Atherosclerosis 193 (2007) 196203

Inammatory and anti-inammatory variable clusters and risk prediction in acute coronary syndrome patients: A factor analysis approach
Dimitrios N. Tziakas a, , Georgios K. Chalikias a , Juan Carlos Kaski b , Angelos Kekes a , Eleni I. Hatzinikolaou a , Dimitrios A. Stakos a , Ioannis K. Tentes c , Alexandros X. Kortsaris c , Dimitrios I. Hatseras a
b

University Cardiology Department, Democritus University of Thrace, Voulgaroktonou 23, 68100 Alexandroupolis, Greece Cardiovascular Biology Research Centre, Division of Cardiac and Vascular Sciences, St. Georges, University of London, London, UK c Department of Biochemistry, Democritus University of Thrace, Alexandroupolis, Greece Received 1 March 2006; received in revised form 4 June 2006; accepted 14 June 2006 Available online 20 July 2006

Abstract Background: Numerous inammatory mediators such as C-reactive protein (CRP), brinogen, interleukin-18 (IL-18), and inter-cellular adhesion molecule-1 (ICAM-1) have been proposed for risk stratication in acute coronary syndrome (ACS) patients. However, interactions between these markers have made it difcult to assess their true role in risk prediction. Factor analysis is a multivariable statistical technique that reduces a large number of intercorrelated variables to a smaller set of independent clusters, underlining physiological relationships. The aim of this study was to investigate, using factor analysis, a clustering of pro-inammatory markers, anti-inammatory cytokines such as interleukin-10 (IL-10) and HDL cholesterol, and to determine their role in prediction of risk of recurrent coronary events in ACS patients. Methods: We assessed 320 consecutive patients (236 men; 67 years; IQ 5874 years) admitted with ACS. The composite of cardiac death and re-hospitalization with non-fatal myocardial infarction, or unstable angina, was the pre-specied study end-point. Serum CRP, brinogen, HDL cholesterol, IL-10, IL-18 and ICAM-1 levels were measured at study entry. We assessed independent predictors of the combined end-point during a 1-year follow-up using multiple logistic regression analysis. Results: Factor analysis identied three clusters which were arbitrarily interpreted as (1) a systemic inammation cluster with positive loadings of CRP and brinogen, (2) a local inammationendothelial dysfunction cluster with positive loadings of IL-18 and ICAM-1 and (3) an anti-inammation cluster comprising IL-10 and HDL cholesterol. Only the anti-inammation cluster was a signicant predictor (OR 0.66, 95% CI: 0.490.89) of adverse cardiac events during a 1-year follow-up and remained signicant (OR 0.65, 95% CI: 0.480.88) in a multivariate model that included all three factors. Conclusions: Although inammatory markers such as CRP predict future cardiovascular events in ACS patients, when all inammatory mediators are taken into account in a prospective analysis of risk, markers reecting anti-inammatory mechanisms are better prognostic markers. 2006 Elsevier Ireland Ltd. All rights reserved.
Keywords: Factor analysis; ACS prognosis; Inammation

1. Introduction Acute coronary syndromes (ACS) such as unstable angina and myocardial infarction are the clinical manifestations of

Corresponding author. Tel.: +30 25510 35596; fax: +30 25510 35596. E-mail address: dtziakas@med.duth.gr (D.N. Tziakas).

destablization of coronary atherosclerotic plaques. There is increased recognition that inammation plays a key role in atherosclerosis and its complications [1,2]. Recently, attention has focused on the potential role of circulating markers of inammation as risk predictors among those who have suffered an acute cardiovascular event [3,4]. Acute phase reactants such as C-reactive protein (CRP)

0021-9150/$ see front matter 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.atherosclerosis.2006.06.016

