SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Limitations of the FRS
  4. Individual Biomarkers
  5. Multiple Biomarker Panels
  6. Imaging Biomarkers: Coronary Artery Calcium Scoring
  7. Conclusions
  8. References

Editor’s Note: The following Point/Counterpoint articles were derived from a debate presentation sponsored by the American Society for Preventive Cardiology at the March 2010 meeting of the American Heart Association Council on Epidemiology and Prevention, titled “Should We Focus on Novel Risk Marker and Screening Tests to Better Predict and Prevent Cardiovascular Disease?” Dr. James de Lemos presented the pro side, titled “Novel Risk Markers and Screening Tests Will Improve the Prediction and Prevention of Cardiovascular Disease,” and Dr. Donald Lloyd-Jones advocated the con side, titled “Better Implementation of Existing Knowledge Will Save More Lives Than All of the Novel Biomarkers in the World.” The following articles include points from the debate, rebuttal, and questions raised by the audience. We thank all authors for sharing this debate with the readership.

Despite many recent advances in the understanding of the pathophysiology of coronary heart disease (CHD), and the enthusiasm generated by novel genetic, proteomic, and imaging technologies, prediction of future CHD events remains largely reliant upon traditional risk factors. Initially, risk estimation involved simply totaling numbers of individual risk factors; however, more sophisticated scoring systems developed from multivariable regression models, such as the Framingham risk score (FRS),1 Systematic Coronary Risk Evaluation (SCORE) system,2 and QRisk (QRESEARCH cardiovascular risk algorithm),3 which led to improved predictive ability for CHD events. Prevention guidelines recommend using these scoring systems to estimate risk in the general population4 and using categories of risk to determine treatment goals (ie, for low-density lipoprotein cholesterol) in the primary prevention of CHD.5 This risk-based approach to cholesterol management targets pharmacologic therapy to those individuals at the highest relative risk for CHD. While this strategy is cost-efficient, many individuals are misclassified as low-risk on the basis of currently available risk assessment algorithms and thus do not receive preventive therapies that may favorably impact risk. With the development of novel biomarkers and imaging modalities, it is hoped that risk prediction might be enhanced so that earlier institution of therapies might be utilized to prevent adverse outcomes.

Limitations of the FRS

  1. Top of page
  2. Abstract
  3. Limitations of the FRS
  4. Individual Biomarkers
  5. Multiple Biomarker Panels
  6. Imaging Biomarkers: Coronary Artery Calcium Scoring
  7. Conclusions
  8. References

The FRS discriminates risk for future CHD events moderately well, with c-statistics typically ranging from 0.7 to 0.8.1,6 This performance clearly leaves room for improvement, and the FRS has several notable limitations (Table I). For example, data from the National Health and Nutritional Examination Survey (NHANES) illustrate that the majority of US adults fall into the low- and intermediate-risk groups as determined by the FRS.7 Although risk of CHD is obviously higher among individuals in the high-risk category, the absolute number of events that occur among low- and intermediate-risk individuals is estimated to be higher than among high-risk individuals, due to the large size of these subgroups and the imperfect classification of risk based on the FRS (Figure 1).8 By definition, individuals judged to be at low risk who subsequently experience a CHD event represent a failure of risk assessment and a potential missed opportunity for prevention.

Table I.   Limitations of Framingham Risk Score
Only modest discrimination for those at risk for future events (c-statistic generally 0.7–0.8)
Age dominates the Framingham risk score, thus marginalizing other potentially modifiable risk factors
Too few women classified at intermediate risk
Lifetime risk not adequately accounted for in risk estimation
Not well calibrated in all ethnic groups
image

Figure 1.  Percentage of 10-year coronary heart disease events by risk category. Percentages are based upon estimates from National Health and Nutrition Examination Survey data and average risk in each group. Int indicates intermediate. Adapted from reference 8.

