the latter scenario, higher levels of IgG-p210nat, reflecting a stronger activation of adaptive immunity against certain epitopes, would counteract some facets of the metabolic syndrome whereas lower levels would lead to a poorer risk factor control. However, in secondary analyses with stratification for the right and left carotid circulation, significant inverse associations were observed between IgG-p210nat levels and progression of left IMTmean, IMTmean–max and Bif-IMTmean. Of note, when interpreting these results, it should be remembered that the baseline measures of cIMT reflect processes that have occurred in the intima-media layers of the artery wall over decades prior to the autoantibody determinations whereas the measures of cIMT progression relate to changes taking place within a couple of years after plasma was drawn for autoantibody determination.
      In contrast, the cIMT results for IgM-p210MDA in the entire cohort showed independent inverse associations to bifurcation IMT measured at baseline, but not to the corresponding progression measures, whereas it was independently related to progression of IMTmean–max and ICA-IMTmax, a composite measure of lesion severity respectively a measure of maximum lesion size in the right and left internal carotid artery segments. Thus, contrary to IgG-p210nat, IgM-p210MDA autoantibody levels were independently related to some cIMT measures, including measures of cIMT progression, higher levels being associated with lower cIMT or slower cIMT progression. In addition, the pattern of independent associations for IgM-p210MDA was further reinforced in secondary analyses, particularly with stratification for gender.
      Studies of autoantibodies against apoB-100 performed so far in human subjects have examined diverse antibody species, patient groups and clinical phenotypes using different ultrasound protocols, rendering comparisons across studies complicated. Of two prior cIMT studies [
      • Fredrikson G.N.
      • Schiopu A.
      • Berglund G.
      • Alm R.
      • Shah P.K.
      • Nilsson J.
      Autoantibody against the amino acid sequence 661-680 in apo B-100 is associated with decreased carotid stenosis and cardiovascular events.
      ,
      • Fredrikson G.N.
      • Hedblad B.
      • Berglund G.
      • et al.
      Association between IgM against an aldehyde-modified peptide in apolipoprotein B-100 and progression of carotid disease.
      ], one focused on IgG- and IgM-p210MDA relationships to baseline cIMT and cIMT progression in symptomless subjects who, however, had a demonstrable carotid plaque [
      • Fredrikson G.N.
      • Hedblad B.
      • Berglund G.
      • et al.
      Association between IgM against an aldehyde-modified peptide in apolipoprotein B-100 and progression of carotid disease.
      ] whereas the other investigated relationships between the IgG- and IgM-p45MDA antibody species, cIMT and incident myocardial infarction (MI) or sudden cardiac death [
      • Fredrikson G.N.
      • Schiopu A.
      • Berglund G.
      • Alm R.
      • Shah P.K.
      • Nilsson J.
      Autoantibody against the amino acid sequence 661-680 in apo B-100 is associated with decreased carotid stenosis and cardiovascular events.
      ]. Comparisons between the two studies measuring IgM-p210MDA reveal similar findings of an inverse relationship with respect to baseline measurements of the bifurcation (Bif-IMT) but discrepant results for the common carotid (CC-IMT).
      High levels of IgG-p210nat autoantibodies have previously been found to be associated with less coronary artery calcification in persons with type-2 diabetes, but no clinical signs of coronary artery disease [
      • Fredrikson G.N.
      • Anand D.V.
      • Hopkins D.
      • et al.
      Associations between autoantibodies against apolipoprotein B-100 peptides and vascular complications in patients with type 2 diabetes.
      ], as well as with less severe coronary lesions in subjects who had suffered MI before the age of 60 [
      • Sjogren P.
      • Fredrikson G.N.
      • Samnegard A.
      • et al.
      High plasma concentrations of autoantibodies against native peptide 210 of apoB-100 are related to less coronary atherosclerosis and lower risk of myocardial infarction.
      ], suggesting a potential protective role of these autoantibodies in atherosclerosis. In the present study, inverse associations were also seen between IgG-p210nat levels and severity of subclinical atherosclerosis, as reflected by the baseline composite measures of cIMT. However, these associations were not independent of established and emerging risk factors. One of the possible mechanisms underlying the proatherogenic effects of IgG-p210nat could involve apolipoprotein H and, thus, facilitated binding of antiphospholipid antibodies to lipids of the endothelial cell membrane, eventually resulting in apoptosis of endothelial cells [
      • Gasparyan A.Y.
      • Stavropoulos-Kalinoglou A.
      • Mikhailidis D.P.
      • Toms T.E.
      • Douglas K.M.
      • Kitas G.D.
      The rationale for comparative studies of accelerated atherosclerosis in rheumatic diseases.
