Local Atherosclerotic Plaques Are a Source of Prognostic Biomarkers for Adverse Cardiovascular Events
Objective— Atherosclerotic cardiovascular disease is a major burden to health care. Because atherosclerosis is considered a systemic disease, we hypothesized that one single atherosclerotic plaque contains ample molecular information that predicts future cardiovascular events in all vascular territories.
Methods and Results— AtheroExpress is a biobank collecting atherosclerotic lesions during surgery, with a 3-year follow-up. The composite primary outcome encompasses all cardiovascular events and interventions, eg, cardiovascular death, myocardial infarction, stroke, and endovascular interventions. A proteomics search identified osteopontin as a potential plaque biomarker. Patients undergoing carotid surgery (n=574) served as the cohort in which plaque osteopontin levels were examined in relation to their outcome during follow-up and was validated in a cohort of patients undergoing femoral endarterectomy (n=151). Comparing the highest quartile of carotid plaque osteopontin levels with quartile 1 showed a hazard ratio for the primary outcome of 3.8 (95% confidence interval, 2.6–5.9). The outcome did not change after adjustment for plaque characteristics and traditional risk factors (hazard ratio, 3.5; 95% confidence interval, 2.0–5.9). The femoral validation cohort showed a hazard ratio of 3.8 (95% confidence interval 2.0 to 7.4) comparing osteopontin levels in quartile 4 with quartile 1.
Conclusion— Plaque osteopontin levels in single lesions are predictive for cardiovascular events in other vascular territories. Local atherosclerotic plaques are a source of prognostic biomarkers with a high predictive value for secondary manifestations of atherosclerotic disease.
Advanced atherosclerotic cardiovascular disease continues to be a major burden to health care expenditures and requires exhaustive forms of medical treatment. A pressing need exists for prognostic biomarkers to identify high-risk patients for aggressive treatment.
Proteins in the plasma are easily accessible and can serve as a surrogate measure of atherosclerotic disease progression, but existing circulating biomarkers do not provide an accurate value of predictive patient risk.1,2 The main focus toward identifying patients with rapidly progressive advanced atherosclerotic disease is based on the known characteristics of the vulnerable or recently ruptured plaque with typically a large lipid core, thin fibrous cap, a high number of inflammatory cells, and thrombus.3–6 The pathological definition of the vulnerable plaque is founded on cross-sectional studies. Subsequently, molecular and cellular features associated with the vulnerable plaque are considered potential diagnostic imaging markers for plaque rupture and plaque thrombosis. However, longitudinal studies supporting the predictive power of these pathological markers have not been executed, and information about the natural history of atherosclerotic disease is therefore incomplete.
The systemic nature of atherosclerotic disease, however, is well-established7–9 through histopathologic observations demonstrating that inflammation,10 morphology,11 and lipid content12 correlate between different arterial segments within 1 individual. This gave rise to the hypothesis that local plaques contain molecular information that is predictive for atherothrombotic events in all vascular territories and that the local atherosclerotic plaque may act as a source of prognostic biomarkers that identify the patient at risk.
To provide evidence for the concept that single plaques contain molecular information that predicts future systemic events, the AtheroExpress biobank was initiated.13,14 In this longitudinal biobank study, we compared plaque proteins from patients who had a cardiovascular event to plaque proteins from patients who remained stable during follow-up using the complementary power of proteomics. As a proof of concept, expression of the identified local plaque protein osteopontin (OPN) was studied in all carotid samples and validated in femoral plaque samples to elucidate its predictive value for the occurrence of systemic cardiovascular events elsewhere in the body.
Materials and Methods
Study Population and Design
AtheroExpress is a longitudinal vascular biobank study that includes biomaterials from patients undergoing carotid endarterectomy and femoral endarterectomy in 2 Dutch hospitals (UMC Utrecht and St. Antonius Hospital, Nieuwegein). The primary objective of the study is to investigate the relation between single plaque characteristics at baseline and clinical outcome during follow-up. The study design has been described previously.13 The study has been approved by the Institutional Review boards of both hospitals and written informed consent was obtained from all patients.
