Relationship of Anti-60 kDa Heat Shock Protein and Anti-Cholesterol Antibodies to Cardiovascular Events
Background— Several recent studies have indicated an association between key inflammatory mediators and atherosclerotic diseases. We evaluated whether high levels of antibodies against heat shock proteins and cholesterol (ACHA) predicted cardiovascular (CV) events.
Methods and Results— We used blood samples from the Heart Outcomes Prevention Evaluation (HOPE) study to conduct a nested case-control study of 386 cases with CV events and 386 age- and sex-matched HOPE study controls without events. We explored the relationship between anti-hsp antibodies, ACHA, and subsequent outcomes (incident myocardial infarction, stroke, or CV death) during a mean follow-up of 4.5 years using conditional logistic regression. High levels of anti-hsp65 antibodies (≥90th percentile) predicted CV events (OR, 2.1; 95% CI, 1.2 to 3.9, P=0.01). Anti-hsp60 antibodies did not predict any event type, whereas incident stroke developed significantly less frequently in patients with high ACHA levels. Anti-hsp antibodies and ACHA did not correlate with inflammatory (fibrinogen, C-reactive protein, interleukin-6, intracellular adhesion molecule-1) or infectious markers (C pneumoniae or cytomegalovirus antibodies). Anti-hsp65 antibodies (≥90th percentile) and fibrinogen (highest tertile) had a strong joint effect: patients with high concentrations of both had more CV events (OR, 5.5; 95% CI, 1.8 to 17.5, P=0.004) than patients with low levels of both. A similar joint effect (OR, 2.7; 95% CI, 1.3 to 5.7, P=0.01) was found for high levels of anti-hsp65 and presence of cytomegalovirus antibodies.
Conclusions— Serum antibodies to hsp65 were associated with subsequent CV events in this study of high-risk patients, independent of conventional cardiovascular risk factors and other inflammatory markers.
Received May 15, 2002; revision received September 10, 2002;, accepted September 11, 2002.
Pathological and clinical data suggest a prominent role for inflammation at every stage of atherogenesis.1 Several studies have found an association between key mediators and markers of the inflammatory process and atherosclerotic disease, including C-reactive protein (CRP),2,3⇓ fibrinogen,3 complement components,4 and 60-kDa heat-shock protein (hsp60), a putative autoantigen.5 Heat shock proteins occur ubiquitously in prokaryotic and eukaryotic cells, have essential functions under normal growth conditions, and play a role in immunological response.6 High levels of autoantibodies against 60-kDa members of the hsp60 family were associated with atherosclerosclerotic vascular diseases.7–10⇓⇓⇓ We11 and others12 recently demonstrated an association between autoantibodies to human hsp60 and both the presence and the severity of coronary artery disease (CAD).
Alving and Swartz13 found that sera of virtually all healthy subjects contain IgM and IgG autoantibodies against cholesterol (ACHA). We found that the serum concentration of ACHA was significantly lower in patients with peripheral atherosclerosis and cerebrovascular disease but significantly higher in patients with CAD compared with age-matched healthy controls.14
Because few studies have addressed the predictive value of high levels of anti-hsp antibodies for subsequent cardiovascular events and no prospective study has examined ACHA, we related the levels of these antibodies at study baseline to future clinical events in the prospective Heart Outcomes Prevention Evaluation (HOPE) study. The HOPE study was a double-blind, randomized trial, conducted to evaluate the effects of ramipril and vitamin E in 9541 patients at high risk for cardiovascular (CV) events. Treatment with ramipril for a mean of 4.5 years significantly reduced the primary outcome of CV death, myocardial infarction (MI), and stroke, whereas vitamin E had no apparent effect on cardiovascular outcomes.15,16⇓
Serum samples frozen at study enrollment from 3168 Canadian patients were used to explore the relationship between inflammatory and infectious markers and future cardiovascular events. The analysis of associations between some inflammatory and infectious markers with subsequent cardiovascular events in this entire patient cohort is summarized separately (Smieja et al, unpublished data, 2002). In this study, we measured anti-hsp antibodies to human hsp60 (anti-hsp60) and mycobacterial hsp65 (anti-hsp65) and anti-cholesterol antibodies (ACHA) in a prospective nested case-control study of HOPE study patients to determine whether these factors predicted cardiovascular outcomes and to examine their relationship with the other inflammatory and infectious risk markers.
