International Journal of Cardiology 168 (2013) 484–489

Contents lists available at ScienceDirect

International Journal of Cardiology

j ourna l homepage: www.e lsev ie r .com/ locate / i j ca rd

Antibodies to malondialdehyde oxidized low-density lipoproteins predict long termcardiovascular mortality in high risk patients

Giuseppe Maiolino b,1, Luigi Pedon a,1, Maurizio Cesari b, Anna Chiara Frigo c, Marlena Barisa b,Giacomo Rossitto b, Teresa Maria Seccia b, Mario Zanchetta a, Gian Paolo Rossi b,⁎a Divisione di Cardiologia Ospedale di Cittadella, Italyb Dept. of Medicine-Internal Medicine University of Padua, Italyc Department of Cardiac, Thoracic and Vascular Sciences, University of Padua, Italy

Abbreviations: BP, blood pressure; CAD, coronary arteryprotein; LDL, low density lipoprotein; MI, myocardial infaOxLDL, oxidized low density lipoprotein; OxLDLAbs, antiblipoprotein.⁎ Corresponding author at: DMCS-ClinicaMedica 4, Uni

2, 35126 Padova, Italy. Tel.: +39 0 49 821 3304x2279; faE-mail address: gianpaolo.rossi@unipd.it (G.P. Rossi)

1 These authors have equally contributed to this man

0167-5273/$ – see front matter © 2012 Elsevier Irelandhttp://dx.doi.org/10.1016/j.ijcard.2012.09.165

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 3 May 2012Received in revised form 3 September 2012Accepted 25 September 2012Available online 25 October 2012

Keywords:Coronary artery diseaseAnti-LDL-oxidized antibodiesPrognosisAtherosclerosis

Aims: Antibodies to oxidized low-density lipoproteins (oxLDLAbs) are detectable in the serum of patientswith and without atherosclerosis, but it is unclear if they play a pathogenic or a protective role in atherogen-esis or if they are simply a marker of atherosclerosis. Therefore, in a prospective cohort study we investigatedif oxLDLAbs titer predicts cardiovascular (CV) events in high-risk coronary artery disease patients.Methods and results: The titer of IgG antibodies to malondialdehyde modified oxidized low-density lipopro-teins was measured in 748 randomly selected patients of the GENICA study who underwent coronary angi-ography and assessment of incident CV events at follow-up. Patients were classified by oxLDLAbs into a lowand a high titer group, corresponding to the first three and the last quartile, respectively. Cardiovascularevent-free survival was compared between oxLDLAbs groups by Kaplan–Meier and multivariate techniqueincluding propensity score matching analysis. During long-term follow-up (median 7.2 years) CV deaths

were observed in 65 patients (11.6%), more commonly in the high than in the low oxLDLAbs group (patientsfree from CV death 83.1% vs. 89% respectively, p=0.025). The incidence of CV events was also higher in theformer than in latter (event-free survival 69.2% vs. 77.7% respectively, p=0.030).Conclusions: An oxLDLAbs titer above the 75th percentile is a marker of LDL oxidation which predicts a worseCV prognosis at long term follow-up in high-risk Caucasian patients referred for coronary angiography.

© 2012 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Accumulation of low-density lipoproteins (LDLs) in the sub-endothelial extracellular space promotes recruitment of monocytesand their differentiation into macrophages. The latter produce reac-tive nitrogen and oxygen species (ROS) that induce oxidation ofLDLs thus generating oxidized LDLs (oxLDLs). OxLDLs are cytotoxicto endothelial cells, chemotactic for monocytes, and mitogenic formacrophages and smooth muscle cells [1–3]. Moreover, their bindingto the scavenger receptors is followed by internalization with ensuingfoam cell formation. Hence, they have multiple effects that are key foratherogenesis [4].

Oxidation of LDLs induces immunogenic epitopes in their molecule[5] with ensuing generation of antibodies against oxLDLs (oxLDLAbs)

disease; HDL, high-density lipo-rction; MDA, malondialdehyde;odies anti-oxidized low density

versity Hospital, via Giustiniani,x: +39 49 880 2252..uscript.

Ltd. All rights reserved.