D.N. Tziakas et al. / Atherosclerosis 193 (2007) 196203

197

and brinogen, are independent predictors of short- and long-term mortality in patients with symptomatic coronary artery disease [5]. Markers of endothelial activation i.e. intracellular adhesion molecule-1 (ICAM-1) have also been found to predict future cardiovascular events in prospective studies [6]. Elevated serum levels of inammatory mediators such as interleukin (IL)-6 [7], IL-18 [8], pregnancy associated protein-A [9], and neopterin [10] have been associated with poor prognosis in this patient group. However, the inammatory cascade has counterbalancing factors that maintain a delicate balance of pro- and anti-inammatory molecules that regulate vascular homeostasis [11] and maintain the integrity of the vessel wall [11,12]. In this respect, elevated serum levels of IL-10, a potent anti-inammatory cytokine, are known to be associated with a more favorable prognosis in patients with ACS in another study [13]. Although investigators have assessed the individual contribution of each of these molecules to cardiovascular risk, the role of clustering of inammatory and anti-inammatory markers in risk prediction has not been systematically explored. Such an assessment is essential because inammatory pathways, anti-inammatory mechanisms, and endothelial activation do not occur in isolation, and cardiovascular risk is likely to depend on the interaction among these variables. We therefore assessed in ACS patients clusters of proinammatory and anti-inammatory mediators, as well as markers of endothelial activation. Using exploratory factor analysis, a variable reduction approach, we assessed the performance of these clusters individually and their association as markers of future cardiovascular risk.

undergo elective non-cardiac surgical procedures during follow-up were excluded from the study. 2.2. Follow-up Patients were followed for up to 1-year after admission using a standardized protocol that included outpatient visits, telephone contacts and the recording of recurrent cardiac events. The primary clinical outcome of interest was the composite end-point of major adverse cardiovascular event such as (i) hospital admission with non-fatal MI dened according to European Cardiology Society criteria [14,15] (e.g. raised cardiac troponins and/or cardiac enzymes, characteristic ECG changes, and prolonged typical chest pain), (ii) hospital admission with Braunwalds class IIIb unstable angina requiring medical treatment and/or urgent revascularization i.e. percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) [16], and (iii) cardiac death, dened as sudden unexplained death, death from fatal myocardial infarction or death from possible myocardial ischemia. 2.3. Biochemical data In all patients peripheral blood samples were drawn within 1 h from admission. After centrifugation at 3000 rpm for 10 min, serum samples were frozen and stored at 70 C until biochemical assessment. Sandwich enzyme immunoassay was performed for measuring concentrations of serum IL-10 using Quantikine HS human IL-10, R&D Systems Inc., Minneapolis, USA [17] commercial kits with monoclonal antibodies, with a minimum detectable concentration <0.5 pg/ml. The intra-assay precision and inter-assay precision of the method were 8.5% and 15.6%, respectively. Sandwich enzyme immunoassay was performed for measuring concentrations of serum IL-18 using MBL Medical and Biological Laboratories Co. Ltd., Nagoya, Japan [18] commercial kits with monoclonal antibodies, with a minimum detectable concentration of 12.5 pg/ml. The intra-assay precision and inter-assay precision of the method were 5.61% and 10.1%, respectively. Sandwich enzyme immunoassay was performed by measuring concentrations of serum ICAM-1 using human ICAM-1 R&D Systems Inc., Minneapolis, USA [19] commercial kits with monoclonal antibodies, with a minimum detectable concentration of 3.3 ng/ml. The intra-assay precision and inter-assay precision of the method were 4.1% and 7.6%, respectively. High sensitivity C-reactive protein (CRP) was measured using Dade Behring, (Marburg, Germany) [20] commercial kits while brinogen was measured using DG-FIB, Grifols (Barcelona, Spain) commercial kits [21]. All other biochemistry measurements were carried out by our biochemistry department using standard methods.

2. Methods 2.1. Patients We assessed consecutive patients who were admitted to the Coronary Care Unit of our institution with a diagnosis of ACS. Serum samples were obtained from every patient at study entry for the assessment of inammatory markers. Myocardial infarction (MI) [ST segment elevation (STEMI) and non-ST segment elevation (NSTEMI)] and unstable angina (UA) were diagnosed using European Society of Cardiology criteria [14,15]. Patient management was carried out in accordance with institutional protocols and specic decisions regarding urgent revascularization were left to the discretion of the managing cardiologist. The study was approved by the Hospitals Ethics Committee and all patients gave written informed consent prior to study entry. We did not include patients with a history of hematological, neoplastic, renal, liver or thyroid disease, or patients receiving treatment with anti-inammatory drugs. Furthermore, patients with infectious or autoimmune diseases, familial hyperlipidaemia, or those scheduled to

