Author + information
- Received July 22, 2009
- Revision received October 26, 2009
- Accepted November 9, 2009
- Published online March 16, 2010.
- Claire L. Heslop, BMSc* (, )
- Jiri J. Frohlich, MD and
- John S. Hill, PhD
- ↵*Reprint requests and correspondence:
Ms. Claire L. Heslop, ASL/Healthy Heart, St. Paul's Hospital, 1081 Burrard Street, Vancouver, British Columbia V6Y 1Y6, Canada
Objectives We evaluated the relative and combined value of oxidative stress biomarkers for predicting cardiovascular mortality in patients undergoing selective coronary angiography.
Background Oxidative stress participates in all stages of cardiovascular disease, from lipoprotein modification to plaque rupture, and biomarkers of oxidative stress predict development of coronary artery disease (CAD). Oxidative stress biomarkers merit investigation for the value they may offer for long-term cardiovascular risk prediction.
Methods Myeloperoxidase (MPO), nitrotyrosine, oxidized low-density lipoprotein, and antioxidant capacity were measured in a prospective cohort of 885 selective coronary angiography patients followed up for >13 years for cardiovascular mortality.
Results MPO independently predicted CAD, and top tertile MPO levels predicted a 2.4-fold risk of cardiovascular mortality (95% confidence interval [CI]: 1.47 to 2.98), compared with patients with lowest tertile MPO levels. MPO also improved risk model discrimination and patient risk category classification. Elevations in multiple oxidative stress biomarkers predicted increased mortality risk; however, the strongest risk prediction was achieved by assessing MPO and C-reactive protein (CRP) together. Patients with either MPO or CRP elevated had 5.3-fold higher cardiovascular mortality risk (95% CI: 1.86 to 14.9), and patients with high levels of both MPO and CRP had a 4.3-fold risk compared with patients with only elevated marker (95% CI: 2.26 to 8.31). These results remained significant with adjustment for cardiovascular risk factors and baseline disease burden.
Conclusions MPO accurately predicted cardiovascular mortality risk in coronary angiography patients. Considering MPO and CRP together may improve long-term risk assessment and CAD patient outcomes.
Patients with stable coronary artery disease (CAD) require ongoing effective risk stratification; however, there remain significant clinical limitations to current treatment strategies (1). Biomarkers that add incremental data to cardiovascular risk prediction have been studied with the goal of enhancing approaches to long-term risk reduction (2).
Based on evidence that oxidative stress is involved in all stages of atherosclerosis (3), oxidative stress biomarkers have been investigated for their potential clinical value. Oxidative modifications of low density lipoproteins and cellular lipids contribute to vascular endothelium dysfunction (4), monocyte invasion (5), foam cell formation (6), and plaque instability (7). Pro-oxidant enzyme myeloperoxidase (MPO) is released by activated neutrophils and macrophages. MPO has been found to predict cardiovascular disease development (8,9) and myocardial infarction in patients with chest pain (10). Nitrotyrosine (N-tyr), a protein residue modification generated by MPO and reactive nitrogen processes, has been localized to atherosclerotic plaques (11), and predicts presence of CAD (12). Elevated plasma oxidized low-density lipoprotein (oxLDL) predicts CAD and severity of acute coronary syndromes (13). Despite this compelling evidence, the prognostic utility of oxidative stress biomarkers for long-term risk assessment in stable CAD has not yet been explored. Finally, circulating antioxidant enzymes and molecules may protect against oxidative stress. Indeed, reduced plasma antioxidant capacity (AOC) is associated with coronary artery calcification (14), but has not yet been demonstrated to predict disease development or outcome.
Oxidative stress biomarkers that identify instability in CAD may be useful for screening stable CAD patients to identify individual patients requiring more aggressive interventions. Improved risk prediction would permit uptitration of medical therapy and improved compliance in higher-risk patients, and a biomarker that could further guide treatment strategies would optimize patient outcome.