[6] that are detectable in the sera of the majority of patients with ad-vanced atherosclerotic lesions [7]. These autoantibodies can thereforebe viewed as an in vivo marker of LDL oxidation, but their biologicalrole remains puzzling [6] since they were found also in children withoutatherosclerosis [8]. It has therefore been contended that oxLDLAbs pro-tect from atherosclerosis by neutralizing the aforementioned pathogenicactions of oxLDL [9,10]. Along this line a reduced atherosclerosis that cor-related with the levels of immunoglobulin G (IgG) oxLDLAbs was foundin rodents immunized with oxLDL [9–12]. Moreover, a monoclonal IgManti-oxLDL antibody (EO6) derived from apoE-deficient mouse wasshown to block the oxLDL binding to the macrophage scavenger recep-tors CD36 and SR-B1. Thus, oxLDLAbs and the innate immune responseto oxLDL may protect from atherosclerosis via the macrophages [13].

Notwithstanding all these premises and the wealth of experimen-tal data, longitudinal studies on the prognostic value of oxLDLAbshave given controversial results [6]: about half of the studies showedan association of oxLDLAbs with cardiovascular (CV) events, but therest showed no association. Only one study in end-stage renal diseasepatients, e.g. highly selected populations, suggested that a lowoxLDLAbs titer could predict CVmortality [14], but results were oppo-site in another study [15]. Moreover, precautions to avoid theconfounding effect of an unbalanced distribution of risk factors across

485G. Maiolino et al. / International Journal of Cardiology 168 (2013) 484–489

the oxLDLAbs groups were adopted in no studies thus far. Therefore,in the prospective branch of the GENICA (Genetic and Environmentalfactors In Coronary Artery disease) study we sought to test the hy-pothesis that oxLDLAbs predict CV events using a statistical analysisthat minimizes the impact of potential confounders.

2. Methods

2.1. Study participants

The protocol of the GENICA study will be briefly recalled as it was previously detailed[16,17]. The study enrolled consecutive Caucasian patients referred for coronary angiographyto investigate chest pain and/or suspected CAD between 1999 and 2001. All signed a consentform to participate in this study and the Medical Ethics Committee approved the protocol.Refusal to participate and previous inclusion in the study were the only exclusion criteria.Among the population potentially eligible (no. 1273) only two patients refused participationto the study. Information onmedical history, smoking habits, presence/absence of arterial hy-pertension, diabetes mellitus, dyslipidemia, and current medications was gathered with astaff administered questionnaire [17]. Definitions for bodymass index (BMI), smoking status,diabetes mellitus, impaired glucose tolerance, hypercholesterolemia, and hypertriglyc-eridemiawere already reported [16,17]. Bloodpressurewasmeasuredbymercury sphygmo-manometer using Korotkoff phase V for diastolic, according to the WHO guidelines.Hypertension was defined as systolic blood pressure≥140 mm Hg and/or diastolic bloodpressure≥90 mm Hg and/or use of antihypertensive drugs.

2.2. Coronary angiography

Angiography and measurement of left ventricular ejection fraction (LVEF) and thegrading of the CAD burden with the modified Duke Prognostic Index score were carriedout as described [18]. This score considers major epicardial coronary arteries with≥50% diameter stenosis and goes from 0 (all major coronary arteries with lesionsb50% diameter stenosis) to 100 (≥95% left main stenosis). It was reported to accurate-ly predict five-year mortality of medically treated patients [19].

2.3. Laboratory measurements

Patients were studied between 8.30 a.m. and noon. Blood samples were taken immedi-ately before coronary angiography, put on ice, and centrifuged at 3000×g (at 4 °C for10 min). Total cholesterol, HDL-cholesterol, triglycerides, glycemia, sodium, potassium,blood urea nitrogen, and creatinine levels were measured with conventional methods. IgGautoantibodies againstmalondialdehyde (MDA)-modified LDL (IgGoxLDLAbs)were assayedby ELISAwith a commercially available kit (Anti-oxLDLAntibody ELISAKIT; IMMCODiagnos-tic Inc.), according to manufacturer's specifications. The intra-assay and inter-assay coeffi-cients of variation of this method were 9% and 15%, respectively, in our hands [16].

Fig. 1. Data analysis flow chart. Flow chart showing the selection pro

2.4. Follow-up data

Information on the long-term outcome of the patients was gathered blindly totheir biochemical profile with a predefined form through review of medical chartsfor the patients regularly seen at referring hospitals, and through telephone interviewsof family doctors, and/or patients, and/or first-degree relatives for those not attendingregular follow-up visits. Predetermined primary endpoints were CV events, includingacute coronary syndromes, stroke, and CV deaths. The latter comprised sudden deathor due to congestive heart failure, acute coronary syndromes, or stroke according tothe Syst-Eur Trial criteria [20]. All events were validated by the adjudication committee(GPR, MZ, and GM) blinded to patients' biochemical profile.