198

D.N. Tziakas et al. / Atherosclerosis 193 (2007) 196203

2.4. Statistical analysis Values are expressed as median values with interquartile ranges. All statistical calculations were performed using SPSS 11.0 statistical software package (SPSS Inc.). A p value <0.05 was considered statistically signicant. Differences in continuous variables between groups were assessed using MannWhitney U test, whereas comparisons between categorical variables were performed by chi-square test or Fischers exact test when required. The KolmogorovSmirnof test was used to examine the normality of the variables distribution. The levels of IL-18, IL-10, HDL cholesterol, ICAM-1, CRP and brinogen were not normally distributed, therefore Spearmans correlation was used to estimate correlation coefcients between inammatory markers. 2.5. Factor analysis

sion, revascularization during follow-up, b-blocker, statin and clopidogrel use, as well as for presence of EF < 40%. The aforementioned variables were signicantly different between patients presenting with and without the study endpoint during the 1-year follow-up, therefore could have acted as cofounders (data not shown). For analysis inammatory marker levels were modelled in quartiles with odds ratios (ORs) referring to relative risk between the upper and lower quartile. The association of cluster scores with risk of developing adverse cardiac events during the 1-year follow-up was assessed using multivariate regression analysis with covariates including diagnosis on admission, revascularization during follow-up, b-blocker, statin and clopidogrel use, as well as for presence of EF < 40%. Cluster scores were modelled as continuous independent variables with odds ratios referring to risk of adverse events per 1 S.D. difference. 2.7. Power analysis

Factor analysis simplies multivariate data consisting of a large number of intercorrelated variables by grouping them into a smaller set of independent factors (clusters) according to basic underlying relationships among variables. Factors are fewer in number than the original variables; account for a signicant proportion of data variance; are useful in subsequent statistical analysis i.e. predictive regression models; and are often helpful in elucidating pathophysiologic mechanisms [22]. Briey, factor analysis was initiated by extraction of an initial solution using principal component analysis. Principal component analysis is a linear transformation of a set of partially correlated variables into an uncorrelated set of factors that are ordered according to progressively decreasing contribution to total variance in the data. In the present analysis screen plots were inspected and factors (clusters) with eigen values greater than 1 were retained [23]. This corresponds to retention of factors that contribute to variance greater than any original variable. Varimax rotation was then performed with identication of variables comprising a factor (cluster) based on loadings greater than 0.5 [23]. The threshold value of 0.5 has been commonly used [23,24]. For each cluster identied, a cluster score was derived. These scores represent the subjects predicted values for each factor (cluster) and are calculated using the factor weights and the original variable values. Of importance, naming and interpretation of the clusters are arbitrary. Although the normality assumption is not a pre-requisite for factor analysis, all of the under investigation variables used in the present study had not an extreme non-normal distribution (skewness <2 and kurtosis <7) [23]. 2.6. Regression analysis The association of each one of the inammatory variables independently with risk of adverse cardiovascular events during the 1-year follow-up was assessed using logistic regression analysis. Multivariate models were constructed with adjustment for covariates including diagnosis on admis-

Firstly, the sample size of the present study (n = 320) was adequate enough for factor analysis based on the KaiserMeyerOlkin test for sampling adequacy (KMO-test) [23]. As far as the logistic regression analyses regarding the independent contribution of each one of the under investigation variables to adverse events, with a sample size of 320 patients, the study had 80% power to detect odds ratios 1.9 or higher and 0.5 or lower assuming cumulative incidence of adverse coronary event rates of 1520% during 1-year follow-up, at the 5% level of signicance (two-sided). Similarly, for the logistic regression analyses regarding the association of clusters with prospective risk of recurrent coronary events, with a sample size of 320 patients, the study had 80% power to detect odds ratios 1.4 or higher and 0.7 or lower assuming cumulative incidence of adverse coronary event rates of 1520% during 1-year follow-up, at the 5% levels of signicance (two-sided). In both power analyses an adjustment was made taking into account the multiple regression of the independent variable of interest on the other independent variables in the logistic regression.

3. Results 3.1. Study population-baseline characteristics We recruited 320 consecutive patients (236 (74%) men) with ACS of whom 123 had STEMI, 117 NSTEMI and 80 UA, with a median age of 67 (5874) years. Sixty ve percent (n = 207) of the patients were hypertensives, 51% (164) had a history of dyslipidaemia, 30% (96) had a history of diabetes and 38% (122) were current smokers. In addition, 49% of the study participants (n = 158) had a history of established coronary artery disease, 30% of them (n = 95) were on lipid lowering treatment (among the 95 patients in the lipid lower-