Download figure to PowerPoint

Other limitations of the FRS are also notable. For example, in women, intermediate risk is seen only in those who are aged 55 years or older with three elevated cardiovascular risk factors; as a consequence, the potential exists for giving false reassurance to and undertreating a significant number of women at higher risk for cardiovascular disease (CVD) events.9 Lloyd-Jones et al.10 have described that many individuals stratified at low 10-year risk for cardiovascular events remain at substantial lifetime risk for cardiovascular events. Potentially, increasing time windows of risk prediction could reduce false reassurance of individuals who are at low short-term but high lifetime risk; however, this remains to be implemented into clinical practice. Moreover, even among individuals with no elevated risk factors at middle age, lifetime risk for CHD remains substantial.11

A very important additional limitation of traditional risk assessment algorithms is the dominant effect of age as compared to the other risk factors. This is exemplified in Cook’s analysis of the individual variables used in the Framingham model using the c-statistic. The c-statistic for age alone was 0.73, with little increment seen by adding other traditional risk factors (c-statistic with all traditional risk factors, 0.78) (Table II). Although the intent of this analysis was to demonstrate limitations of the c-statistic as a performance metric, these data highlight the challenges inherent in identifying new risk predictors in cohorts with wide age ranges.12

Table II.   Contributions of Individual Variables to CV Risk Prediction
Variableχ2RR per 2 SDC Statistic
  1. Abbreviations: HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; RR, relative risk; SD, standard deviation; SBP, systolic blood pressure; TC, total cholesterol. Adapted from reference 12.

Age3964.00.70
+SBP1492.50.74
+Smoking1212.90.73
+HDL-C860.50.73
+TC331.60.72
+LDL-C291.50.71
Framingham0.78
−TC391.60.77
−HDL640.60.77
−Smoking1002.60.76
−SBP1142.20.76
−Age2573.20.73

Individual Biomarkers

  1. Top of page
  2. Abstract
  3. Limitations of the FRS
  4. Individual Biomarkers
  5. Multiple Biomarker Panels
  6. Imaging Biomarkers: Coronary Artery Calcium Scoring
  7. Conclusions
  8. References

As standard risk factors alone fail to identify a large number of apparently low- and intermediate-risk individuals who will still experience CVD events, opportunities exist to develop and validate novel biomarkers and screening tools to improve risk assessment. Of the individual biomarkers studied to date, C-reactive protein (CRP) has certainly been the most widely studied and discussed. Unfortunately, results suggest only modest improvement in risk prediction when CRP is added to standard risk factor models. In the Reykjavik prospective study, 2459 patients with a history of nonfatal myocardial infarction or who died of CHD were compared to 3969 control participants; CRP was found to be only moderately associated with CHD (adjusted odds ratio [OR], 1.45; 95% confidence interval [CI], 1.25–1.68 for those in the highest tertile of CRP levels compared to those in the lowest tertile). This magnitude of risk associated with higher CRP levels was lower than that associated with increased total cholesterol concentration (adjusted OR, 2.35; 95% CI, 2.03–2.74) and cigarette smoking (OR, 1.87; 95% CI, 1.62–2.16).13 The association between CRP levels and risk of cardiovascular disease was further defined in a meta-analysis of 54 prospective studies, which included 160,309 individuals without a history of CVD who were followed for incidence of CHD.14 After adjustment for age and sex, a 1–standard deviation increase in CRP was associated with a relative risk (RR) of 1.68 (95% CI, 1.59–1.78) for CHD. After adjustment for conventional risk factors, the RR was reduced to 1.36 (95% CI, 1.22–1.52) and to 1.23 (95% CI, 1.07–1.42) after further adjustment for fibrinogen levels. The attenuation of the association after adjustment suggests that the predictive value of CRP may be largely dependent upon the contribution of conventional risk factors and partially explained by residual confounding by factors such as fibrinogen.

