Author + information
- Received February 19, 1999
- Revision received July 2, 1999
- Accepted September 7, 1999
- Published online December 1, 1999.
- Mustafa Porsch-Oezçueruemez, MD∗,* (, )
- Dagmar Kunz, MD†,‡,
- Hans-Ulrich Kloer, MD† and
- Claus Luley, MD‡
- ↵*Reprint requests and correspondence: Dr. Mustafa Porsch-Oezçueruemez, Institut für Klinische Chemie und Blutbank, Universitätsklinikum Regensburg, Franz-Josef-Strauss-Allee 11, 93053 Regensburg, Germany
The diagnostic importance of circulating solubilized tumor necrosis factor-alpha receptor II (sTNF-αRII) and interleukin-2 receptor in coronary heart disease (CHD) was evaluated. In addition, these variables were correlated with solubilized adhesion molecule levels (i.e., intracellular adhesion molecule [ICAM], vascular cell adhesion molecule [VCAM], and E-selectin).
Atherosclerosis is considered to be a chronic inflammatory process. Because the immunologic network presents an abundance of positive and negative feedback mechanisms, information obtained from different immunologic variables might be highly redundant.
In a cross-sectional study design, 60 patients with angiographically proven CHD were compared with 60 individuals who had undergone coronary angiography but in whom no evidence of stenosis could be found (control subjects). Angiography was performed at least one year before the study. Cytokine levels were determined by enzyme-linked immunosorbent assay technique and evaluated by univariate and multivariate statistical methods.
All immunologic variables except E-selectin (p = 0.08) significantly discriminated between patients and control subjects. Coronary artery bypass graft surgery after angiography did not lead to significant differences in the variables investigated in the patients with bypass surgery as compared with the subjects without bypass surgery. Receiver-operating characteristics analysis showed comparable test accuracy for solubilized adhesion molecules and cytokine receptors. Multivariate logistic regression analysis, including age, revealed that only ICAM and sTNF-αRII were independently correlated with the presence of CHD. Patients belonging to the third tertile of at least one of these two variables demonstrated a 1.6- to 2-fold increased relative risk for the presence of CHD.
We concluded that both circulating ICAM and TNF-αRII should be further evaluated as markers for atherosclerotic changes in the coronary system.
Atherosclerosis can be described as a chronic inflammatory process (1). During recent years there has been growing interest in the evaluation of suitable markers of inflammation, which might be associated with atherogenesis. Because the adhesion of monocytes to the endothelium through adhesion molecules such as vascular cell adhesion molecule (VCAM), intracellular adhesion molecule (ICAM) and E-selectin is an early event in atherogenesis, these markers are of particular interest. Many studies have reported a relation between these adhesion molecules and acute or chronic atherosclerotic vascular disease (2–4). All of these variables are rapidly synthesized in response to cytokines (5). However, most of these cytokines appear only transiently in blood and have a short half-life. Circulating cytokine receptors, which are released after contact with the corresponding cytokine, also show basal serum levels in healthy subjects. Because their serum concentrations correlate with the cellular response to cytokines, circulating cytokine receptors might also provide additional information in chronic inflammatory processes such as atherosclerosis. In this study we evaluated the importance of solubilized tumor necrosis factor-alpha receptor II (sTNF-αRII) (p75) as an indirect marker of monocyte/macrophage stimulation and solubilized interleukin-2 (IL-2) receptor (CD25) as a marker for T-lymphocyte activation in patients with coronary heart disease (CHD). In addition, we analyzed circulating adhesion molecules (VCAM, ICAM and E-selectin) and correlated all immunologic variables in a multivariate model to evaluate whether additional information can be obtained by cytokine receptor levels or whether the measurement of these variables provides only redundant information concerning the presence of an inflammatory state of atherosclerosis.
All subjects included in this study were suspected of having CHD. As a consequence, these patients had to undergo coronary angiography at our hospital. To obtain a sufficient amount of patients and control subjects, 805 consecutive records of patients with coronary angiography performed at least one year but no longer than two years before recruitment were reviewed. Information on angiographic findings like degree and localization of a stenosis of the coronary arteries was obtained from standardized report forms, which had to be filled out by an experienced cardiologist directly after evaluation of the angiographic study. All procedures were performed according to the recommendations of quality management in the heart catheterization laboratory (6). The proportion of angiographic studies without significant pathologic findings ranged between 10% and 15% during the last two years, which is in accordance with performance statistics of heart catheterization laboratories in Germany (7).