On the basis of the preceding evidence, we evaluated whether plasma oxidative stress biomarkers, separately and in combination, predict future fatal cardiovascular events in a large, prospective cohort of nonemergent coronary angiography patients.
Coronary angiography cohort
Between 1992 and 1995, a cohort of 1,117 patients (797 men and 320 women) was recruited from consecutive selective coronary angiography subjects at 2 Vancouver teaching hospitals. Indications for angiography included stable angina and previous myocardial infarction (MI), as well as aortic stenosis and/or regurgitation, and mitral valve regurgitation. Patients with unstable angina and/or MI within the preceding 2 months were excluded (n = 98). Analyses of this cohort by investigators in our research group are reported elsewhere (15,16). Lesions visualized in major epicardial vessels were assessed semiquantitatively for percent stenosis, rounded to the nearest 10%. Presence of CAD was defined by the presence of any lesion ≥20% stenosis, and severe CAD was defined by presence of any lesion ≥50% stenosis.
Data regarding cigarette smoking history (past, current, or never), self-identified ethnicity, previous diagnosis of hypertension, previous diagnosis of diabetes mellitus, and history of premature CAD in patient's family (males before 45 years of age and females before 55 years of age), were obtained by self-report. Blood pressure was also measured, and hypertension classified by pre-existing diagnosis and/or baseline blood pressure ≥140/90 mm Hg. Height and weight were measured, and body mass index calculated. Waist circumference was measured midway between the iliac crest and the bottom of the ribcage. Medication use was recorded from patient's charts. All participants gave written informed consent, and the Research Ethics Board of St. Paul's Hospital, Vancouver, approved this research.
Identifying data were linked to the British Columbia Vital Statistics Agency mortality database to determine deaths to the end of 2007. Cardiovascular deaths were identified by World Health Organization International Classification of Disease-10th revision mortality codes I20 to I25 and I60 to I69 (17).
Fasting blood samples were collected before angiography in ethylenediamine tetra-acetate Vacutainer tubes (Becton Dickinson, Franklin Lakes, New Jersey). After centrifugation, plasma was aliquoted and stored at −70°C, with a single freeze-thaw cycle before biomarker analyses. Lipid and lipoprotein measurements were completed immediately, and inflammatory marker C-reactive protein (CRP) was measured in 2002, using high-sensitivity methods, as we have described previously (16).
The MPO and N-tyr were measured using solid-phase enzyme-linked immunosorbent assays (Hycult, Uden, the Netherlands). The oxLDL was measured using a competitive enzyme-linked immunosorbent assay (Mercodia, Uppsala, Sweden), based on monoclonal antibody 4E6. The AOC was measured by a colorimetric assay (Alpco, Salem, New Hampshire) on the basis of reaction of sample antioxidants with hydrogen peroxide. Assay performance was assessed by linearity studies using dilutions within assay linear ranges, and samples were diluted to obtain measurements within assay linear ranges: MPO 0.4 to 100 ng/ml; N-tyr 2 to 1,500 nmol/l; oxLDL 0.5 to 7.6 U/l; and AOC 130 to 393 μmol/l. Intra-assay and interassay coefficients of variation were as follows: MPO (7.2%, 7.8%), N-tyr (7.9%, 11.1%), oxLDL (5.4%, 4.2%), and AOC (7.4%, 7.5%). All measurements were performed blinded to patient data, and in duplicate. Stability of biomarkers in stored plasma was verified by comparing assay performance (coefficients of variation and linearity) with results from an angiography cohort collected more recently in our laboratory (n ≥20 for comparisons).
Normal range of biomarkers
For the purpose of characterizing biomarker assay performance, a group of 78 patients (37 women and 41 men) was selected to represent the “healthiest” controls. These patients reported no history of cardiovascular disease, hypertension, diabetes, or renal insufficiency, and they had no lesions >10% stenosis at baseline coronary angiography.