2.5. Statistical analysis

Serum triglycerides, HDL- and LDL-cholesterol, age, creatinine, CADDuke Index score,LVEF, and oxLDLAbs were examined after log or square root transformation to achieve aGaussian distribution. A random sample comprising 748 of the CAD patients originallyrecruited in the GENICA study was obtained with single random number generationthrough SPSS. Patients in this sample were selected for the oxLDLAbs titer measurement,whichwas performed blindlywith respect to clinical and anthropometric data (see Fig. 1).

Standardized z scores were calculated to identify univariate outliers and exclusionof cases with z scores exceeding ∣3.29∣ that corresponded to a pb0.001, was decided apriori. Mahalanobis distance was assessed by regression analysis to identify multivar-iate outliers; cases with χ2 in excess of 32.909 (12 df at α=0.001) were consideredoutliers and removed from further analysis (14 patients) according to the techniqueof Tabachnick and Fidell [21].

Comparison of quantitative variables across groups was done by ANOVA followed byBonferroni's post hoc test. Chi-square analysiswas used to compare the frequencies of cat-egorical CAD risk factors.

Standard multiple regression analysis was used beforehand to verify the as-sumption that cases lost at follow-up did not differ significantly from thoseavailable for survival analysis. Propensity score was calculated with logistic re-gression analysis including all available variables (including gender, age, BMI,LDL- and HDL-cholesterol, triglycerides, serum creatinine, homocysteine, glyce-mia, serum sodium concentration, heart rate, arterial hypertension, smokinghabit, LVEF, the Duke Prognostic Index of coronary atherosclerotic burden, lengthof follow-up, history, and treatment variables) that are known to potentially affectthe outcomes. To correct for the imbalanced distribution of variables between thepatients with low and high OxLDLAbs we did a greedy matching without replace-ment using a caliper of 0.2 standard deviations of the logit of the propensityscore [22].

The distribution of measured baseline covariates was then compared betweenlow and high OxLDLAbs groups in the matched samples, assessing the balance inmeasured variables with standardized differences (see Supplemental table) [23].We plotted the Kaplan–Meier curve and compared the event-free survival for thematched set with the test proposed by Klein andMoeschberger [24]. Statistical signif-icance was defined as pb0.05. SPSS 18 for Windows (SPSS Italy Inc., Bologna, Italy)was used for all analyses.

cess by which the patients were submitted to statistical analysis.

486 G. Maiolino et al. / International Journal of Cardiology 168 (2013) 484–489

3. Results

3.1. Clinical characteristics

After exclusion of univariate and multivariate outliers follow-updata were available in 570 patients (78%) of those randomly selectedfor measurement of the oxLDLAbs titer. Eleven further cases of periop-erative death at coronary artery by-pass surgery (CABG) were excludedfrom the analysis as death was judged being related to CABG-surgery orits complications in these patients. Therefore, the final analysis was car-ried out in 559 patients. The anthropometric and clinical features ofthese patients divided according to the oxLDLAbs quartiles are shownin Table 1. Overall there were no significant differences across quartileswith the obvious exception of the oxLDLAbs titer.

In this cohort of patients the coronary angiography findings didnot differ from those of the whole population of the GENICA study[16–18] and we could fully confirm the lack of association betweenthe CAD burden and the oxLDLAbs titer at baseline (Table 1), as de-scribed in a smaller cohort [16].

3.2. Follow-up data

The median follow-up length was 7.2 years (range 1–12.7 years).Of the 559 patients 65 (11.6%) had a CV death and 153 (27.4%) a CVevent. The CV death and events according to quartiles of oxLDLAbstiter are shown in Fig. 2. The patients in the highest quartile ofoxLDLAbs titer had a lower CV death-free survival (82.6% vs. 90.5%,respectively, p=0.007 at Kaplan–Meyer analysis) and CV event-freesurvival (66.4% vs. 74.3%, respectively, p=0.036 at Kaplan–Meyeranalysis) than those in the low oxLDLAbs group.