All patients undergoing carotid endarterectomy between April 1, 2002, and March 1, 2006, in 1 of the 2 hospitals were considered for inclusion in the AtheroExpress study (n=685 carotid endarterectomy, n=200 femoral endarterectomy). Exclusion criteria for follow-up were unwillingness or physical incapability to participate (eg, severe dementia).
The criteria to perform carotid endarterectomy were based on the recommendations by the ACAS study15 for asymptomatic patients and the NASCET study16 for symptomatic patients. At baseline, clinical parameters including cardiovascular risk factors and medication use were recorded.
All patients underwent clinical follow-up 1 year after surgical intervention and completed postal questionnaires 1, 2, and 3 years after the operation. When patients did not respond to the questionnaire, the general practitioner was contacted by phone. Adjudication of the outcome events was performed by an outcome event committee consisting of 3 authors (F.L.M., J.P.d.e.V., W.E.H.) who were blinded to laboratory results. All end points were independently assessed by 2 members of the committee; in case of disagreement, a third opinion was obtained.
The primary outcome was a composite encompassing all cardiovascular events and interventions, including vascular death, nonfatal myocardial infarction, nonfatal stroke, and vascular intervention that was not planned at the time of inclusion.
Secondary outcome was any major cardiovascular event, such as vascular death, nonfatal myocardial infarction, nonfatal stroke, and nonfatal aneurysm rupture. Definitions and assessment procedures of the outcome events were described previously.13 This end point encompasses the same end points as the composite primary end point, excluding the peripheral surgical interventions.
Protein Extraction, Purification, and Proteomics
After dissection, plaques were cut in 0.5-cm segments and processed in a standardized way.13 For the proteomics analyses, plaque proteins were compared from carotid endarterectomy patients who had a primary outcome vs patients who were event-free. For this purpose, the first 80 subsequent patients in the cohort who had an event before March 2006 were selected in the event group. In the same cohort, 80 control patients were selected who matched for age, gender, and the duration of follow-up time after the event.
A detailed description of the proteomics is given in the online Appendix, available in the Data Supplement at http://atvb.ahajournals.org. In sum, after digestion, the peptides from each of the pooled samples were separated offline on high-performance liquid chromatography using a strong cation exchange column in 15 salt fractions. Subsequently, each salt fraction was injected onto the second dimension separation (C18 column) and sprayed online in the LTQ mass spectrometer (Thermo Scientific) for protein identification.
From the raw mass spectrometry data, Sequest in Bioworks 3.3 was used to generate a list of identified proteins per salt fraction that was collected in a Microsoft Excel file. Omniviz was used to combine the lists of identified proteins in the different salt fractions into 1 file with all different identified proteins (≈1200 identified proteins) identified per pooled sample and selected the identified proteins that were only in the event or only in the controls (Omniviz list). To select from this Omniviz list that included the proteins for validation in the individual patients, we used Ingenuity networks (Ingenuity Systems Inc). Ingenuity selected 90 proteins that had a previously described connection with the other proteins in the literature and visualized these connections in a large network.
Within this network, OPN was selected based on its previously described relation with atherosclerosis17–20 and the availability of an assay, as well as macrophage migration inhibitory factor (MIF).21,22
Osteopontin Enzyme-Linked Immunosorbent Assay
A commercial Osteopontin enzyme-linked immunosorbent assay (DOST00; R&D Systems) was used according to the described procedures. From each patient, 1 μg of Tripure (Roche) isolated protein was used per well.
From all patients who underwent carotid endarterectomy, plaques were used to assess the predictive value of plaque OPN levels for reaching the primary and secondary outcome events. Plaques obtained from patients undergoing femoral endarterectomy were used for external validation. Since part of the carotid artery samples had been used for proteomic analyses, a subgroup analysis was performed on those patients whose plaque material had not been used for the proteomic analyses.