D.N. Tziakas et al. / Atherosclerosis 193 (2007) 196203 Table 1 Baseline laboratory values of study population (n = 320) Laboratory variable Interleukin-10 (IL-10) (pg/ml) Interleukin-18 (IL-18) (pg/ml) C-reactive protein (CRP) (mg/l) Fibrinogen (mg/dl) Troponin T (ng/ml) Total cholesterol (mg/dl) LDL cholesterol (mg/dl) HDL cholesterol (mg/dl) Triglycerides (mg/dl) Inter-cellular adhesion molecule-1 (ICAM-1) (ng/ml) 4.7 (3.95.8) 236.2 (153347.8) 2.9 (1.14.7) 486 (399578) 0.78 (0.021.2) 212 (183232) 137 (119151) 43 (3848) 142 (108201) 611 (451886) Table 3 Factor (cluster) analysis results Cluster 1 Cluster 1: systemic inammation CRP 0.875 Fibrinogen 0.888 Cluster 2 0.150 0.046 Cluster 3 0.187 0.136 0.022 0.045 0.625 0.820

199

Cluster 2: local inammationendothelial dysfunction IL-18 0.039 0.804 ICAM-1 0.148 0.731 Cluster 3: anti-inammation IL-10 0.172 HDL cholesterol 0.083 0.408 0.212

Values are expressed as medians with interquartile range.

Loadings 0.500 are in bold. IL, interleukin; CRP, C-reactive protein; HDL, high density lipoprotein; ICAM, inter-cellular adhesion molecule.

ing arm, 92 patients (29%) were on statins, ve patients (2%) were on brates, three patients (1%) were on ezetimibe and one patient (0.3%) was on omega-3 marine triglycerides), 23% (n = 75) were on b-blockers and 30% (n = 96) were on angiotensin converting enzyme inhibitors before study entry. Table 1 summarizes biochemistry results at study entry. 3.2. Correlation analysis Table 2 shows Spearman correlation coefcients between the inammation, anti-inammation and endothelial dysfunction markers that were assessed on admission. The strongest associations were observed between CRP and brinogen (r = 0.590) and IL-18 and ICAM-1 (r = 0.308). More modest but signicant correlations were observed between CRP and ICAM-1 (r = 0.178), IL-10 and IL-18 (r = 0.194) and IL-10 and HDL (r = 0.100). With the same magnitude, albeit inversely, CRP was correlated with IL-10 (r = 0.151) and HDL (r = 0.091). 3.3. Factor analysis Table 3 summarizes factor analysis results for the total group of 320 patients. There was a reduction in variable number from the original 6 to 3 composite clusters, which was supported by the retention criteria described in Section 2. These three clusters taken together explained about 70% of the variance in the original data (cluster 1: 27.52%, cluster 2: 23.34% and cluster 3: 18.28%). These clusters were interpreted as (1) a systemic inammation cluster
Table 2 Spearman correlations between assessed markers (n = 320) Marker IL-10 IL-18 CRP Fibrinogen HDL cholesterol ICAM-1 IL-10 1 0.194* 0.151* 0.011 0.100* 0.084 IL-18 1 0.080 0.045 0.066 0.308* CRP

with positive loadings of CRP and brinogen, (2) a local inammationendothelial dysfunction cluster with positive loadings of IL-18 and ICAM-1 and (3) a protective or antiinammation cluster, with positive loadings of IL-10 and HDL cholesterol. 3.4. Logistic regression analysis for adverse cardiovascular events Eighty-four patients had one or more adverse cardiac events and 236 remained event free during the 1-year followup. During follow-up, there were 60 cardiac deaths (18 during hospitalization and 42 after hospital discharge during the 1year follow-up), and 14 patients were hospitalized for UA and 10 for a new MI. No differences in clinical and biochemical variables were found between patients with recurrent cardiac events and those without, except for the fact that a higher proportion of patients with recurrent events had STEMI on admission (57% versus 32%, p < 0.001), and also had a higher prevalence of reduced ejection fraction (EF) < 40% (36% versus 15%, p < 0.001) compared to patients who were event free during follow-up. The medications taken by both groups during follow-up were similar except for b-blocking agents (29% versus 49%, p = 0.001), statins (25% versus 52%, p < 0.001) and clopidogrel (21% versus 36%, p = 0.014) which were signicantly more common in the event free group. A higher proportion of patients without recurrent events had a revascularization procedure done (CABG or PCI) during follow-up (25% versus 43%, p = 0.004).

Fibrinogen

HDL cholesterol

ICAM-1

1 0.590* 0.091* 0.178*

1 0.067 0.028

1 0.032

IL, interleukin; CRP, C-reactive protein; HDL, high density lipoprotein; ICAM, inter-cellular adhesion molecule. * p < 0.05.