Baseline continuous variables are presented as mean ± SE, skewed variables as median (interquartile range [IQR]), and categorical variables as number (percentage). The MPO, N-tyr, AOC, and CRP had skewed distribution and were therefore log transformed. Baseline characteristics were tested for relationships with angiographic CAD and cardiovascular mortality using Student ttests for continuous variables, Mann-Whitney rank-sum tests for skewed continuous variables, and chi-square statistics for categorical variables. Logistic regression models were used to calculate odds ratios (ORs) and corresponding confidence intervals (CIs) for associations between biomarkers and angiographic CAD, with adjustment for the following cardiovascular risk factors: age, sex, total cholesterol/high-density lipoprotein cholesterol ratio (TC:HDL-C), body mass index, smoking, diabetes, and hypertension. These covariates were chosen for demonstrated relationships with cardiovascular risk and oxidative stress pathways.
Biomarker tertiles were defined within the cohort and tested for association with mortality by log-rank tests. Biomarkers associated with mortality were further tested by Cox proportionate hazards analyses, with adjustment for risk factors listed above, with further adjustment for CRP and severe CAD (≥50% stenosis), where noted. The proportionate hazards assumption was confirmed by correlating Schoenfeld residuals with time, and partial residual plots.
To evaluate biomarkers for risk discrimination, c-statistics from time-adjusted receiver-operating characteristic curves were generated from survival models with and without biomarkers (18). Then c-statistics were compared using nonparametric methods for comparing curves generated from the same patients (19). Risk model calibration was also assessed by Hosmer-Lemeshow statistics (20).
Finally, survival models were assessed for accuracy of patient risk estimation, using cross-classification tables (21). Patient outcome risk was calculated from covariate adjusted models, classified into categories of ≤5%, 5% to 10%, 10% to 20%, and ≥20% risk. Rates of patient reclassification with biomarkers were calculated, and significance of model improvement was determined by the net reclassification index (NRI), as described by Pencina et al. (22). Although they do not adjust for right censoring, NRI analyses permit risk models to be compared for how accurately they predict an outcome at a given follow-up time.
Data were analyzed using SPSS version 14.0 (SPSS Inc., Chicago, Illinois), and R version 2.8.1 (R Foundation for Statistical Computing, Vienna, Austria), with 2-tailed p values ≤0.05 considered statistically significant.
Biomarkers and angiographic CAD
Of the patients recruited, 885 had samples available for biomarker measurement. Of those patients, 651 had angiographic CAD, and 604 had severe lesions (≥50% stenosis). Baseline characteristics are presented in Table 1.Patients with severe lesions had higher MPO (p = 0.022), oxLDL (p = 0.031), and AOC (p = 0.003). The oxLDL and AOC were not associated with CAD in age- and sex-adjusted risk models. However, MPO predicted angiographic CAD (OR per SD: 1.97, 95% CI: 1.08 to 3.60; p = 0.03) and severe CAD (OR per SD: 1.85, 95% CI: 1.09 to 3.18; p = 0.021) in covariate adjusted analyses. Oxidative stress biomarker levels among the 78 healthiest control patients were as follows: median MPO 76.9 (IQR 55.3 to 124.7), N-tyr 76.6 (IQR 66.4 to 98.9), oxLDL 65.7 (IQR 54.3 to 82.1), and AOC 287.9 (IQR 263.8 to 314.0).
Biomarkers and baseline characteristics
Female patients had lower AOC; median was 280.2 mol/l (IQR 256.1 to 303.7 mol/l) for women versus 302.7 mol/l (IQR 277.7 to 325.7 mol/l) for men (p < 0.001), which supports findings reported elsewhere (23). No other sex differences were observed for biomarkers. No ethnicity differences were tested, as 87% of patients reported Caucasian ethnicity.