A subject was considered as a control if no stenosis of the coronary arteries was reported. In addition, no clinical evidence such as alterations in the electrocardiogram were apparent since angiography. Patients with CHD were defined as those individuals exhibiting a significant stenosis of >50% of at least one main branch of the coronary arteries. Patients with myocardial infarction or unstable angina within the last three months were excluded from further study. Furthermore, none of the patients with CHD had a history of coronary artery bypass graft surgery (CABG) or percutaneous transluminal coronary angioplasty within the last year.
The subjects examined were invited by phone for an informal visit to the hospital regarding the aims of the study. The recruitment of participants was continued until a balanced distribution of 30 men and 30 women in both the patient and the control group was reached. All of the 60 patients and 60 control subjects gave written consent in compliance with the local Human Studies Committee and were included in the study between December 1996 and October 1997. Age ranged from 35 to 78 years in the control group and from 42 to 76 years in the CHD group. Of the 60 patients with CHD, 28 (47%, 14 men and 14 women) had a history of CABG >1 year before recruitment. Subjects with acute or chronic inflammatory diseases, current smokers, subjects with malignancies, renal, hepatic or thyroid diseases or diabetes mellitus or those treated with immunosuppressive or cytotoxic drugs were excluded.
Blood sampling and laboratory determinations
Blood was drawn in the sitting position from the antecubital vein between 8:00 and 9:00 amin the fasting state. All medications were withdrawn for at least 12 h. The blood samples were immediately cooled on ice. Serum was obtained by subsequent centrifugation at 3,000gfor 10 min. All samples were stored at −70°C until assayed.
Total cholesterol and triglyceride levels were measured enzymatically by the CHOD-PAP and GPO-PAP methods (Boehringer Mannheim, Germany). Low density lipoprotein and high density lipoprotein cholesterol levels were determined as recommended by the guidelines of the Lipid Research Clinics Program (8). C-reactive protein (CRP) was measured by nephelometry (Boehringer Mannheim, Mannheim, Germany). Leukocyte count was performed using a CellDyne 3000 auto-analyzer (Abbott, Wiesbaden, Germany). Serum levels of solubilized adhesion molecules (VCAM, ICAM and E-selectin) and solubilized cytokine receptors (TNF-αRII and IL-2 receptor) were determined by commercially available enzyme-linked immosorbent assay (DPC Biermann, Bad Nauheim, Germany).
The Statistical Package for the Social Science (SPSS for Windows 6.01) was employed for statistical analysis. Variables were tested by the Mann-Whitney Utest to analyze differences of the distribution between the two groups. Differences in prevalence were tested by the nonparametric chi-square test. Multiple linear regression analysis and multiple logistic regression analysis were performed to evaluate the independent effects of the distinct immunologic variables investigated with respect to CHD. Receiver-operating characteristics (ROC) analysis was performed according to Beck and Shultz (9)to estimate the potential of the variables tested to discriminate between patients with CHD and control subjects (10). All analyses were tested two-sided and p values <0.05 were considered statistically significant.
Table 1presents the clinical characteristics of the subjects investigated. Patients and control subjects differed in their age distribution with significantly higher median values in the patient group (65 vs. 62 years). Fifty (83%) of the investigated 60 patients with CHD had a history of past myocardial infarction. Patients with one-vessel disease (n = 28) slightly prevailed over patients with two- or three-vessel disease (n = 16 each). There was a subset of patients with a reduced ejection fraction (<60%); however, no significant differences in prevalence were obvious between patients and control subjects.
Anthropometric variables (body mass index, waist to hip ratio) and systolic and diastolic blood pressure were comparable between the groups. Lipid profiles were similar in patients and control subjects. However, there was a significantly higher prevalence of subjects receiving lipid-lowering drugs in the CHD group (48% vs. 8%). In contrast, use of diuretic agents and beta-blocker did not differ significantly between the subgroups investigated. Smoking habits were also comparable with respect to the prevalence of former smokers, because current smokers were not included in this study.
White blood cell counts were within the reference range (<10 cells/nl) in all subjects investigated and did not differ statistically between patients and control subjects. In parallel, CRP levels were also comparable between the subgroups. Three subjects in the CHD group and two subjects in the control group exhibited slightly elevated CRP levels (10 to 20 mg/l) without any obvious clinical correlate.
Univariate analysis of serum levels of solubilized adhesion molecules and cytokine receptors
As shown in Table 2, significant differences in plasma levels were observed for solubilized VCAM, ICAM, TNF-αRII and IL-2 receptor, with higher median values in the CHD group. Solubilized E-selectin did not reach the level of significance (p = 0.086), even though median values were 30% (63 vs. 48 ng/ml) higher in patients with CHD as compared with control subjects. However, there was a wide overlap between the two groups. As shown in Figure 1, control subjects with a reduced ejection fraction tended to have higher adhesion molecule and cytokine receptor levels, which in fact reduced the median differences between patients and control subjects. Interestingly, vessel disease without former myocardial infarction was also accompanied by elevated adhesion molecule and cytokine receptor levels (with the exception of solubilized ICAM).