Study Population and Design
Baseline serum and plasma samples from 3168 Canadian patients enrolled in the HOPE study were frozen and stored at −70°C in the HOPE Central Laboratory, Hamilton General Hospital, Hamilton Health Sciences, Hamilton, Canada. A total of 386 cases of incident MI, stroke, or CV deaths were matched one to one with study controls (patients who did not develop any event enlisted above) by sex and age. A total of 386 matched pairs, or 772 patients, were analyzed for this study. Patient characteristics from the entire HOPE study and the nested case-control substudy were similar in most characteristics. No differences were found between the two groups in the occurrence of diabetes, hypertension, and previous stroke and the proportion of patients treated with ramipril.
Because cases were overrepresented in the present nested case-control study, the proportion of males, current smokers, and patients with previous CAD and previous MI was higher in the participants of the present study than among the entire HOPE study population.
Determination of Anti-hsp 60/65 Antibodies
The amounts of IgG-type antibodies reacting with the hsp60 (recombinant human hsp60, StressGen, SPP-740, Victoria, BC, Canada) and hsp65 (recombinant M bovis hsp65, Lionex, Braunschweig, Germany) were assessed by ELISA, as described previously.17,18⇓ Serial dilutions of serum samples of healthy blood donors having high antibody levels against the tested heat shock proteins were used as standard. Data obtained as optical density values were calculated to arbitrary unit per milliliter values related to this standard. Because levels of these antibodies are expressed in arbitrary units, higher apparent titers of one antibody does not mean that its true serum concentration is higher, too.
Determination of Cholesterol-Specific Antibodies
The measurement of ACHA by a solid-phase enzyme immunoassay (EIA) was performed as described previously14 by using polystyrene plates (Greiner) coated with 5 μg/well cholesterol dissolved in 100 μL absolute ethanol. Data obtained as optical density values were expressed in arbitrary units per milliliter (AU/mL) related to the serial dilution of purified immunoglobulin.
Measurement of CRP, Fibrinogen, Intracellular Adhesion Molecule-1, and Interleukin-6
CRP was measured with a high-sensitivity automated rate nephelometric immunoassay (Dade-Behring high sensitivity CRP, BNII Nephelometer System). Fibrinogen was measured by the Clauss method on the Sigma Amax (Sigma, reagent from Diagnostica Stago, Asmieres-Sur-Seine, France). Intracellular adhesion molecule (ICAM)-1 and interleukin (IL)-6 were measured on a microplate with a quantitative sandwich EIA (Reagent Kit, R&D Systems).
Measurement of Chlamydia pneumoniae and Cytomegalovirus Antibodies
Serum was assayed for C pneumoniae IgG and IgA antibodies by micro-immunofluorescence (MIF) end-point titration, as described previously.19,20⇓ Analysis of IgG antibodies to cytomegalovirus (CMV) were done by EIA (DiaSorin Saluggia).
Baseline characteristics were listed as mean±SD for continuous variables and frequency table for discrete variable. Because anti-hsp and ACHA levels were not normally distributed, their levels are given as median and interquartile range. On the basis of our previous studies,11,21⇓ serum samples with levels of variables exceeding the 90th percentile were considered high. Difference between cases and controls was tested using McNemar test for categorical variables and paired t test or Wilcoxon test for continuous variables. Conditional logistic regression models were fitted to evaluate the association between the occurrence of a CV event (incident MI, stroke, or CV death) and the various antibodies, with adjustment for baseline characteristics. Spearman correlations were conducted to investigate the association between anti-hsp65, anti-hsp60, and anti-cholesterol (ACHA) antibody levels and the infectious or inflammatory markers. Joint effects of these markers on clinical outcomes were explored by testing their interaction terms in conditional logistic models. All analyses were performed in SAS version 8 (SAS/STAT Users Guide, Version 8).
Comparison of the Baseline Characteristics of Cases and Controls
Baseline characteristics of cases and controls are summarized in Table 1. Cases and controls were well matched for sex distribution, age, history of previous coronary artery disease, stroke, prevalence of type II diabetes, and hypertension. Fewer cases than controls were treated with ramipril than placebo, whereas current smoking and previous MI were significantly more common among cases than controls (Table 1).
Concentration of Anti-hsp Antibodies and ACHA in Cases and Controls
Median serum concentration of anti-hsp65 antibodies was significantly (P=0.02) higher in the baseline serum samples of cases than controls (Table 2). High levels (≥90th percentile) of anti-hsp65 antibodies occurred significantly more often among cases than controls (13.2% versus 6.6%, P=0.008), whereas median levels of anti-hsp60 antibodies and ACHA were similar in cases and controls (Table 2).