Further analysis was carried out in 544 patients who had data com-prehensive of all the variables necessary to compute the propensityscore. Among these patients 97% of the 140 patients in the highest quar-tile of oxLDLAbs could be matched with an equal number (n=136) ofpatients in the low oxLDLAbs group by propensity score. After thismatching 38 deaths were observed in the cohort, corresponding to adeath rate of 14.0%. Patients in the high oxLDLAbs group again showeda significantlyworse CVdeath-free survival (83.1% vs. 89.0%, respective-ly, p=0.025) and a significantlyworse CV event-free survival (69.2% vs.77.7%, respectively, p=0.030) than those in the low oxLDLAbs group(Fig. 3A and B).

Table 1Demographic and clinical characteristics of the subjects classified by oxLDLAbs quartiles. ReBP, blood pressure; HDL, high density lipoprotein; LDL, low density lipoprotein; EF, ejection

Quartile of oxLD

Variable 1st(n=179)

2nd(n=188)

Age, (years) 62.3±10.6 63.1±9.5Gender, (M/F) 147(82%)/32(18%) 144(76%)/45(24Non-smokers/smokers/ex, (%) 36/12/52 39/13/48BMI, (kg/m2) 26.7±4.0 27.2±3.9Serum creatinine, (μmol/L) 92±26 95±29Serum K+, (mmol/L) 4.2±0.4 4.3±0.4Serum Na+, (mmol/L) 140±3 140±3Heart rate, (b/min) 65±10 67±11Systolic BP, (mm Hg) 133±18 135±19Diastolic BP, (mm Hg) 78±10 78±9Serum glycemia, (mg/dL) 6.4±1.8 6.4±2.3Total cholesterol, (mg/dL) 207±40 207±42HDL-cholesterol, (mg/dL) 47±11 46±12LDL cholesterol, (mg/dL) 133±32 132±33Triglycerides, (mg/dL) 144±84 148±93Homocysteine, (μmol/L) 11.9±7.9 11.6±6.8Left ventricular EF, (%) 62±14 62±14OxLDLAbs, (EU/mL) 1.8±0.7 4.4±1.0Duke CAD score 37±21 36±21Follow-up, (years) 7.4±2.3 7.0±2.5

4. Discussion

This study involved patients at clearly high-risk, as confirmed by theobservation of a high rate of CV events and an 11.6% death rate at along-term follow-up. Exhaustive information on many potentially rele-vant confounders including the coronary atherosclerotic burden atbaseline (by use of a modified Duke Index score to take into accountalso the left main lesions) was collected. With the power conferred bythese features and a careful matching for potential confounders, wecould address the question of whether oxLDLAbs titer carries any prog-nostic information in high-risk coronary artery disease patients. Wemeasured oxLDLAbs with a commercially available kit in order toallow for both independent replication and clinical use of the studyresults. Our data are novel in that they demonstrate for the first timethat in a cohort of high-risk Caucasian patients referred for coronary an-giography a titer of oxLDLAbs exceeding the 75th percentile of thevalues distribution is associated with a worse CV prognosis, as com-pared to the patients with lower titers. This negative prognostic effectwas independent of major CV risk factors, as shown by adjustment forpotential confounders in our analysis.

The strength of this conclusion is supported by the prospectivedesign of our study and the application of a statistical analysis basedon the propensity score matching, which allows the balancing of thegroups based on the known baseline determinants of CV events atfollow-up. Thus, the observation that the titer of oxLDLAbs predictsCV event- and CV death-free survival in a high-risk CAD population,independently of other established CV risk factors, is a major hardpiece of evidence.

After the initial contention by Witzum et al. that oxLDL can play arole in the initiation and progression of the atherosclerotic process[25], data on the association of oxLDLAbs with atherosclerosis or CVevents have been highly conflicting (for rev. [6]). Some cross-sectional studies suggested a direct association of oxLDLAbs with ath-erosclerosis in different vascular beds [7,26–28], but others reportedeither opposite results on the association of oxLDLAbs with stroke[29], restenosis [30], and cardiovascular events in diabetics [31,32],or no association with coronary atherosclerotic burden in CAD pa-tients [16].

The longitudinal studies notoriously are held to be less prone to selec-tion bias and serendipitous findings. To date they suggested a prognosticrole of a high oxLDLAbs titer for the progression of atherosclerosis

sults are expressed as mean±SD. BMI, body mass index; K+, potassium; Na+, sodium;fraction; oxLDLAbs, antibodies against oxidized LDL; CAD, coronary artery disease.