200

D.N. Tziakas et al. / Atherosclerosis 193 (2007) 196203

After adjustment for all the variables that were associated with presence or absence of recurrent coronary adverse events (diagnosis on admission, revascularization during follow-up, b-blocker, statin and clopidogrel use, as well as presence of EF < 40%), multiple logistic regression analysis was performed to assess relative contribution of each one of the under investigation markers to adverse events. IL-18 (OR upper versus lower quartile 3.69, 95% CI: 1.580.68, p = 0.003) and CRP levels (OR upper versus lower quartile 9.85, 95% CI: 3.269.8, p < 0.001) were independent predictors of recurrent coronary events. HDL cholesterol levels (OR upper versus lower quartile 0.28, 95% CI: 0.120.65, p = 0.003) and IL10 levels (OR upper versus lower quartile 0.32, 95% CI: 0.150.74, p = 0.007) were negative predictors. In contrast, raised brinogen levels (OR upper versus lower quartile 1.33, 95% CI: 0.622.84, p = 0.466) and ICAM-1 levels (OR upper versus lower quartile 0.84, 95% CI: 0.391.8, p = 0.659) were not associated with recurrent adverse cardiac events in the multivariable analysis. The association of clusters with prospective risk of recurrent coronary events was assessed using logistic regression analysis. In separate multivariate analyses only the protectiveanti-inammation cluster was a signicant negative predictor (OR 0.66, 95% CI: 0.490.89, p = 0.007) while the systemic inammation cluster (OR 1.15, 95% CI: 0.841.55, p = 0.381) and the local inammationendothelial dysfunction cluster (OR 1.28, 95% CI: 0.951.73, p = 0.109) did not signicantly predict adverse cardiac events during 1-year follow-up (Fig. 1A). In a multivariate model including all three clusters, there was a modest change in the magnitude of associations, however, the protectiveanti-inammation cluster continued to be a signicant negative predictor (OR 0.65, 95% CI: 0.480.88, p = 0.003) while the systemic inammation cluster (OR 1.17, 95% CI: 0.851.60, p = 0.343) and the local inammationendothelial dysfunction cluster (OR 1.33, 95% CI: 0.961.83, p = 0.078) continued not to predict future adverse cardiovascular events (Fig. 1B).

Fig. 1. Cluster odds ratios (ORs) for adverse cardiovascular events during 1-year follow-up in patients with acute coronary syndrome. ORs refer to risk of recurrent coronary events for 1 S.D. difference. (A) Separate multivariate logistic regression analysis including one only cluster at a time adjusted for diagnosis on admission, revascularization during follow-up, b-blocker, statin and clopidogrel use, as well as for presence of EF < 40%. (B) Multivariate logistic regression analysis including all three clusters adjusted for diagnosis on admission, revascularization during follow-up, b-blocker, statin and clopidogrel use, as well as for presence of EF < 40%.

4. Discussion Accumulating evidence that inammation plays a crucial role in pathogenesis of most of the cases of ACS, led to the assessment of various circulating inammatory markers as stable, independent risk predictors in these syndromes [2,25]. However, the fact that all these inammatory mediators are not expressed in isolation but rather concurrently, and are continuously interacting with each other, has rendered their comparison and appraisal problematic. As atherosclerosis-related inammation is a complex process given its pathophysiological mechanisms that involve multiple pathways, statistical assessment of markers is problematic. Factor analysis may be a useful approach to increase our understanding of the contributory role of the various

molecules involved in the atheromatic process and plaque disruption. The multivariate statistical technique of factor analysis is a natural approach for investigations when multiple highly correlated variables are concerned, and several studies have appeared recently that successfully used this approach [26,27]. Factor analysis in this study was performed to reduce the complexity of multivariable risk analysis. One of the major advantages of this type of analysis is minimization of correlations among variables. Thus, factor analysis is often useful in settings where collinearity of independent variables is problematic, as in the present study involving several inammatory markers. Clustering of independent markers into composite variables represented by factors could potentially be useful in risk assessment by utilization of all information in the data set while at the same time minimizing collinearity problems. Another advantage of factor analysis is the ability to dene relationships between variables related to underlying pathophysiologic mechanisms. Our study is among the rst to evaluate the clustering of multiple inammatory markers in the ACS setting, and their prognostic ability in these patients. Three clusters have