Association Between the Anti-hsp Antibodies and ACHA With Various Outcomes
The relationship of high levels of antibodies with the composite and individual outcomes of the study are shown in Table 3. High levels of anti-hsp65 antibodies predicted the composite outcome of incident MI, stroke, or CV death (OR, 2.1; 95% CI, 1.2 to 3.9; P=0.01). The results were consistent among those receiving ramipril or placebo. Analyzing types of outcome events separately, high anti-hsp65 levels predicted MI (OR, 2.3; 95% CI, 1.1 to 4.8; P=0.02) with a consistent trend for other outcomes (Table 3). In contrast, levels of anti-hsp60 antibodies did not predict any of the CV events.
Overall, ACHA levels were not associated with subsequent CV events. However, patients with high levels of ACHA had a significantly lower risk of developing stroke compared with those with normal serum concentration of ACHA (OR, 0.07; 95% CI, 0.01 to 0.97; P=0.047; see Table 3).
Correlation Between Anti-hsp Antibodies and ACHA With Inflammatory and Infectious Markers
In each serum sample, high-sensitivity CRP, fibrinogen, IL-6, sICAM-1, C pneumoniae IgG and IgA, and CMV IgG antibody levels were determined in parallel to anti-hsp and ACHA measurements (Table 4). A weak but highly significant positive correlation between IL-6 and anti-hsp65 and a very weak positive correlation of marginal significance was found between CRP and anti-hsp65. None of the other inflammatory and infectious markers correlated significantly to anti-hsp60, anti-hsp65, or ACHA levels. Anti-hsp60 and anti-hsp65 antibody levels significantly correlated with each other and serum concentration of ACHA; the correlation coefficients were, however, very weak, less than 0.20 at each comparison (Table 4).
Results of Combined Analysis
The Figure summarizes the joint predictive values of high (≥90th percentile) anti-hsp65 levels and high concentrations (highest tertile) of fibrinogen (Figure, A) or presence of CMV-IgG antibody (Figure, B). The latter two were both independent predictors of CV events in a multivariable Cox proportional hazards model among 3168 HOPE study patients (Smieja et al, unpublished data, 2002). The joint effects were evaluated using logistic regression analysis adjusted for clinical variables, which were significantly different in the univariate analysis (Table 1). As can be seen in the Figure, part A, the OR of the patients with high anti-hsp65 antibody and high plasma fibrinogen levels (highest tertile) for developing new myocardial infarction, stroke, or cardiovascular death was 5.5 (95% CI, 1.8 to 17.5, P=0.004) compared with patients with normal anti-hsp65 and fibrinogen concentrations. A similar additive effect was observed with the anti-hsp65 and anti-CMV antibodies as well, although the ORs for all outcomes were somewhat lower (Figure, B).
In the present study we investigated the predictive value of antibodies to heat shock protein and ACHA and the subsequent cardiovascular events in the HOPE study. High levels (≥90th percentile) of antibodies against mycobacterial hsp65, but not those of anti-human hsp60 or ACHA, predicted the development of new MI, stroke, or CV death. High ACHA levels had a protective effect against development of new stroke. Antibodies to hsp65 and hsp60 significantly correlated with each other and ACHA, but correlation coefficients were very weak. Previously11 we also measured low correlation coefficients (0.16) between anti-hsp60 and anti-hsp65 in severe CAD patients and found that these antibodies differ in their antigen specificity.17 High serum concentration of anti-hsp65 and fibrinogen and to a lesser extent high concentrations of anti-hsp65 and anti-CMV had a strong joint effect in predicting cardiovascular events.