There were also no significant gender-specific differences in adhesion molecule and cytokine receptor levels (data not shown). Solubilized TNF-αRII and IL-2 receptor levels significantly correlated with age (p < 0.01), even though the resulting correlation coefficients of linear regression analysis were only 0.266 and 0.296, respectively.
Previous CABG in patients with CHD (n = 28) did not result in significant differences in the distribution of all immunologic variables investigated in comparison to patients with CHD without CABG (data not shown). The use of lipid-lowering drugs, beta-blockers, angiotensin-converting enzyme inhibitors or diuretic agents also had no obvious effect on the results obtained.
The ROC analysis was performed to provide an index for assessment of the accuracy of the variables investigated to discriminate between subjects with angiographically proven CHD and those without it. The cutoff values and the areas under the curve (AUC) are given in Table 3. The AUC represents the probability that a randomly chosen patient with disease is correctly rated or ranked with greater suspicion than a randomly chosen subject without disease. There were no obvious differences in AUC values between the variables listed, which ranged from 0.557 (sVCAM) to 0.606 (sTNF-αRII). The discriminative power of age was comparable to the immunologic variables (AUC = 0.605).
Logistic regression analysis
To assess the independent interrelation between the immunologic variables investigated and CHD, multivariate logistic regression analysis was performed, including age in the model as another covariate with statistical significance in univariate analysis.
The power of the multiple logistic regression model used was calculated as 99.7%. The estimation of model discrimination as a means to evaluate the model performance showed that 68% of all subjects were correctly classified. The goodness-of-fit of this model, as quantified by the Nagelkerke R2coefficient, estimated with a p value of 0.005 that 23% of the variation in the logistic regression model is explained by the variables included in the model. This proportion reflects a “median” influence of the covariates investigated on the dependent variable as categorized by Cohen (11). The cutoff values assuming a weak, median or strong influence of covariates included in a regression model are given as 2% to 15%, 16% to 35% and >35%, respectively, in this reference.
As indicated in Table 4, of all the dependent variables only serum levels of solubilized TNF-αRII and ICAM remained statistically significant. Both variables contributed equally to the goodness-of-fit of the model as indicated by R values of partial correlation of 0.121 in both. Categorizing serum levels of solubilized TNF-αRII and ICAM by tertiles (cutoff values of tertiles were 2,977/3,619 ng/ml for sTNF-αRII and 137/168 ng/ml for sICAM) revealed a 1.6- to 2.0-fold elevated relative risk attributable to both variables as compared with the group belonging to the first tertile. No obvious additive effects of both variables were observed (data not shown).
In this study we focused on the role of several immunologic variables that reflect distinct parts of the immunologic system and their correlation with angiographically proven CHD. Because the immunologic network presents an abundance of positive and negative feedback mechanisms, information obtained from different immunologic variables might be highly redundant. A large number of growth factors, cytokines and vasoregulatory molecules are known to participate in atherogenesis (1).
Solubilized adhesion molecules and cytokine receptors discriminate patients with CHD
Although leukocyte count and CRP levels, as the two classic indicators of inflammatory diseases, could discriminate patients with CHD (12,13), these variables are obviously insensitive in a cohort such as that investigated in our study. C-reactive protein levels were relatively high as compared with those in other studies (14). This might in part result from differences in the methods used, because nephelometry, which was used in our study, overestimates CRP levels in comparison to approaches using enzyme-linked immunosorbent assay (15).
To our knowledge, this is the first study to investigate the diagnostic value of solubilized adhesion molecules and cytokine receptors for CHD in the same study group. As confirmed by our study, significant differences between patients with CHD and control subjects in serum levels of these immunologic variables investigated were apparent using univariate statistical procedures. In particular, we did not find any differences between patients with and without CABG. Therefore, it is unlikely that elevated levels of circulating adhesion molecules and cytokine receptors in patients with CHD in our study were caused by this intervention.
Because subjects were not matched for age, differences in the age distribution between groups were also expected. However, as a limitation of this study, it has to be mentioned that the selection criteria, including the setting of the study, might have biased the age distribution of the control group and diminished the validity of statements concerning an independent influence of age on the presence of atherosclerotic lesions in coronary vessels.