      ]. Based on the findings listed above, it could be speculated that IgG-p210nat autoantibodies may have a more important protective role in advanced as opposed to early subclinical atherosclerosis. Alternatively, autoantibodies directed against apoB-100 may exert different effects in the coronary as opposed to the carotid circulation. It should also be emphasized in this context that neither previous clinical studies nor the present one have demonstrated associations between high IgG-p210nat and slower disease progression or lowered risk of future clinical events. This fact can be interpreted to further question the cause–effect relationship between IgG-p210nat antibody levels and various cardiovascular phenotypes. The level of IgG-210nat was found to be lower in subjects with type 2 diabetes. The reason for this remains to be clarified. However, it has previously been shown that diabetic subjects with a poor metabolic control have lower levels of IgG-210nat levels [
      • Fredrikson G.N.
      • Anand D.V.
      • Hopkins D.
      • et al.
      Associations between autoantibodies against apolipoprotein B-100 peptides and vascular complications in patients with type 2 diabetes.
      ], which may have contributed to the findings in the present study. It is also interesting to note that IgG-210nat levels have been reported to be lower in type 2 diabetes patients with a high degree of coronary calcification (>400 Agatston units) than in those with low to moderate calcification [
      • Fredrikson G.N.
      • Anand D.V.
      • Hopkins D.
      • et al.
      Associations between autoantibodies against apolipoprotein B-100 peptides and vascular complications in patients with type 2 diabetes.
      ].
      The strengths of the present study include the sample size and the uniquely detailed ultrasound protocol. Whereas the European setting allows for data to be interpreted in a wider context than in previous studies and thus would add to the strengths of the study, it also potentially introduced heterogeneity in terms of population substructure. Of note, the first MDS dimension, a measure of genetic distance between individuals, was significantly correlated with IgG-p210nat autoantibody levels. Thus, studies of the genetic regulation of IgG-p210nat are warranted. The inter-individual variation in IgG-p210nat levels was greater than that for IgM-p210MDA. This contributed to a higher inter-assay variation of the IgG-p210nat ELISA. The focused autoantibody panel, which did not include the p45 epitope, although based on findings made by our group in relation to MI, constitutes a limitation. An additional limitation of the present study is that antibody levels were only determined at baseline; hence, we lack data regarding changes in antibody levels during follow-up. However, previous studies analyzing changes in autoantibodies against p210 over a 12-month period suggest that these remain relatively stable [
      • Fredrikson G.N.
      • Hedblad B.
      • Berglund G.
      • et al.
      Association between IgM against an aldehyde-modified peptide in apolipoprotein B-100 and progression of carotid disease.
      ]. In conclusion, the present study provides further evidence of an involvement of autoantibodies against the native and MDA-modified peptide 210 in atherosclerosis. Furthermore, the results obtained here provide stronger indications of a direct causal protective role for IgM-p210MDA than for IgG-p210nat. However, whereas mainly IgM-p210MDA autoantibody levels were inversely and independently related to some cIMT measures, including measures of cIMT progression, higher IgG-p210nat levels were consistently related to lower values of baseline global measures of cIMT, but these associations disappeared when controlled for the effects of cardiovascular risk factors. Further large-scale prospective cohort studies of autoantibodies directed towards apoB-100 are thus warranted that focus on both the p45 and the p210 epitopes. Such studies should combine a comprehensive analysis of clinical and biochemical risk factors and detailed cardiovascular imaging with adequate numbers of incident cardiovascular events and sequential blood sampling. Due to the existing evidence of accelerated progression of atherosclerosis in some rheumatic diseases characterized by inflammation [
      • Gasparyan A.Y.
      • Stavropoulos-Kalinoglou A.
      • Mikhailidis D.P.
      • Toms T.E.
      • Douglas K.M.
      • Kitas G.D.
      The rationale for comparative studies of accelerated atherosclerosis in rheumatic diseases.
      ], it is also worth examining to what extent antibodies against native and MDA-modified peptide 210 may be involved in inflammation-induced atherosclerosis. Until such studies have been completed, the mechanistic role and biomarker status of antibodies directed against the native and MDA-modified peptide epitopes on apoB-100 remain unresolved. However, these restrictions notwithstanding, the present study provides further support for exploring apoB-100 peptide 210 as a target for immunization against atherosclerosis in humans.