Plaque OPN levels were divided into quartiles with cut-off values at 3.2, 8.2, and 14.9 ng/mL. In all analyses, quartiles 1 and 2 had similar risk implications. Kaplan-Meier survival analysis was used to estimate cumulative event rates after 3 years’ follow-up. Cox regression was used to calculate the hazard ratio (HR) with 95% confidence interval (95% CI) for the association between plaque characteristics and follow-up. In multivariable analysis, we adjusted simultaneously for those potential confounders that had a crude HR significant at the P<0.10 in univariate analysis. For the multivariate analysis, we included all patients and created a specific class of absent for those variables missing. Spearman test was used to compare OPN plaque and serum data
The study design is outlined in Figure 1. A total of 685 patients undergoing carotid endarterectomy were found to be eligible for inclusion. Forty patients were not willing to participate, and in 64 patients no plaque material was available for protein analyses. In the majority of these cases, the dissected plaque sample was formalin-fixed and paraffin-embedded for immunohistochemistry. Seven patients (1.2%) of the remaining 581 patients were lost to follow-up, resulting in a final cohort of 574 patients (Data Supplement Figure IA, available at http://atvb.ahajournals.org; Table 1). In the cohort undergoing femoral artery surgery, 200 patients were eligible, of whom 19 were excluded (17 not willing to participate, 2 with severe dementia). In the resulting 181 patients, 29 had insufficient plaque material and 1 was lost to follow-up, yielding a final cohort of 151 patients (Data Supplement Figure IB). Baseline characteristics did not differ between patients of the study cohort and patients with insufficient plaque material or who were lost to follow-up (data not shown).
The proteomics and bioinformatics search revealed a list of 90 candidate plaque proteins that could be upregulated in patients who had a cardiovascular event during follow-up. Next, we selected proteins that have been identified earlier in experimental set-ups with any cardiovascular phenotype but not necessarily with atherosclerosis (OPN17–20; MIF21). Then, we searched whether a detection tool was commercially available (eg, a commercial enzyme-linked immunosorbent assay or Luminex). OPN was selected for the current proof-of-concept study.
The mean plaque OPN levels differed consistently between plaques obtained from patients who had any cardiovascular event during follow-up (mean level, 15.0±1.0 ng/mL) and patients who had no event (mean level, 9.2±0.4 ng/mL). Mean OPN levels were 15.1±1.3 ng/mL for patients who had a secondary outcome, 15.5±1.6 ng/mL for those who had a stroke, 13.5±1.5 ng/mL for patients with a coronary event or intervention, and 15.2±1.6 ng/mL for peripheral interventions.
Table 2 shows the plaque OPN levels in association with the baseline clinical characteristics and plaque characteristics (HR, 1.45; 95% CI, 1.28–1.64) per 10 ng/mL increase of OPN (P<0.0005). The CI for OPN plaque levels are the 95% CI for the OPN plaque levels within each subgroup.
OPN as a Predictive Plaque Marker in Carotid Plaques
Differential expression of carotid plaque OPN between pooled carotid plaques of patients that had a secondary event and controls without a secondary event was validated in individual carotid plaque samples. Carotid plaque OPN levels (n=574) were strongly related with the occurrence of the primary outcome, and this relationship was stronger with higher levels of OPN. The HR was 2.1 (95% CI, 1.2–3.7) for quartile 3 vs quartile 1 (Table 3) and 3.8 for the highest quartile (quartile 4; 95% CI, 2.3–6.4; Table 3). Adjustment for potential confounders did not change these findings: the multiple adjusted HR for quartile 4 vs quartile 1 was 3.5 (95% CI, 2.0–5.9).
Osteopontin levels were also related with the occurrence of the secondary outcome (quartile 3 vs quartile 1, adjusted HR, 2.1; 95% CI, 1.0–4.5; quartile 4 vs quartile 1, adjusted HR ratio, 3.4 95% CI, 1.6–6.9).