The test accuracy of all variables was comparable, as confirmed by ROC analysis (Table 3). In contrast, there were no obvious differences in risk factors such as lipids, which was partly owing to the treatment with lipid-lowering drugs in the CHD group (Table 1). Data on the influence of hepatic hydroxymethyl glutaryl coenzyme A (HMG-CoA) reductase inhibitors on inflammatory processes are rare. Fukuo et al. (16)did not observe an influence of simvastatin on the cytokine production by human monocytes and macrophages. Giroux et al. (17)reported a diminished phagocytotic activity of human monocyte-derived macrophages after in vitro treatment with simvastatin. These observations provide some evidence that HMG-CoA reductase inhibitors either do not influence the inflammatory response at all or decrease it.
Role of solubilized adhesion molecules and cytokine receptors in atherosclerosis
When introducing serum levels of solubilized adhesion molecules, cytokine receptor and age in a multivariate logistic regression model, only solubilized TNF-αRII and ICAM remained significantly correlated with the presence of CHD. This implies that these variables are more independently related to the appearance of atherosclerosis than the other variables investigated. TNF-αRII is expressed by many different cell types (18). Its intracellular domain, in concert with TNF-αRI, is proposed to be responsible for the signal transduction of TNF-alpha (19). The proteolytic cleavage of the receptors from the cell surface seems to be a downregulating mechanism of the cell response to TNF-alpha (20). The solubilized receptor is therefore closely related to TNF-alpha levels. The capability of sTNF-αRII to independently discriminate patients with CHD might derive from the central role of monocytes and macrophages as a source of TNF-alpha in all stages of atherogenesis (21). In contrast to TNF-alpha, which has a very short half-life (22), the solubilized receptor II shows stable basic serum levels (23)and is therefore more suitable for laboratory investigations.
Solubilized IL-2 receptors have been extensively studied in transplant atherosclerosis (24). The IL-2 receptor is upregulated by IL-2, and its shedding is proposed to be a regulatory means of IL-2 response (25). The presence of CD25+ T lymphocytes in atherosclerotic lesions supported the theory of atherosclerosis as a chronic inflammatory process (26), even though the association with serum levels of the shedded receptor is unclear. In our study, however, the significant differences between patients with CHD and subjects without it were largely diminished by the other variables of our study, as also observed for sVCAM and sE-selectin. Elevated serum levels of IL-2 receptor were also observed in patients with dilated cardiomyopathy and correlated positively with New York Heart Association functional class (27). This observation was also suggestively seen in the other adhesion molecule and cytokine receptor levels investigated and fits well with current concepts on the pathogenesis of the clinical syndrome of heart failure (28).
Although there are many pathophysiologic connections between adhesion molecules and atherogenesis (1,29), there are still conflicting results reported with regard to the discriminative value of distinct solubilized adhesion molecules (2–4). In our collective, only sICAM was independently related to the presence of atherosclerosis in a multivariate model including age and serum levels of three adhesion molecules and two cytokine receptors. Most of the other studies did not use multivariate analysis to confirm the independent correlation of these variables to atherosclerosis. In our study E-selectin and VCAM serum levels seem to be obviously affected by the other covariates. Morisaki et al. (2)also showed that sICAM is related to atherosclerosis independent of sVCAM in a multivariate analysis. However, neither their study nor our own can solve the question of causality regarding atherosclerosis. It also has to be stated that although sICAM and TNF-αRII levels emerged as two independent discriminators for the presence of atherosclerotic lesions in coronary arteries, in our study the pathophysiologic background of the absence of an expected additive effect of both variables remains unclear and must be further elucidated.
The present results indicate that both circulating ICAM and TNF-αRII should be further evaluated as two potential sensitive immunologic markers for atherosclerotic changes in the coronary system.
- area under the curve
- coronary artery bypass graft surgery
- coronary heart disease
- C-reactive protein
- hepatic hydroxymethyl glutaryl coenzyme A
- (solubilized) intracellular adhesion molecule
- receiver-operating characteristics
- (solubilized) tumor necrosis factor-alpha receptor II
- (solubilized) vascular cell adhesion molecule
- Received February 19, 1999.
- Revision received July 2, 1999.
- Accepted September 7, 1999.
- American College of Cardiology
- Morisaki N,
- Saito I,
- Tamura K,
- et al.
- Peter K,
- Nawroth P,
- Conradt C,
- et al.
- Erbel R,
- Sommerfeld U,
- Ashry M,
- et al.
- ↵National Institutes of Health. Manual of Laboratory Operations. Vol. 1. Washington, DC: Lipid Research Clinics Program, DHEW, NIH publication no. 75-628, 1974.
- Zweig M.H,
- Campbell G
- Cohen J
- Ledue T.B,
- Weiner D.L,
- Sipe J.D,
- et al.
- Tartaglia L.A,
- Pennica D,
- Goeddel D.V
- Beutler B.A,
- Milsark I.W,
- Cerami A