D.N. Tziakas et al. / Atherosclerosis 193 (2007) 196203

201

emerged in the present analysis from concurrent admission measurement of inammation mediators levels. The rst included CRP and brinogen, and was interpreted as a marker of systemic inammation, the second consisted of IL18 and ICAM-1, which was interpreted as an indicator of local inammationendothelial dysfunction and the third comprising both IL-10 and HDL cholesterol, was considered to represent a protective anti-inammatory marker. Overall, we found that the protective anti-inammatory cluster was signicantly associated with a good prognosis in ACS patients during 1-year follow-up. In addition, the anti-inammatory cluster had better prognostic ability when compared to the systemic inammation and the local inammationendothelial dysfunction clusters. Clustering of HDL cholesterol with IL-10, a potent antiinammatory cytokine is in accordance with studies in the literature [28,29]. In terms of anti-inammatory effects, HDL cholesterol reduces cytokine-mediated up-regulation of cell adhesion molecules, blocks the nuclear-factor kappa B signaling cascade and is associated with reduced matrix metalloproteinases levels [28,29]. In addition, IL-10, which has been shown to be low in ACS patients compared to stable angina [17], is also capable of down-regulating numerous inammatory pathways that play an important role in the progression and instability of the atherosclerotic plaque. These include, among others, the inhibition of the pro-inammatory transcription factor nuclear factor kappa B, suppression of cytokine production, reduced matrix metalloproteinase production, reduced tissue factor expression, suppressed expression of adhesion molecules, and inhibition of the release of potent chemoattractants [30]. The above mechanisms may possibly account for the protective role of the antiinammation cluster in ACS patients. Clustering of IL-18 with ICAM-1 is in agreement with studies that showed expression of IL-18 in human atherosclerotic plaques [31], as well as IL-18 induced expression of adhesion molecules in vitro [32]. Taken account this experimental background, this cytokine appears to be an excellent candidate liniking local vessel pathology with inammation. Furthermore, clustering of CRP and brinogen, two acute phase proteins, reects the presence of systemic low grade inammation in the coronary arteries or elsewhere [33]. These ndings are in keeping with prior data published by our group [34] and others [35], suggesting that evaluating the imbalance between pro-inammatory and antiinammatory mediators concurrently, may be of signicance in risk prediction in ACS patients. Although previous studies have examined prospective associations between inammatory clusters and risk of cardiovascular events [26,27], these studies were conducted in highly selective populations and neither included a true anti-inammatory mediator. The major contribution of this study is the inclusion of a representative marker of the anti-inammatory pathway and HDL cholesterol levels, in addition to several markers of inammation.

The nding of the present study of a positive association between serum levels of IL-18 and IL-10 is of particular interest. Different and various direct associations have been reported in the current literature regarding interactions between circulating concentrations of proand anti-inammatory markers. Heeschen et al. reported an inverse association between CRP and IL-10 [13]. Similarly, Yamashita et al. showed that levels of the pro-inammatory markers IL-6, CRP and IL-12 correlated negatively with IL10 levels [36], while others have reported no association at all [37]. In accordance to our study, Biasucci et al. showed that in unstable angina patients circulating levels of the antiinammatory marker IL-1 receptor antagonist was directly associated with IL-1 and IL-6 probably reecting disease activity [7]. In addition, obesity which is characterized by increased levels of CRP and IL-6 is also associated with a parallel increase of IL-10 [11]. It is possible that expression of anti-inammatory molecules to reect a counter-regulatory mechanism in response to inammatory stimuli. Recent studies supported the notion that a lack of an appropriate antiinammatory counter-balance to a certain pro-inammatory reaction rather than the pro- or the anti-inammatory mechanism per se might be the culprit of the increased coronary risk [34,35]. The ndings of the present study should be interpreted in light of certain limitations. Firstly, our investigation due to the relative small sample size was subject to type II error regarding the prognosis assessment of the local inammationendothelial dysfunction cluster. Even with the relatively small sample size the present study shows that markers reecting anti-inammatory mechanisms are better predictors of adverse events when compared to markers of endothelial dysfunction. Secondly, the present study design did not allow us to assess the predictive value of the under investigation clusters in sub-group of patients i.e. statin or b-blocker users. Furthermore, ICAM-1 is also expressed by a great variety of non-endothelial cells therefore does not reect specically endothelial dysfunction-activation as E-selectin [38]. However, serum concentration of soluble ICAM-1 is a better predictor of future cardiovascular events compared to E-selectin levels [39,40]. Therefore, the use of Eselectin, an endothelial-specic activation marker could minimize the ability of the local inammationendothelial dysfunction cluster to predict adverse coronary events. Finally, other strong prognostic markers established in risk stratication in ACS patients such as brain natriuretic peptide, or circulating indices reecting plaque activation i.e. matrix metalloproteinases and soluble CD40 ligand have not been measured. Inclusion of the above markers into the factor analysis derived clusters would be of importance. Although this was not possible, the present study showed that multivariate risk appraisal is vital because the various risk markers tend to cluster and the risk associated with any particular marker varies widely depending on the amount of clustering. In conclusion, our study further supports the concept that in patients with ACS when all inammatory mediators are