LAbs titer

3rd(n=181)

4th(n=185)

p=

64.4±10.3 63.2±10.0 NS%) 136(75%)/45(25%) 148(80%)/37(20%) NS

38/19/43 34/13/53 NS26.9±3.8 26.8±3.4 NS91±26 96±27 NS4.2±0.4 4.2±0.4 NS140±2 139±3 NS66±10 66±10 NS134±18 133±17 NS78±10 77±9 NS5.9±1.7 6.3±2.2 NS209±43 205±49 NS45±11 45±9 NS134±34 131±34 NS142±67 146±126 NS11.5±5.2 12.8±6.9 NS61±15 61±14 NS8.7±1.3 23.3±18.3 b0.00136±21 38±19 NS7.1±2.5 6.8±2.6 NS

Fig. 2. Cardiovascular events. Cardiovascular death and events rate by quartiles of oxLDLAbs titer (the absolute number of events is shown above each column). AMI: acute myo-cardial infarction.

487G. Maiolino et al. / International Journal of Cardiology 168 (2013) 484–489

[33–37], but for mortality only in patients with end stage renal disease[15]. The latter finding, however, was not confirmed in another longitudi-nal study of end stage kidney disease patients that even reported anincreased survival in those with a high oxLDLAbs titer [14]. Otherstudies reported no prognostic role for oxLDLAbs titer on CV events[29,32,33,38–42].

These controversial results might derive from the diversity of humanoxLDLAbs [43,44], which entail different classes and subclasses, includingIgA, IgG1, IgG2, IgG3, and IgM, and to specific idiotypes [45]. It could alsobe that this heterogeneity translates into opposing biological actions: forexample, oxLDL contains phospholipids and IgG2 is the dominant sub-class of auto antibodies made in response to epitopes containing phos-pholipids, while the IgG3 subclass have more pathogenic propertiesthan the IgG2 subclass because it fixes complement better and binds Fcreceptors more avidly [46]. These differences among anti-oxLDL anti-bodies and their correlation with clinical manifestations have not beenclarified thus far and therefore there are no robust answers to the ques-tion on whether different oxLDLAbs have different effects on CV events.

Fig. 3. Cardiovascular death- (Panel A) and event-free survival (Panel B). Kaplan–Meier cur(Panel B) in the propensity score-matched high-risk patients divided into the high anddeath-free survival and a significantly lower CV event-free survival.

The available conflicting data could also be attributed to differ-ences in study design, insufficient power, and lack of taking into prop-er consideration the potentially confounding effect of an unevendistribution of CV risk factors between patients with different levelsof oxLDLAbs titer.

The predictive role of the oxLDLAbs on CV death evidenced in thisstudy might seem odd given the previously reported lack of associationof these antibodies with angiographically assessed CAD [16]. It should,however, be pointed out that triggering of MI and athero-thromboticstroke is usually due to rupture of vulnerable atherosclerotic plaques,which is enhanced by several factors involvingmatrixmetalloproteinaseactivation [47]. Hence, even though the determinants of plaque destabi-lization causing CV events and those of plaque growth, e.g. of the athero-sclerotic burden, can overlap, they differ significantly, as previouslynoticed [47]. Indeed, it is worth remembering that the culprit vulnerableplaque causing acute coronary syndromes commonly occurs at sites de-termining stenoses b50% [48,49]. Thus, the present finding thatoxLDLAbs titer did predict fatal CV events while they did not correlate

ves showing cardiovascular (CV) death-free survival (Panel A) and event-free survivallow oxLDLAbs titer group. Patients with low oxLDLAbs had a significantly lower CV

488 G. Maiolino et al. / International Journal of Cardiology 168 (2013) 484–489

with the CAD atherosclerotic burden is just a further example of the dis-sociation of the determinants of stable and unstable CAD.

4.1. Limitations and strengths of the study

Some caveats concerning the present conclusions must beunderscored: given the aforementioned heterogeneity of oxLDLAbsour results apply only to IgG antibodies against MDA-modified oxLDLand not to other classes or idiotypes of oxLDLAbs. In addition, they per-tain to a high-risk population of Caucasian patients referred for coro-nary artery angiography and therefore, it remains to be verified inother cohorts of different ethnicity and with a lower risk profile.

It might also be argued that 22% of the patients lost at follow-up,representing a non-trivial part of our cohort, could have biased ourresults. Considering, however, the length of our follow-up this dropoutrate seems a reasonable one; moreover, the cases lost at follow-up didnot differ significantly from those available at follow-up at baseline,thus making a selection bias unlikely. Finally, this study was notaimed at investigating the cellular or molecular mechanisms by whichoxLDLAbs can affect outcome, which remain a goal for future studies.