Antioxidant capacity correlated with body mass index (r = 0.149; p < 0.001), as did oxLDL (r = 0.173; p < 0.001), which supports previous research (24). Higher oxLDL levels were observed in patients with a smoking history versus nonsmokers (73.4 mU/l [IQR 61.8 to 91.0 mU/l] vs. 70.9 mU/l [IQR 58.7 to 85.6 mU/l]; p = 0.03); however, these patients also had higher AOC (301.3 mol/l [IQR 274.3 to 328.4 mol/l] vs. 295.2 mol/l [IQR 268.6 to 314.7 mol/l]; p = 0.01). CRP correlated with MPO (r = 0.140; p < 0.001), as observed elsewhere (25).
Patients with angiographic CAD were more likely to report statin use at the time of initial assessment (19% vs. 5%; p < 0.001); however, statin users did not have different rates of cardiovascular mortality (hazard ratio [HR]: 0.96 [95% CI: 0.59 to 1.67]). Although patients with angiographic disease were not more likely to report using antihypertensive medications at baseline, use of angiotensin-converting enzyme inhibitors predicted cardiovascular mortality (HR: 2.54 [95% CI: 1.73 to 3.74]; p < 0.001). No other class of medication was associated with mortality, and no differences in biomarkers were observed according to medication use at baseline.
Biomarkers and cardiovascular mortality
After a median follow-up time of 12.9 years (IQR 11.1 to 13.1 years), there were 257 deaths, and 117 were cardiovascular deaths. Patients who died of cardiovascular causes were older (p < 0.001), and more likely to have diabetes (p = 0.026). Also, they had higher TC:HDL-C (p < 0.001), CRP (p < 0.001), and MPO (p = 0.001).
Risk for cardiovascular mortality increased across tertiles of MPO (p = 0.007) and N-tyr (p = 0.029) (Fig. 1),with significant differences between highest and lowest tertiles appearing within the first year for MPO, and after 4 years for N-tyr. Risk for cardiovascular mortality among patients with severe CAD also increased across tertiles of MPO (p = 0.029) and N-tyr (p = 0.035). Adjustment for covariates attenuated relationships between N-tyr and cardiovascular mortality (Table 2).
The unadjusted HR for cardiovascular mortality for the top MPO tertile was 2.38 (95% CI: 1.47 to 2.98) compared with the bottom tertile (p < 0.001). The risk remained with covariate adjustment (HR: 1.96, 95% CI: 1.15 to 3.37; p = 0.012). With further adjustment for severe CAD (≥50% stenosis) the risk remained high (HR: 2.06, 95% CI: 1.26 to 3.36; p = 0.004).
Finally, the risk for elevated MPO remained significant with further adjustment to include CRP (HR: 1.75, 95% CI: 1.16 to 3.10; p = 0.010), and although CRP attenuated the risk for the second tertile, the trend remained linear (p = 0.036). Survival analyses for MPO are displayed in Table 3.
Left ventricular ejection fraction (LVEF) values were collected from coronary angiography reports for a subset of angiography patients (n = 415). Decreased LVEF was associated with cardiovascular mortality (p < 0.001), but LVEF did not correlate with oxidative stress biomarkers. >In covariate adjusted survival analyses including adjustment for CAD severity, CRP, and LVEF, risk associated with the highest MPO tertile remained strong (HR: 2.88, 95% CI: 1.18 to 7.03; p = 0.02), and the trend remained linear across tertiles (p = 0.030).
In covariate-adjusted survival analyses including adjustment for CAD severity, CRP, and use of angiotensin-converting enzyme inhibitors at baseline, risk associated with the highest MPO tertile remained strong (HR: 2.02, 95% CI: 1.22 to 3.34; p = 0.001), and the trend remained linear (p = 0.022).
Combined biomarkers and cardiovascular mortality risk
Oxidative stress scores, assigned by adding the number of pro-oxidant markers (MPO, N-tyr, and oxLDL) in the highest tertile with a score for AOC in the lowest tertile, predicted cardiovascular mortality in a linear fashion (p = 0.010). Patients with the highest scores (3 and 4, n = 89) had a 2.8-fold cardiovascular mortality risk (95% CI: 1.53 to 5.39; p = 0.001) compared with patients who had the lowest multimarker score (n = 262). This relationship persisted with adjustment for covariates and severity of CAD (p = 0.037 for linear trend). However, the trend was attenuated by adjustment for MPO, suggesting multiple markers offered no significant benefit.