Figure 2A and 2B illustrate the time-to-event curves for the primary and secondary outcome. Additional curves (Figure 2C–F) also demonstrate consistent outcomes for the other clinical outcomes (Table 3). Almost half of all patients in quartile 4 reached the primary end point within 3 years. Approximately one-quarter in quartile 4 had a major cardiovascular event during follow-up compared with 6% in quartiles 1 and 2. A post hoc analysis among patients who had asymptomatic carotid artery stenosis (n=118) showed a crude HR for the primary outcome of 7.1 (95% CI, 3.1–16.4) in quartile 4; the multiple adjusted HR was 7.1 (95%CI, 2.9–17.2).
External Validation: Predictive Value of OPN in Femoral Plaques
We validated these observations on basis of carotid plaque material in patients who had femoral endarterectomy. Mean follow-up time in this cohort was 2.3±0.9 years. Eighty-two patients (54%) reached the primary outcome during follow-up. The HR of quartile 3 vs quartile 1 was 2.8 (95% CI, 1.5–5.6), and it was 3.8 (95% CI, 2.0–7.4) for quartile 4. The time-to-event curve for the primary outcome in the femoral artery cohort is shown in Figure 2G (Table 3). Multivariable analyses did not materially change these findings (data not shown).
OPN Blood Levels: Predictive Value Of OPN in Serum
Having established that OPN plaque levels are associated with future cardiovascular events, we now measured serum OPN in an unbiased subgroup of our plaque cohort. Within the cohort of 574 carotid patients, we collected serum from 305 patients and measured OPN serum levels. This revealed (Table 4) that for OPN serum levels, the HR of quartile 3 vs quartiles 1 and 2 was 1.4 (95% CI, 0.8–2.7), and for quartile 4 vs quartiles 1 and 2 it was 2.2 (95% CI, 1.2–3.8). In the same group, however, HR was 2.3 for quartile 3 and 4.0 for quartile 4 for plaque OPN levels. Comparison of plaque and serum OPN levels within the patient revealed a correlation coefficient of 0.149 (P=0.009).
Our study indicates that high levels of OPN from atherosclerotic plaques strongly predict the risk of new vascular complications. These findings pertain to all parts of the vascular bed and were observed for carotid and femoral artery plaques. Osteopontin confirms our hypothesis that local plaques contain molecular information predictive of new atherothrombotic events elsewhere in the vascular tree independent of age and any other risk factor for atherosclerotic disease.
This is in line with previous observations in which the positive association between carotid artery intima-media thickness and cardiovascular risk factors7 and events8 has been established. In addition, asymptomatic carotid artery stenosis is an independent predictor of vascular events,9 and carotid plaque morphology is an independent predictor for noncerebral events.11
We identified a new concept and strategy in the search for progressive atherosclerotic disease biomarkers. The following observations support this view. First, following the notion that local plaques contain information on the stability of the entire atherosclerotic vascular system, we assumed that the biomarker should be detectable in every advanced atherosclerotic plaque independent of localization. We showed that in addition to carotid plaque OPN levels, femoral plaque OPN levels were also predictive for cardiovascular events. This strongly suggests that there is predictive information in every advanced plaque for future cardiovascular events in other vascular territories, independent of plaque localization. Second, the relation of plaque OPN with future adverse events was weak or independent of plaque features that are considered the hallmark of the vulnerable plaque, eg, presence of macrophages, large areas of lipid, and smooth muscle cell content. The current observations do not challenge the value of the current definitions of the vulnerable plaque for the understanding of pathogenesis but rather indicate that the local plaque hides strong molecular biomarkers for disease progression that might be independent of the pathological characterization. The weak association of OPN with a large lipid core and macrophage infiltration is probably attributable to the presence of OPN in only a subgroup of macrophages (Data Supplement Figure II). Third, the relation between OPN levels and the outcome measure was evident not only in symptomatic but also in asymptomatic patients. Expression levels were studied in advanced lesions in search for markers predictive for secondary clinical manifestations. Considering the promising results in asymptomatic patients, the next phase in this biomarker discovery approach could be a longitudinal study specifically addressing asymptomatic less advanced lesions. The high HR (7.1) for high OPN plaque levels in the asymptomatic patient group was confirmatory but surprising. Although supportive of our conclusions, this is a subgroup analyses with low patient numbers; therefore, results require verification to become conclusive.