Strengths of the study include, besides those already mentionedabove, the long-term follow-up, the comprehensive information onanthropometric, biochemical, and clinical features at baseline and apainstaking assessment of CV events at follow-up. Moreover, itmight seem that the study size was small; however, matching asmany as 136 patients with the propensity score for all the relevantvariables that were used to its calculation is, indeed, an accomplish-ment that must be appreciated. The fact that after this matching theresults of the univariate Kaplan–Meier analysis were fully confirmedprovides therefore strong new evidence in an area that has been high-ly controversial thus far.

5. Conclusions

By using for the first time a propensity score matching analysis thisstudy shows that the titer of antibodies against MDA modified LDLpredicts CV death and CV events independent of the other commonmajor CV risk factors and of the CAD atherosclerotic burden inhigh-risk Caucasian patients with CAD. Therefore, these results shednovel light in a highly controversial field. Nonetheless, further researchshould be aimed at clarifying if this finding applies also to other types ofoxLDLAbs, as these antibodies are highly heterogeneous in nature [45].Investigation of themechanisms bywhich oxLDLAbs impact on athero-sclerotic plaque erosion and rupture is worthy, because it can open theway to novel preventive and therapeutic approaches for the first causeof death in westernized countries [50].

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.ijcard.2012.09.165.

Acknowledgments

This study was supported by grants of the Italian Cabinet of Univer-sity and Scientific Research (MURST) to ACP (9906193152_001/06) andby research grants from Regione Veneto (863/01/98), Unindustria ofTreviso and F.O.R.I.C.A. to Prof. G.P. Rossi.

The authors of this manuscript have certified that they complywith the Principles of Ethical Publishing in the International Journalof Cardiology.

References

[1] Hessler JR, Morel DW, Lewis LJ, Chisolm GM. Lipoprotein oxidation andlipoprotein-induced cytotoxicity. Arteriosclerosis 1983;3:215-22.

[2] Chatterjee S, Ghosh N. Oxidized low density lipoprotein stimulates aortic smoothmuscle cell proliferation. Glycobiology 1996;6:303-11.

[3] Yui S, Sasaki T, Miyazaki A, Horiuchi S, Yamazaki M. Induction of murine macro-phage growth by modified LDLs. Arterioscler Thromb 1993;13:331-7.

[4] Steinberg D, Witztum JL. Oxidized low-density lipoprotein and atherosclerosis.Arterioscler Thromb Vasc Biol 2010;30:2311-6.

[5] Erkkila AT, Narvanen O, Lehto S, Uusitupa MI, Yla-Herttuala S. Autoantibodiesagainst oxidized low-density lipoprotein and cardiolipin in patients with coro-nary heart disease. Arterioscler Thromb Vasc Biol 2000;20:204-9.

[6] Strobel NA, Fassett RG, Marsh SA, Coombes JS. Oxidative stress biomarkers as pre-dictors of cardiovascular disease. Int J Cardiol 2011;147:191-201.

[7] Inoue T, Uchida T, Kamishirado H, Takayanagi K, Hayashi T, Morooka S. Clinicalsignificance of antibody against oxidized low density lipoprotein in patientswith atherosclerotic coronary artery disease. J Am Coll Cardiol 2001;37:775-9.

[8] Shoenfeld Y, Sherer Y, George J, Harats D. Autoantibodies associated with athero-sclerosis. Ann Med 2000;32(Suppl. 1):37-40.

[9] Zhou X, Caligiuri G, Hamsten A, Lefvert AK, Hansson GK. LDL immunization in-duces T-cell-dependent antibody formation and protection against atherosclero-sis. Arterioscler Thromb Vasc Biol 2001;21:108-14.

[10] Caligiuri G, Nicoletti A, Poirier B, Hansson GK. Protective immunity against atheroscle-rosis carried by B cells of hypercholesterolemic mice. J Clin Invest 2002;109:745-53.

[11] George J, Afek A, Gilburd B, et al. Hyperimmunization of apo-E-deficient micewith homologous malondialdehyde low-density lipoprotein suppresses early ath-erogenesis. Atherosclerosis 1998;138:147-52.

[12] Shoenfeld Y, Harats D, George J. Heat shock protein 60/65, beta 2-glycoprotein Iand oxidized LDL as players in murine atherosclerosis. J Autoimmun 2000;15:199-202.