High levels of anti-hsp65 were significantly associated with an increased risk of development of MI and the composite of cardiovascular events. Similar effects were observed for stroke or cardiovascular death individually, but these were not statistically significant. In line with this observation, elevated anti-hsp65 levels were observed by several groups in patients with severe CAD.9–11⇓⇓ However, we did not find a difference in the anti-hsp65 antibody levels between stroke patients and age-matched controls.22 Our present findings indicate that antibodies to hsp65 may facilitate the progression of established atherosclerotic vascular diseases, such as in the HOPE study subjects. These findings are in agreement with the results of Xu et al,8 who found that high anti-hsp65 levels predict carotid atherosclerotic progression during 5 years of follow-up. According to the hypothesis of Wick et al,23 the association between high levels of anti-hsp65 antibodies and atherosclerotic vascular diseases is attributable to an autoimmune reaction to endothelial cells that express high levels of hsp60 due to stress, such as oxidized LDL, free radicals, local infections, cytokines, or hemodynamic stress.24
In a separate analysis of 3168 patients from HOPE, high baseline plasma fibrinogen and high serum concentration of IgG antibodies against CMV were independent predictors of cardiovascular events. High serum anti-hsp65 antibody levels were associated with 2.1-fold risk (present study), whereas high fibrinogen levels were associated with 1.4-fold risk for development of new myocardial infarction, stroke, or cardiovascular death (Smieja et al, unpublished data, 2002). By contrast, subjects with combined high baseline anti-hsp65 antibody and high fibrinogen levels had an OR of 5.5 compared with patients with low levels of both variables (Figure). A similar but less marked joint effect was observed with anti-hsp65- and anti-CMV antibodies (Figure, B). These findings support the concept of the additive nature of different inflammatory and infectious risk factors of atherosclerotic vascular diseases.21,25⇓
We also studied the relationship of serum concentration of the anti-hsp antibodies to other markers of inflammation. No significant correlation was found between the levels of these antibodies and high levels of IgG and IgA antibodies to C pneumoniae. This finding is in complete agreement with recent findings.12,21,26⇓⇓ No correlation was found between serum concentration of anti-hsp60 or anti-hsp65 with the serum levels of other inflammatory markers such as CRP, fibrinogen, and sICAM. There was, however, a significant positive correlation between IL-6 and anti-hsp65 levels.
Recently, high levels of antibodies to human-hsp60 were found to be associated with CAD.11,12,21⇓⇓ In the present study, however, the serum concentration of autoantibodies to human hsp60 did not predict the development of new cardiovascular events. There are no published studies to assess the possible predictive value of IgG type anti-hsp60 antibodies in CAD. In two unpublished primary prevention trials, high levels of these autoantibodies were found to predict future coronary events. In the Second Northwick Park Heart Prospective Study, George Miller et al (Miller et al, unpublished data, 2001) found a 1.5 times (95% CI, 1.02 to 2.25, P=0.04) adjusted higher risk to develop coronary events during a follow-up period of 52 months in men with high levels of complement activating anti-hsp60 antibodies than in those with lower concentration. Huittinen et al,25 in 227 matched case-control pairs from the Helsinki Heart Study, found an association between high baseline anti-hsp60 IgA antibody levels and future coronary events during 8.5 years of follow-up (OR, 1.95; 95% CI, 1.12 to 3.40).
The lack of predictive value of anti-hsp60 measurement may be explained by the fact that more than half of patients in the present nested case-control substudy had a previous history of MI ever or in the year before study enrollment. Tissue necrosis could result in the release of high amounts of intracellular hsp60 antigen into the circulation, immune complex formation, and consequent decrease of serum concentrations of anti-hsp60 autoantibodies.10 Alternatively, it is possible that the lack of association between anti-hsp60 antibodies and subsequent CV events is the consequence of impaired formation of anti-hsp60 antibodies in patients with previous clinical events. Indeed, Schett et al27 demonstrated a strong downregulation of hsp60 humoral and cellular immune response after immunization of hsp65 primed rats with hsp60 containing coronary eluates. The difference between the results of Huittinen et al25 and our present findings can be attributable to the class (IgA versus IgG, respectively) of anti-hsp60 antibodies measured.
Autoantibodies to cholesterol are present in almost every healthy individual. They are natural antibodies, and their levels are independent of the amount of cholesterol in various lipoprotein fractions.14 These antibodies are different from anti-phospholipid antibodies.14 Recently we determined serum concentrations of IgG type ACHA by an ELISA in 600 patients with atherosclerotic vascular diseases. ACHA-IgG levels were lower in patients with peripheral occlusive atherosclerosis and ischemic stroke but considerably higher in CAD patients than in the age-matched controls.14 Low levels of ACHA in stroke patients could be a secondary phenomenon because of adsorption of ACHA to damaged cells in CNS or may be associated with the pathomechanism of ischemic stroke. The inverse association between high ACHA levels (Table 2) and new stroke may support the latter explanation. However, because few patients developed new stroke in the present study, any putative protective role of high ACHA levels for stroke development requires independent validation.