Combined value of MPO and CRP
The CRP tertiles were defined within the cohort as <1.04, 1.04 to 3.30, and >3.30 mg/l. MPO not only predicted cardiovascular mortality independent of CRP, but also combining MPO and CRP tertiles predicted cardiovascular mortality risk (p < 0.001). Patients who had either marker elevated had a 5.33-fold risk of cardiovascular mortality compared with patients who had both biomarkers in the lowest tertile (95% CI: 1.86 to 14.9; p = 0.002). Patients who had both markers elevated had a 4.33-fold risk of cardiovascular mortality compared with patients who had either biomarker elevated (95% CI: 2.26 to 8.31; p < 0.001). Cumulative survival curves for this comparison are displayed in Figure 2.
These observations remained significant after covariate adjustment, including adjustment for either CRP or MPO levels. Among patients with severe CAD, covariate adjusted HR was 3.83 for elevation of 1 marker (95% CI: 1.33 to 11.02; p = 0.013), and 6.41 for both markers (95% CI: 2.18 to 18.78; p = 0.001) compared with low levels of both markers. Further adjustment for LVEF did not attenuate the relationship.
Evaluation of survival models
The covariate adjusted model including CAD severity yielded an area under the curve of 0.715, which improved to 0.761 with MPO (p = 0.031), and to 0.781 with combined MPO and CRP tertiles (p = 0.004). No deviations from risk calibration were detected for any risk models.
Finally, covariate adjusted models (including CAD severity) were used to classify patients into risk categories for cardiovascular mortality across the entire study duration. The NRI was calculated from patient reclassification rates conferred by biomarkers. Patients reclassified by MPO are displayed in Table 4.The MPO tertiles improved risk prediction across all categories, with a NRI of 14.4% (p = 0.003). Combining MPO and CRP tertiles also improved patient risk classification over risks estimated by covariates and CRP alone (NRI 9.6%; p = 0.05).
MPO has been localized to human atherosclerotic lesions, and implicated in proatherogenic modifications to lipoprotein particles (26). Also, catabolism of nitric oxide by MPO may impair its vasodilatory and anti-inflammatory functions in the vasculature (27). Indeed, elevated MPO independently predicts reduced endothelial function (28), whereas CRP has not been consistently found to do so (29,30). In this prospective study of 885 coronary angiography patients, baseline MPO independently predicted cardiovascular mortality beyond traditional risk factors and degree of angiographic stenosis. Furthermore, MPO enhanced survival model discrimination and improved patient risk classification across meaningful risk categories. Although multiple markers of oxidative stress incrementally influenced cardiovascular mortality risk, the greatest value was offered by evaluation of MPO. Finally, MPO and CRP had independent and combined predictive value, and improved accuracy of patient risk estimations.
MPO has previously been demonstrated to identify patients with acute coronary syndromes and MI who are at higher risk for adverse outcomes (31,32); however, one study found risk conferred by MPO was only for events during the initial 72 h after onset of symptoms (33). Although elevations of MPO secondary to recruitment and activation of neutrophils in ACS may explain these results (34), MPO may also be induce plaque rupture by activating metalloproteases (35) and triggering endothelial cell apoptosis (36). Level of MPO is also elevated in patients with heart failure (37), and may have value for heart failure screening, in combination with CRP (38). We demonstrate for the first time the value of MPO for long-term risk estimation for patients with stable CAD, in comparison with traditional risk factors and CRP. Multiple oxidative stress markers have been associated with CAD in a smaller, cross-sectional study (39); however, our study is the first prospective investigation of multiple oxidative stress biomarkers in angiography patients. Our findings extend and refine the current evidence implicating oxidative stress in cardiovascular disease, and we clarify the relative prognostic utility of oxidative stress biomarkers compared with conventional risk factors.