[13] Binder CJ, Chang MK, Shaw PX, et al. Innate and acquired immunity in atherogen-esis. Nat Med 2002;8:1218-26.

[14] Shoji T, Fukumoto M, Kimoto E, et al. Antibody to oxidized low-density lipoproteinand cardiovascular mortality in end-stage renal disease. Kidney Int 2002;62:2230-7.

[15] Bayes B, Pastor MC, Bonal J, Foraster A, Romero R. Oxidative stress, inflammationand cardiovascular mortality in haemodialysis — role of seniority and intravenousferrotherapy: analysis at 4 years of follow-up. Nephrol Dial Transplant 2006;21:984-90.

[16] Rossi GP, Cesari M, De Toni R, et al. Antibodies to oxidized low-density lipopro-teins and angiographically assessed coronary artery disease in white patients.Circulation 2003;108:2467-72.

[17] Rossi GP, Cesari M, Zanchetta M, et al. The T-786C endothelial nitric oxidesynthase genotype is a novel risk factor for coronary artery disease in Caucasianpatients of the GENICA study. J Am Coll Cardiol 2003;41:930-7.

[18] Cesari M, Zanchetta M, Burlina A, et al. Hyperhomocysteinemia is inversely relatedwith left ventricular ejection fraction and predicts cardiovascular mortality inhigh-risk coronary artery disease hypertensives. Arterioscler Thromb Vasc Biol2005;25:115-21.

[19] Califf RM, Armstrong PW, Carver JR, D'Agostino RB, Strauss WE. 27th BethesdaConference: matching the intensity of risk factor management with the hazardfor coronary disease events. Task Force 5. Stratification of patients into high, me-dium and low risk subgroups for purposes of risk factor management. J Am CollCardiol 1996;27:1007-19.

[20] Staessen JA, Fagard R, Thijs L, et al. Randomised double-blind comparison of pla-cebo and active treatment for older patients with isolated systolic hypertension.The Systolic Hypertension in Europe (Syst-Eur) Trial Investigators. Lancet1997;350:757-64.

[21] Tabachnick BG, Fidell LS. Using multivariate statistics; 2001. p. 56–110.[22] Austin PC. Some methods of propensity-score matching had superior perfor-

mance to others: results of an empirical investigation and Monte Carlo simula-tions. Biom J 2009;51:171-84.

[23] Heinze G, Juni P. An overview of the objectives of and the approaches to propen-sity score analyses. Eur Heart J 2011;32:1704-8.

[24] Klein JP, Moeschberger ML. Survival analysis: techniques for censored and trun-cated data. New York: Springer Verlag; 1997.

[25] Witztum JL, Steinberg D. Role of oxidized low density lipoprotein in atherogene-sis. J Clin Invest 1991;88:1785-92.

[26] Inoue T, Yaguchi I, Uchida T, et al. Clinical significance of the antibody against oxi-dized low-density lipoprotein in acute myocardial infarction. Cardiology 2002;98:13-7.

[27] Nishi K, Itabe H, Uno M, et al. Oxidized LDL in carotid plaques and plasma associ-ates with plaque instability. Arterioscler Thromb Vasc Biol 2002;22:1649-54.

[28] Tsimikas S, Bergmark C, Beyer RW, et al. Temporal increases in plasma markers ofoxidized low-density lipoprotein strongly reflect the presence of acute coronarysyndromes. J Am Coll Cardiol 2003;41:360-70.

[29] Ahmed E, Trifunovic J, Stegmayr B, Hallmans G, Lefvert AK. Autoantibodies againstoxidatively modified LDL do not constitute a risk factor for stroke: a nested case–control study. Stroke 1999;30:2541-6.

[30] Boullier A, Walters-Laporte E, Hamon M, et al. Absence of relationship betweenplasma Lp(a), Lp-AI, anti-oxidized LDL autoantibodies, LDL immune complexesconcentrations and restenosis after percutaneous transluminal coronary angio-plasty. Clin Chim Acta 2000;299:129-40.

[31] Lopes-Virella MF, Virella G, Orchard TJ, et al. Antibodies to oxidized LDL andLDL-containing immune complexes as risk factors for coronary artery disease indiabetes mellitus. Clin Immunol 1999;90:165-72.

[32] Uusitupa MI, Niskanen L, Luoma J, et al. Autoantibodies against oxidized LDL donot predict atherosclerotic vascular disease in non-insulin-dependent diabetesmellitus. Arterioscler Thromb Vasc Biol 1996;16:1236-42.