202

D.N. Tziakas et al. / Atherosclerosis 193 (2007) 196203 [17] Smith DA, Irving SD, Sheldon J, Cole D, Kaski JC. Serum levels of the anti-inammatory cytokine interleukin-10 are decreased in patients with unstable angina. Circulation 2001;104:7469. [18] Mallat Z, Henry P, Alouani S, et al. Increased plasma concentrations of interleukin-18 in acute coronary syndromes. Heart 2002;88: 4679. [19] OMalley T, Ludlam CA, Riemermsa RA, Fox KA. Early increase in levels of soluble inter-cellular adhesion molecule-1 (sICAM-1); potential risk factor for the acute coronary syndromes. Eur Heart J 2001;22:122634. [20] Ledue TB, Weiner DL, Sipe JD, et al. Analytical evaluation of particle enhanced immunonephelometric assays for C-reactive protein, serum amyloid A and mannose-binding protein in human serum. Ann Clin Biochem 1998;35:74553. [21] Bod V, Sanchis J, Llacer A, et al. Risk stratication in non-ST elevation y acute coronary syndromes. Predictive power of troponin I, C-reactive protein, brinogen and homocysteine. Int J Cardiol 2005;98:277 83. [22] Cureton EE, DAgostino RB. Factor analysis. In: Factor analysis: an applied approach. Hillside, NJ: Lawrence Erlbaum Associates; 1983. [23] Field AP. Factor analysis on SPSS. In: Discovering statistics using SPSS. 2nd ed. London, UK: Sage; 2005. [24] Stevens J. Factor analysis. In: Applied multivaried statistics for the social sciences. Hillsdale, NJ: Lawrence Erlbaum Associates; 1986. [25] Biasucci LM. CDC/AHA workshop on markers of inammation and cardiovascular disease application to clinical and public health practice clinical use of inammatory markers in patients with cardiovascular diseases: a background paper. Circulation 2004;110:e5607. [26] Kip KE, Marroquin OC, Shaw LJ, et al. Global inammation predicts cardiovascular risk in women: a report from the Womens Ischemia Syndrome Evaluation (WISE) study. Am Heart J 2005;150:9006. [27] Corsetti JP, Zareba W, Moss AJ, Rainwater DL, Sparks CE. Elevated HDL is a risk factor for recurrent coronary events in a subgroup of non-diabetic post-infarction patients with hypercholesterolemia and inammation. Atherosclerosis 2006;187:1917. [28] Tziakas DN, Chalikias GK, Hatzinikolaou HI, et al. Anti-inammatory cytokine prole in acute coronary syndromes: behavior of interleukin10 in association with serum metalloproteinases and pro-inammatory cytokines. Int J Cardiol 2003;92:16975. [29] Xia P, Vadas MA, Rye KA, Barter PJ, Gamble JR. High density lipoproteins (HDL) interrupt the sphingosine kinase signaling pathway. J Biol Chem 1999;274:331437. [30] Terkeltaub RA. IL-10: an immunologic scalpel for atherosclerosis? Arterioscler Thromb Vasc Biol 1999;19:28235. [31] Mallat Z, Corbaz A, Scoazec A, et al. Expression of interleukin-18 in human atherosclerotic plaques and relation to plaque instability. Circulation 2001;104:1598603. [32] Gerdes N, Sukhova GK, Libby P, et al. Expression of interleukin (IL18) and functional IL-18 receptor on human vascular endothelial cells, smooth muscle cells and macrophages: implication for atherogenesis. J Exp Med 2002;195:24557. [33] Gabay C, Kushner I. Acute phase proteins and other systemic responses to inammation. N Engl J Med 1999;340:44854. [34] Chalikias GK, Tziakas DN, Kaski JC, et al. Interleukin-18: interleukin10 ratio and in-hospital adverse events in patients with acute coronary syndrome. Atherosclerosis 2005;182:13543. [35] Kilic T, Ural D, Ural E, et al. Relation between pro- to anti-inammatory cytokine ratios and long term prognosis in patients with non-ST elevation acute coronary syndrome. Heart 2006;(March 17) (epub ahead of print). [36] Yamashita H, Shimada K, Seki E, Mokuno H, Daida H. Concentrations of interleukins, interferon and C-reactive protein in stable and unstable angina pectoris. Am J Cardiol 2003;91:1336. [37] Fichtlscherer S, Breuer S, Heescen C, Dimmeler S, Zeiher AM. Interleukin-10 serum levels and systemic endothelial vasoreactivity