Although CRP has received most of the attention among potential new biomarkers for population screening, other biomarkers may actually have more value. For example, recent investigations have shown that B-type natriuretic peptides (BNP) may have greater predictive value than CRP as an individual biomarker. In a population-based cohort of 537 individuals without cardiovascular disease, N-terminal prohormone BNP (NT-proBNP) was more strongly associated with CVD than CRP, with an adjusted hazard ratio (HR) of 3.24 (95% CI, 1.80–5.79) for elevated NT-proBNP compared to 1.02 (95% CI, 0.56–1.85) for CRP.15 The predictive value of NT-proBNP was further evaluated by Rutten and colleagues16 in a community-based cohort of more than 5000 individuals without a history of CVD. Participants were followed for incident cardiovascular events, and after adjustment for conventional risk factors, hazard ratios of 2.32 (95% CI, 1.55–2.70) were observed for men and 3.08 (95% CI, 1.91–3.74) for women with elevated NT-proBNP levels. When NT-proBNP was added to a multivariable model containing traditional risk factors, a significant increase in the c-statistic was seen (from 0.66 to 0.69 in men and from 0.73 to 0.76 in women). Moreover, significant net reclassification improvement (NRI) of 0.092 (95% CI, 0.035–0.149; P=.001) in men and 0.133 (95% CI, 0.059–0.208%; P<.001) in women was also noted. Thus, NT-proBNP may show greater promise for predicting cardiovascular events compared to CRP. Given the strong associations of NT-proBNP with heart failure events, this biomarker would likely be of even more value for identifying individuals at risk for broader adverse cardiovascular disease outcomes than only CHD. As shown in a meta-analysis of 40 prospective studies with 87,474 individuals, participants in the highest tertile of BNP/NT-proBNP levels were found to have an HR of 2.82 (95% CI, 2.40–3.33) for CVD events (myocardial infarction, coronary death, and stroke) compared to those in the lowest tertile.17

Cardiac troponins are currently used predominantly in the hospital setting to diagnose myocardial infarction and to risk-stratify patients with suspected acute coronary syndromes. However, recent data suggest the possibility that troponin testing may be of value for population screening. In a population-based sample of 3557 participants from the Dallas Heart Study, the prevalence of troponin elevation in the general population was observed to be 0.7%.18 After multivariable regression, troponin elevation was found to correlate with existing cardiovascular disease (ie, left ventricular hypertrophy and congestive heart failure) or conditions associated with high CV risk (diabetes and chronic kidney disease) (Figure 2). Subsequently, Zethelius and colleagues19 reported results from a community-based study of 1203 men without cardiovascular disease at baseline who were followed for 10 years for first CHD event and all-cause mortality. After adjustment for conventional risk factors, elevated cardiac troponin I was found to be an independent predictor of death (HR, 1.26; 95% CI, 1.08–1.46; P=.003) and first CHD event (HR, 1.31; 95% CI, 1.11–1.54; P=.001). In another population-based cohort of 957 individuals from the Rancho Bernardo Study, troponin T and NT-proBNP levels were examined for prediction of CVD.20 Daniels and colleagues reported that those with elevated troponin T levels (≥0.01 ng/mL) had an increased risk of cardiovascular death (HR, 2.06; 95% CI, 1.03–4.12, P=.040); as previously described, elevated NT-proBNP was also found in this study to predict an increased risk of cardiovascular mortality (adjusted HR per unit-log increase in NT-proBNP, 2.51; 95% CI, 1.55–4.08; P<.001).20 It is possible that application of more sensitive assays for cardiac troponins that are in development, which have recently been shown to strongly predict risk in patients with chronic heart failure21 and CHD,22 will demonstrate even more robust performance for population screening than currently available assays.

image

Figure 2.  Prevalence of cardiac troponin elevation in the general population. In the Dallas Heart Study, prevalence of troponin elevation increased with number of risk determinants present. Inset table shows odds ratios for elevated troponin level ≥0.01 μg/dL. LV indicates left ventricular; cTnT indicates cardiac tropinin T. Adapted from reference 18.

Download figure to PowerPoint

Multiple Biomarker Panels

  1. Top of page
  2. Abstract
  3. Limitations of the FRS
  4. Individual Biomarkers
  5. Multiple Biomarker Panels
  6. Imaging Biomarkers: Coronary Artery Calcium Scoring
  7. Conclusions
  8. References

With individual biomarkers having only moderate predictive value, further studies have investigated the impact of combining multiple biomarkers as a means of improving upon traditional risk factors and individual biomarkers. In a secondary analysis of 3199 patients in the Heart Outcomes Prevention Evaluation (HOPE) study, Blankenberg and colleagues investigated the incremental benefit of adding several biomarkers to traditional risk factor models for the prediction of cardiovascular death, myocardial infarction, and stroke over a 4.5-year follow-up period.23 Measured biomarkers included acute-phase reactants (CRP, fibrinogen, and interleukin 6), proinflammatory markers (soluble tumor necrosis factor receptor-1 and -2, soluble interleukin 1 receptor antagonist, and interleukin 18), and mediators of endothelial activation (soluble vascular adhesion molecule-1 and soluble intercellular adhesion molecule-1), NT-proBNP, and microalbuminuria. Although they hypothesized that a multimarker approach incorporating multiple inflammatory biomarkers would substantially improve risk prediction, it was found that the only biomarker that added significant incremental value was NT-proBNP (Figure 3).

image

Figure 3.  Addition of N-terminal prohormone B-type natriuretic peptide (NT-proBNP) to conventional risk factor model adds incremental predictive value. ROC indicates receiver operating characteristic. Adapted from reference 23.