[33] Salonen JT, Yla-Herttuala S, Yamamoto R, et al. Autoantibody against oxidised LDLand progression of carotid atherosclerosis. Lancet 1992;339:883-7.

[34] Puurunen M, Manttari M, Manninen V, et al. Antibody against oxidizedlow-density lipoprotein predicting myocardial infarction. Arch Intern Med1994;154:2605-9.

489G. Maiolino et al. / International Journal of Cardiology 168 (2013) 484–489

[35] Orchard TJ, Virella G, Forrest KY, Evans RW, Becker DJ, Lopes-Virella MF. Anti-bodies to oxidized LDL predict coronary artery disease in type 1 diabetes: a nestedcase–control study from the Pittsburgh Epidemiology of Diabetes ComplicationsStudy. Diabetes 1999;48:1454-8.

[36] Doo YC, Han SJ, Lee JH, et al. Associations among oxidized low-density lipoproteinantibody, C-reactive protein, interleukin-6, and circulating cell adhesion mole-cules in patients with unstable angina pectoris. Am J Cardiol 2004;93:554-8.

[37] Wu R, Nityanand S, Berglund L, Lithell H, Holm G, Lefvert AK. Antibodies againstcardiolipin and oxidatively modified LDL in 50-year-old men predict myocardialinfarction. Arterioscler Thromb Vasc Biol 1997;17:3159-63.

[38] Erkkila AT, Narvanen O, Lehto S, Uusitupa MI, Yla-Herttuala S. Antibodies againstoxidized LDL and cardiolipin and mortality in patients with coronary heart dis-ease. Atherosclerosis 2005;183:157-62.

[39] Wilson PW, Ben-Yehuda O, McNamara J, Massaro J, Witztum J, Reaven PD. Auto-antibodies to oxidized LDL and cardiovascular risk: the Framingham OffspringStudy. Atherosclerosis 2006;189:364-8.

[40] Mayr M, Kiechl S, Tsimikas S, et al. Oxidized low-density lipoprotein autoanti-bodies, chronic infections, and carotid atherosclerosis in a population-basedstudy. J Am Coll Cardiol 2006;47:2436-43.

[41] Tsimikas S, Brilakis ES, Lennon RJ, et al. Relationship of IgG and IgM autoanti-bodies to oxidized low density lipoprotein with coronary artery disease andcardiovascular events. J Lipid Res 2007;48:425-33.

[42] Wu T, Willett WC, Rifai N, Shai I, Manson JE, Rimm EB. Is plasma oxidizedlow-density lipoprotein, measured with the widely used antibody 4E6, an inde-pendent predictor of coronary heart disease among U.S. men and women? J AmColl Cardiol 2006;48:973-9.

[43] Mironova M, Virella G, Lopes-Virella MF. Isolation and characterization of humanantioxidized LDL autoantibodies. Arterioscler Thromb Vasc Biol 1996;16:222-9.

[44] Wu R, Lefvert AK. Autoantibodies against oxidized low density lipoproteins(oxLDL): characterization of antibody isotype, subclass, affinity and effect on themacrophage uptake of oxLDL. Clin Exp Immunol 1995;102:174-80.

[45] Wu R, Shoenfeld Y, Sherer Y, et al. Anti-idiotypes to oxidized LDL antibodies in in-travenous immunoglobulin preparations — possible immunomodulation of ath-erosclerosis. Autoimmunity 2003;36:91-7.

[46] Henriksen T, Mahoney EM, Steinberg D. Enhanced macrophage degradation oflow density lipoprotein previously incubated with cultured endothelial cells: rec-ognition by receptors for acetylated low density lipoproteins. Proc Natl Acad SciU S A 1981;78:6499-503.

[47] Rossi GP, Maiolino G, Zanchetta M, et al. The T(-786)C endothelial nitric oxidesynthase genotype predicts cardiovascular mortality in high-risk patients. J AmColl Cardiol 2006;48:1166-74.

[48] Little WC, Constantinescu M, Applegate RJ, et al. Can coronary angiography predictthe site of a subsequent myocardial infarction in patients with mild-to-moderatecoronary artery disease? Circulation 1988;78:1157-66.

[49] Ambrose JA, Tannenbaum MA, Alexopoulos D, et al. Angiographic progression ofcoronary artery disease and the development of myocardial infarction. J Am CollCardiol 1988;12:56-62.

[50] Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke statistics — 2011update: a report from the American Heart Association. Circulation 2011;123:e18-209.