Another substudy was performed on MIF plaque levels. MIF is studied in atherosclerotic models and is known to be involved in the recruitment of inflammatory cells and to induce matrix degrading enzymes like matrix metalloproteinases.21 This revealed (Data Supplement Figure III) that plaque MIF levels were strongly associated with secondary cardiovascular events and showed that the concept not only applies for OPN. Also, plaque proteins were found that were not associated with secondary events, indicating that predictive value is only found in a limited group of plaque proteins (Data Supplement Figures IV, V).
OPN was arbitrarily selected for this proof-of-concept biomarker study. The function of OPN has not been explored in this study. OPN, also known as early T-lymphocyte activator 1, is a secreted multifunctional glycoprotein.23 Mouse studies provided genetic evidence17–20 for a causal role of OPN in the development of atherosclerotic plaques. OPN plasma levels have shown to be related to coronary artery disease24,25 and to be an independent predictor of future cardiovascular events in patients with chronic stable angina26 with HR of 1.8. The predictive value of OPN plaque levels was higher than OPN levels in the serum.
Interestingly, we did not observe a strong correlation between blood and plaque OPN levels, which indicates that the plaque OPN levels really hide different biomarker properties compared with serological OPN. The predictive value of plaque OPN levels for future cardiovascular events has not been described yet. Furthermore, OPN plaque levels were not related to the Framingham risk score. Addition of Framingham risk score to the multivariable model did not change the association between plaque OPN and clinical outcome.
Several limitations of this study need to be discussed. Plaque phenotype is associated with the clinical disease status at baseline. Patients who had a stroke have more plaques with unstable characteristics compared to asymptomatic patients. Although our results were independent of clinical presentation, the heterogeneity of the population merits consideration. The results presented are limited to an elderly atherosclerotic patient population that makes extrapolation to a population that has traditional risk factors but without a clinical manifestation difficult. Because the plaque used for the local protein marker measurement associated with secondary events is removed from the patient, data regarding treatment are not collected during follow-up. The predictive value of OPN plaque levels in asymptomatic patients, however, remains high and suggests that these findings might apply to a broader population with advanced atherosclerotic plaques.
The proteomic study included pooled samples of 160 patients who were also part of the total group in which we analyzed OPN and MIF levels on an individual level. Therefore, the full cohort of 574 patients cannot be considered as a validation group. Outcome data of the 160 patients used in the pooled proteomics data did not differ from the newly added 414 patients (Data Supplement Table II). In addition, our results in the carotid patient group were supported by the data obtained in the patient group that underwent femoral endarterectomy.
Discovery of new plaque biomarkers that identify patients at risk for secondary clinical manifestations of advanced atherosclerotic disease brings new clinical implementations within reach. Imaging modalities like MRI, CT, SPECT, and ultrasound27,28 may apply these plaque markers to stratify patients at risk for secondary events by detection of plaque protein levels. Our observations also prove the concept that local plaque material may be used for prognostic screening and research because predictive plaque proteins can be detected in material of advanced plaques harvested after atherectomy, percutaneous coronary intervention, coronary artery bypass grafting,29 and other types of vascular surgery to further stratify these patient populations for secondary events. In conclusion, using OPN and MIF as an example, we show that local atherosclerotic plaques are a potential source of highly predictive biomarkers for the occurrence of adverse cardiovascular events in other vascular territories.
The authors thank A. Stubbs and S. Swagemakers for their contribution in bioinformatic analysis, L.W. Stanton, T.T. Aye, and T.Y. Low for their support in the proteomics, and W.M. Peeters for statistics and database analysis.