Not all studies are consistent with our findings. Kubala et al. (25) did not find elevated MPO in CAD patients, and a recent study of 382 stable CAD patients found MPO did not predict total mortality (40). However, we followed up our larger cohort of patients for a longer time, and determined cause for all cases of mortality. Also, we included patients with moderate lesions, which are more often implicated in MI (41). By investigating patients across a range of risk, we elucidated the prognostic value of MPO, and identified that MPO offers the improvements across meaningful risk categories—including intermediate categories for which more challenging therapeutic decisions must be made, and for which interventional countermeasures such as revascularization may reduce cardiovascular mortality risk. Most important, we demonstrate for the first time the complementary value of MPO and CRP for identifying patients at risk for cardiovascular mortality.
We could not ascertain biomarker changes over time, nor could we adjust for baseline renal function, a major influence on oxidative stress levels. Because of limitations inherent to our samples, we were also not able to measure and compare our findings with every emerging cardiovascular and oxidative stress biomarker. Certain data regarding medication use, interventions, and other cardiovascular events after initial assessment are also not available.
Although there is good evidence from multiple studies that MPO has value for risk assessment in cardiovascular cohorts, the transportability of our results to a more diverse cohort or to lower risk populations has yet to be shown. Finally, the research grade assays used for these analyses are not yet approved for clinical application, and interpretation of results from different assays must be done with caution.
The oxidative stress biomarker MPO powerfully enhances cardiovascular risk prediction, and is additive with CRP for predicting risk in selective angiography patients. Future studies are essential to demonstrate whether outcomes may be improved when MPO and CRP measurements are utilized together for prognostic assessment and therapeutic decision making.
This study was supported by a grant from Heart and Stroke Foundation of British Columbia and Yukon, Vancouver, British Columbia, Canada. Recruitment of study participants was funded by the Canadian Institutes of Health Research, Ottawa, Ontario, Canada.
- Abbreviations and Acronyms
- antioxidant capacity
- coronary artery disease
- confidence interval
- C-reactive protein
- hazard ratio
- interquartile range
- left ventricular ejection fraction
- net reclassification index
- oxidized low-density lipoprotein
- total cholesterol/high-density lipoprotein cholesterol ratio
- Received July 22, 2009.
- Revision received October 26, 2009.
- Accepted November 9, 2009.
- American College of Cardiology Foundation
- Quinn M.T.,
- Parthasarathy S.,
- Fong L.G.,
- Steinberg D.
- Hazell L.J.,
- Stocker R.
- Meuwese M.C.,
- Stroes E.S.,
- Hazen S.L.,
- et al.
- Ehara S.,
- Ueda M.,
- Naruko T.,
- et al.
- Valabhji J.,
- McColl A.J.,
- Richmond W.,
- Schachter M.,
- Rubens M.B.,
- Elkeles R.S.
- Lee K.W.,
- Hill J.S.,
- Walley K.R.,
- Frohlich J.J.
- ↵(1992) International Statistical Classification of Disease and Related Health Problems, Tenth Revision (ICD-10) (World Health Organization, Geneva).
- Hosmer D.,
- Lemeshow S.
- Cook N.R.
- Eiserich J.P.,
- Baldus S.,
- Brennan M.L.,
- et al.
- Vita J.A.,
- Brennan M.L.,
- Gokce N.,
- et al.
- Prasad A.,
- Zhu J.,
- Halcox J.P.,
- Waclawiw M.A.,
- Epstein S.E.,
- Quyyumi A.A.
- Mocatta T.J.,
- Pilbrow A.P.,
- Cameron V.A.,
- et al.
- Baldus S.,
- Heeschen C.,
- Meinertz T.,
- et al.
- Fu X.,
- Kassim S.Y.,
- Parks W.C.,
- Heinecke J.W.
- Sugiyama S.,
- Kugiyama K.,
- Aikawa M.,
- Nakamura S.,
- Ogawa H.,
- Libby P.
- Little W.C.,
- Constantinescu M.,
- Applegate R.J.,
- et al.