      5. Sources of funding

      The IMPROVE study was supported by the European Commission (Contract number: QLG1-CT-2002-00896), the Swedish Heart-Lung Foundation , the Swedish Research Council (projects 8691, 0593, 0125 and 8311), the Knut and Alice Wallenberg Foundation , the Foundation for Strategic Research , the Stockholm County Council (project 592229), the Strategic Cardiovascular and Diabetes Programmes of Karolinska Institutet and Stockholm County Council, the European Union Framework Programme 7 (FP7/2007–2013) for the Innovative Medicine Initiative under grant agreement n° IMI/115006 (the SUMMIT consortium), the Magnus Bergvall Foundation , the Academy of Finland (Grant # 110413 ) the British Heart Foundation ( RG2008/08 , RG2008/014 ) and the Italian Ministry of Health (Ricerca Corrente).

      Acknowledgments

      None.

      Appendix A. Supplementary data

      The following is the supplementary data related to this article:

      References

        • Nilsson J.
        • Hansson G.K.
        Autoimmunity in atherosclerosis: a protective response losing control?.
        J Intern Med. 2008; 263: 464-478
        • Hansson G.K.
        • Hermansson A.
        The immune system in atherosclerosis.
        Nat Immunol. 2011; 12: 204-212
        • Ketelhuth D.F.
        • Tonini G.C.
        • Carvalho M.D.
        • Ramos R.F.
        • Boschcov P.
        • Gidlund M.
        Autoantibody response to chromatographic fractions from oxidized LDL in unstable angina patients and healthy controls.
        Scand J Immunol. 2008; 68: 456-462
        • Binder C.J.
        Natural IgM antibodies against oxidation-specific epitopes.
        J Clin Immunol. 2010; 30: S56-S60
        • Stemme S.
        • Faber B.
        • Holm J.
        • Wiklund O.
        • Witztum J.L.
        • Hansson G.K.
        T lymphocytes from human atherosclerotic plaques recognize oxidized low density lipoprotein.
        Proc Natl Acad Sci U S A. 1995; 92: 3893-3897
        • Fredrikson G.N.
        • Hedblad B.
        • Berglund G.
        • et al.
        Identification of immune responses against aldehyde-modified peptide sequences in apoB associated with cardiovascular disease.
        Arterioscler Thromb Vasc Biol. 2003; 23: 872-878
        • Nilsson J.
        • Hansson G.K.
        • Shah P.K.
        Immunomodulation of atherosclerosis: implications for vaccine development.
        Arterioscler Thromb Vasc Biol. 2005; 25: 18-28
        • Hansson G.K.
        • Libby P.
        The immune response in atherosclerosis: a double-edged sword.
        Nat Rev Immunol. 2006; 6: 508-519
        • Goncalves I.
        • Nitulescu M.
        • Ares M.P.
        • et al.
        Identification of the target for therapeutic recombinant anti-apoB-100 peptide antibodies in human atherosclerotic lesions.
        Atherosclerosis. 2009; 205: 96-100
        • Hansson G.K.
        • Jonasson L.
        The discovery of cellular immunity in the atherosclerotic plaque.
        Arterioscler Thromb Vasc Biol. 2009; 29: 1714-1717
        • Palinski W.
        • Miller E.
        • Witztum J.L.
        Immunization of low density lipoprotein (LDL) receptor-deficient rabbits with homologous malondialdehyde-modified LDL reduces atherogenesis.
        Proc Natl Acad Sci U S A. 1995; 92: 821-825
        • Ameli S.
        • Hultgardh-Nilsson A.
        • Regnstrom J.
        • et al.
        Effect of immunization with homologous LDL and oxidized LDL on early atherosclerosis in hypercholesterolemic rabbits.
        Arterioscler Thromb Vasc Biol. 1996; 16: 1074-1079
        • Zhou X.
        • Caligiuri G.
        • Hamsten A.
        • Lefvert A.K.
        • Hansson G.K.
        LDL immunization induces T-cell-dependent antibody formation and protection against atherosclerosis.
        Arterioscler Thromb Vasc Biol. 2001; 21: 108-114