In summary, among patients with previous CV events or at high risk of such events, high serum concentration of antibodies to mycobacterial hsp65 had a higher risk of developing new CV events. This risk was even higher when combined with high levels of another inflammatory marker, fibrinogen, or with the presence of anti-CMV antibodies. Measurement of anti-hsp65 antibodies may improve the prediction of future CV events in secondary prevention studies and provide additional insight into the role of heat shock protein in human atherosclerosis.
This work was supported by the following grants: OTKA F029030 (Dr Prohászka), OTKA T032661 (Dr Füst), OTKA T029044 (Dr Karádi), FKFP 0138/2001 (Dr Prohászka), ETT 222/2000 (Dr Füst), and ETT 238/2000 (Dr Romics). Dr Prohászka is a Bolyai János Research Fellow, Dr Smieja was a Research Fellow of the Heart and Stroke Foundation of Canada, and Dr Yusuf is supported by a Senior Scientist award of the Canadian Institutes of Health Research and an endowed chair of the Heart and Stroke Foundation of Ontario.
- ↵Ridker PM. High-sensitivity C-reactive protein. potential adjunct for global risk assessment in the primary prevention of cardiovascular disease. Circulation. 2001; 103: 1813–1818.
- ↵Danesh J, Whincup J, Walker M, et al. Low grade inflammation and coronary heart disease: prospective study and updated meta-analysis. BMJ. 2000; 321: 199–204.
- ↵Torzewski J, Browyer DE, Waltenberger J, et al. Processes in atherogenesis: complement activation. Atherosclerosis. 1997; 132: 129–136.
- ↵Morimoto RI, Tissieres A, Georgopoulos G. The Biology of Heat-Shock Proteins and Molecular Chaperones. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1994.
- ↵Xu Q, Kiechl S, Mayr M, et al. Association of serum antibodies to heat-shock protein 65 with carotid atherosclerosis: clinical significance determined in a follow-up study. Circulation. 1999; 100: 1169–1174.
- ↵Zhu J, Arshed A, Rott D, et al. Antibodies to human heat-shock protein 60 are associated with the presence and severity of coronary artery disease. Circulation. 2001; 103: 1071–1075.
- ↵Prohászka Z, Duba J, Lakos G, et al. Antibodies against human hsp60 and mycobacterial hsp65 differ in their antigen specificity and complement activating ability. Int Immunol. 1999; 11: 1363–1370.
- ↵Fagerberg B, Gnarpe J, Gnarpe H, et al. Chlamydia pneumoniae but not cytomegalovirus antibodies are associated with future risk of stroke and cardiovascular disease: a prospective study in middle-aged to elderly men with treated hypertension. Stroke. 1999; 30: 299–305.
- ↵Burian K, Kis Z, Virok D, et al. Independent and joint effects of antibodies to human heat-shock protein 60 and Chlamydia pneumoniae infection in the development of coronary atherosclerosis. Circulation. 2001; 103: 1503–1508.
- ↵Kramer J, Harcos P, Prohászka Z, et al. Frequencies of certain complement protein alleles and serum levels of anti-heat-shock protein antibodies in cerebrovascular diseases. Stroke. 2000; 31: 2648–2652.
- ↵Wick G, Millonig G, Xu Qingbo. The autoimmune pathogenesis of atherosclerosis: an evolutionary Darwinian concept. In: Schoenfeld Y, Harats D, Wick G, eds. Atherosclerosis and Autoimmunity. Amsterdam: Elsevier; 2001: 5–16.
- ↵Hochleitner BW, Hochleitner EO, Obrist P, et al. Fluid shear stress induces heat shock protein 60 expression in endothelial cells in vitro and in vivo. Arterioscler Thromb Vasc Biol. 2000; 20: 617–623.
- ↵Huittinen T, Leinonen M, Tenkanen L, et al. Autoimmunity to human heat shock protein 60, Chlamydia pneumoniae infection and inflammation in predicting coronary risk. Arterioscler Thromb Vasc Biol. 2002; 22: 431–437.
- ↵Jantos CA, Krombach C, Wuppermann FN, et al. Antibody response to the 60-kDa heat-shock protein of Chlamydia pneumoniae in patients with coronary artery disease. J Infect Dis. 2000; 181: 1700–1705.
- ↵Schett G, Metzler B, Kleindienst R, et al. Myocardial injury leads to release of heat shock protein (hsp) 60 and a suppression of the anti-hsp65 immune response. Cardiovasc Res. 1999; 42: 685–695.