taken into account in a prospective analysis of risk, markers reecting anti-inammatory mechanisms are very important determinants of patient outcome. Although at present CRP is the only marker endorsed for cardiovascular risk stratication in clinical practice [25], our ndings provide evidence for the prognostic importance of the imbalance between pro-inammatory and anti-inammatory mechanisms in the ACS setting. Larger studies are needed to evaluate in clinical practice, if therapeutic interventions that improve patient outcome after ACS are associated with potentiation of antiinammatory mechanisms.

References
[1] Davies MJ. Stability and instability: the two faces of coronary atherosclerosis. Circulation 1996;94:201320. [2] Ross R. Atherosclerosis: an inammatory disease. N Engl J Med 1999;340:11526. [3] Blake GJ, Ridker PM. Novel clinical markers of vascular wall inammation. Circ Res 2001;89:76371. [4] Lind L. Circulating markers of inammation and atherosclerosis. Atherosclerosis 2003;169:20314. [5] Blake GJ, Ridker PM. C-reactive protein and other inammatory risk markers in acute coronary syndromes. J Am Coll Cardiol 2003;41:37S42S. [6] Wallen NH, Held C, Rehnqvist N, Hjemdahl P. Elevated serum intercellular adhesion molecule-1 and vascular adhesion molecule-1 among patients with stable angina pectoris who suffer cardiovascular death or non-fatal myocardial infarction. Eur Heart J 1999;20:103943. [7] Biasucci LM, Liuzzo G, Fantuzzi G, et al. Increasing levels of interleukin (IL)-1Ra and IL-6 during the rst 2 days of hospitalization in unstable angina are associated with increased risk of in-hospital coronary events. Circulation 1999;99:207984. [8] Blankenberg S, Tiret L, Bickel C, et al. Interleukin-18 is a strong predictor of cardiovascular death in stable and unstable angina. Circulation 2002;106:2430. [9] Cosin-Sales J, Christiansen M, Kaminski P, et al. Pregnancy-associated plasma protein A and its endogenous inhibitor, the proform of eosinophil major basic protein (proMBP), are related to complex stenosis morphology in patients with stable angina pectoris. Circulation 2004;109:17248. [10] Avanzas P, Arroyo-Espliguero R, Quiles J, Roy D, Kaski JC. Elevated serum neopterin predicts future adverse events in patients with chronic stable angina pectoris. Eur Heart J 2005;26:45763. [11] Mills R, Bhatt DL. The Yin and Yang of arterial inammation. J Am Coll Cardiol 2004;44:502. [12] Tedgui A, Mallat Z. Anti-inammatory mechanisms in the vascular wall. Circ Res 2001;88:87787. [13] Heeschen C, Dimmeler S, Hamm CW, et al. Serum levels of the anti-inammatory cytokine interleukin-10 is an important prognostic determinant in patients with acute coronary syndromes. Circulation 2003;107:210914. [14] ACC Clinical Data Standards. American College of Cardiology key data elements and denition for measuring the clinical management and outcomes of patients with acute coronary syndromes. J Am Coll Cardiol 2001;38:211430. [15] Myocardial infarction redeneda consensus document of the Joint European Society of Cardiology/American College of Cardiology Committee for the redenition of myocardial infarction. Eur Heart J 2000;21:150213. [16] Braunwald E. Unstable angina: a classication. Circulation 1989;80:4104.

D.N. Tziakas et al. / Atherosclerosis 193 (2007) 196203 in patients with coronary artery disease. J Am Coll Cardiol 2004;44: 449. [38] Niessen H, Krijnen P, Visser C, Meijer C, Hack C. Intercellular adhesion molecule-1 in the heart. Ann N Y Acad Sci 2002;973:57385. [39] Hwang SJ, Ballantyne CM, Sharrett RA, et al. Circulating adhesion molecules VCAM-1, ICAM-1 and E-selectin in carotid

203

atherosclerosis and incident coronary heart disease cases. Circulation 1997;96:421925. [40] OMalley T, Ludlam CA, Riemermsa RA, Fox KAA. Early increase in levels of soluble inter-cellular adhesion molecule-1 (sICAM-1). Potential risk factor for the acute coronary syndromes. Eur Heart J 2001;22:122634.