Sources of Funding
EU FP6-2004-Mobility-6 #021773 (D.d.K.), Ministry of Education of Singapore (ARC: T206B3211 to S.K.S.), EU FP6-037400 Immunath-2006 (to D.d.K. and G.P.), and the UMCU strategic Investment Prevention (to G.P.).
F. Moll, M. Daemen, P. van der Spek, G. Pasterkamp, and D. de Kleijn are founders and consultants of a recently established small biotech company with a limited amount of shares.
Received April 29, 2009; revision accepted November 23, 2009.
Shah PK, Falk E, Badimon JJ, Fernandez-Ortiz A, Mailhac A, Villareal-Levy G, Fallon JT, Regnstrom J, Fuster V. Human monocyte-derived macrophages induce collagen breakdown in fibrous caps of atherosclerotic plaques. Potential role of matrix-degrading metalloproteinases and implications for plaque rupture. Circulation. 1995; 92: 1565–1569.
Schwartz SM, Galis ZS, Rosenfeld ME, Falk E. Plaque rupture in humans and mice. Arterioscler Thromb Vasc Biol. 2007; 27: 705–713.
Goessens BM, Visseren FL, Kappelle LJ, Algra A, van der Graaf Y. Asymptomatic carotid artery stenosis and the risk of new vascular events in patients with manifest arterial disease: the SMART study. Stroke. 2007; 38: 1470–1475.
Mauriello A, Sangiorgi G, Fratoni S, Palmieri G, Bonanno E, Anemona L, Schwartz RS, Spagnoli LG. Diffuse and active inflammation occurs in both vulnerable and stable plaques of the entire coronary tree: a histopathologic study of patients dying of acute myocardial infarction. J Am Coll Cardiol. 2005; 45: 1585–1593.
Verhoeven BA, Velema E, Schoneveld AH, de Vries JP, de Bruin P, Seldenrijk CA, de Kleijn DP, Busser E, van der Graaf Y, Moll F, Pasterkamp G. Athero-express: differential atherosclerotic plaque expression of mRNA and protein in relation to cardiovascular events and patient characteristics. Rationale and design. Eur J Epidemiol. 2004; 19: 1127–1133.
Myers DL, Harmon KJ, Lindner V, Liaw L. Alterations of arterial physiology in osteopontin-null mice. Arterioscler Thromb Vasc Biol. 2003; 23: 1021–1028.
Matsui Y, Rittling SR, Okamoto H, Inobe M, Jia N, Shimizu T, Akino M, Sugawara T, Morimoto J, Kimura C, Kon S, Denhardt D, Kitabatake A, Uede T. Osteopontin deficiency attenuates atherosclerosis in female Apolipoprotein E-deficient mice. Arterioscler Thromb Vasc Biol. 2003; 23: 1029–1034.
Zernecke A, Bernhagen J, Weber C. Macrophage migration inhibitory factor in cardiovascular disease. Circulation. 2008; 117: 1594–1602.
Minoretti P, Falcone C, Calcagnino M, Emanuele E, Buzzi MP, Coen E, Geroldi D. Prognostic significance of plasma osteopontin levels in patients with chronic stable angina. Eur Heart J. 2006; 27: 802–807.
Saia F, Schaar J, Regar E, Rodriguez G, De Feyter PJ, Mastik F, Marzocchi A, Marrozzini C, Ortolani P, Palmerini T, Branzi A, van der Steen AF, Serruys PW. Clinical imaging of the vulnerable plaque in the coronary arteries: new intracoronary diagnostic methods. J Cardiovasc Med (Hagerstown). 2006; 7: 21–28.
Altwegg LA, Neidhart M, Hersberger M, Müller S, Eberli FR, Corti R, Roffi M, Sütsch G, Gay S, von Eckardstein A, Wischnewsky MB, Lüscher TF, Maier W. Myeloid-related protein 8/14 complex is released by monocytes and granulocytes at the site of coronary occlusion: a novel, early, and sensitive marker of acute coronary syndromes. Eur Heart J. 2007; 28: 941–948.