        • Binder C.J.
        • Horkko S.
        • Dewan A.
        • et al.
        Pneumococcal vaccination decreases atherosclerotic lesion formation: molecular mimicry between Streptococcus pneumoniae and oxidized LDL.
        Nat Med. 2003; 9: 736-743
        • Sun J.
        • Hartvigsen K.
        • Chou M.Y.
        • et al.
        Deficiency of antigen-presenting cell invariant chain reduces atherosclerosis in mice.
        Circulation. 2010; 122: 808-820
        • Wigren M.
        • Kolbus D.
        • Duner P.
        • et al.
        Evidence for a role of regulatory T cells in mediating the atheroprotective effect of apolipoprotein B peptide vaccine.
        J Intern Med. 2011; 269: 546-556
        • Gasparyan A.Y.
        • Stavropoulos-Kalinoglou A.
        • Mikhailidis D.P.
        • Toms T.E.
        • Douglas K.M.
        • Kitas G.D.
        The rationale for comparative studies of accelerated atherosclerosis in rheumatic diseases.
        Curr Vasc Pharm. 2010; 8: 437-449
        • Klingenberg R.
        • Lebens M.
        • Hermansson A.
        • et al.
        Intranasal immunization with an apolipoprotein B-100 fusion protein induces antigen-specific regulatory T cells and reduces atherosclerosis.
        Arterioscler Thromb Vasc Biol. 2010; 30: 946-952
        • Fredrikson G.N.
        • Schiopu A.
        • Berglund G.
        • Alm R.
        • Shah P.K.
        • Nilsson J.
        Autoantibody against the amino acid sequence 661-680 in apo B-100 is associated with decreased carotid stenosis and cardiovascular events.
        Atherosclerosis. 2007; 194: e188-192
        • Fredrikson G.N.
        • Hedblad B.
        • Berglund G.
        • et al.
        Association between IgM against an aldehyde-modified peptide in apolipoprotein B-100 and progression of carotid disease.
        Stroke. 2007; 38: 1495-1500
        • Sjogren P.
        • Fredrikson G.N.
        • Samnegard A.
        • et al.
        High plasma concentrations of autoantibodies against native peptide 210 of apoB-100 are related to less coronary atherosclerosis and lower risk of myocardial infarction.
        Eur Heart J. 2008; 29: 2218-2226
        • Fredrikson G.N.
        • Anand D.V.
        • Hopkins D.
        • et al.
        Associations between autoantibodies against apolipoprotein B-100 peptides and vascular complications in patients with type 2 diabetes.
        Diabetologia. 2009; 52: 1426-1433
        • Baldassarre D.
        • Nyyssonen K.
        • Rauramaa R.
        • et al.
        Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study.
        Eur Heart J. 2010; 31: 614-622
        • Sans S.
        • Kesteloot H.
        • Kromhout D.
        The burden of cardiovascular diseases mortality in Europe. Task force of the European Society of Cardiology on Cardiovascular Mortality and Morbidity Statistics in Europe.
        Eur Heart J. 1997; 18: 1231-1248
        • Beux F.
        • Carmassi S.
        • Salvetti M.V.
        • et al.
        Automatic evaluation of arterial diameter variation from vascular echographic images.
        Ultrasound Med Biol. 2001; 27: 1621-1629
        • Renauda O.
        • Victoria-Feserb M.-P.
        A robust coefficient of determination for regression.
        J Stat Plan Inference. 2010; 140: 1852-1862
        • Tibshirani R.
        • Friedman J.H.
        The elements of statistical learning: data mining, inference, and prediction. Springer series in statistics.
        2nd ed. Springer, New York, NY2009
        • Huber P.J.
        Robust estimation of a location parameter.
        Ann Math Stat. 1964; 35: 73-101
        • Purcell S.
        • Neale B.
        • Todd-Brown K.
        • et al.
        PLINK: a tool set for whole-genome association and population-based linkage analyses.
        Am J Hum Genet. 2007; 81: 559-575