Download figure to PowerPoint

Also investigating the utility of incorporating multiple biomarkers into a risk prediction model, Wang and associates24 studied 3209 individuals from the Framingham Heart Study. Biomarkers evaluated included CRP, BNP, N-terminal pro atrial natriuretic peptide, aldosterone, renin, fibrinogen, D-dimer, plasminogen-activator inhibitor type 1, homocysteine, and urinary albumin-to-creatinine ratio. After a mean follow-up period of 7.4 years, 169 individuals experienced the outcome of first major cardiovascular event. After adjustment for conventional risk factors, BNP was again associated with increased cardiovascular risk (adjusted HR, 1.25 per 1–standard deviation increment in the log values), as was urinary albumin-to-creatinine ratio (1.20). A “multimarker” score was created based upon weighted values of biomarkers selected from a backward elimination multivariable model: individuals in the highest quintile for the biomarker score had increased risk of death (adjusted HR, 4.08; P<.001) and major cardiovascular events (adjusted HR, 1.84; P=.02) compared with those with scores in the lowest twoi quintiles. However, the addition of multimarker scores to conventional risk factors resulted in only small increases in the c-statistic (0.80 without biomarkers vs 0.82 with biomarkers). Although this was interpreted as a “negative” study, limitations of the c-statistic have been well-described,12 and it is plausible that reevaluation of these data using novel statistical metrics, including those assessing clinical reclassification,25 may demonstrate clinical utility of the multimarker score, given the relatively robust associations with cardiovascular mortality observed.

Although the studies above demonstrated only marginal success for the multimarker panels, other studies have demonstrated greater potential utility. This is likely due to the inclusion in these studies of biomarkers with more robust individual predictive value, such as NT-proBNP and troponin. A recent study by Zethelius and colleagues studied the benefit of adding several biomarkers (troponin I, NT-proBNP, CRP, and cystatin C) to a conventional risk factor model for prediction of cardiovascular death. Compared to the use of traditional risk factors alone, the addition of these selected biomarkers significantly improved discrimination for cardiovascular death (c-statistic with vs without biomarkers, 0.77 vs 0.66; P<.001).26 One explanation for the more favorable results of this study is that the biomarkers selected for this panel simply represent better individual biomarkers than those studied previously, which is plausible because they largely reflect existing cardiac (troponin I, NT-proBNP) or renal (cystatin C) disease, rather than nonspecific inflammatory processes. This study also highlights a notable issue regarding the contribution of age to risk prediction models. Age as a variable tends to dominate risk prediction and consequently the ability to evaluate other variables (including biomarkers) may be limited. The approach utilized in this study of enrolling participants within a restricted age range (all participants were aged approximately 71 years) may serve as a methodological model for isolating the contribution of other variables to cardiovascular risk.27

Newly emerging biomarkers continue to be investigated, as are newer statistical metrics for evaluating these markers. This was demonstrated by Melander and colleagues in a recent investigation of contemporary biomarkers using methods of clinical reclassification based upon findings from multivariable regression models.28 This study of 5067 participants included measurement of novel markers such as lipoprotein-associated phospholipase-2, midregional proadrenomedullin (MR-proADM), midregional pro-atrial natriuretic peptide, as well as previously described markers of CRP, cystatin C, and NT-proBNP. Multivariable regression analysis was performed to determine the association with first cardiovascular events (myocardial infarction, stroke, coronary death) during the 12.8-year follow-up period. Conventional risk factor models had c-statistics of 0.76 (95% CI, 0.73–0.78) for total cardiovascular events and 0.76 (95% CI, 0.73–0.79) for coronary events. After adjustment, CRP and NT-proBNP marginally improved predictive value for cardiovascular events (increased the c-statistic by 0.007; P=.04). A modest proportion of individuals were reclassified (8% for cardiovascular risk, 5% for coronary risk), but net reclassification improvement was nonsignificant for cardiovascular events and coronary events. Again, this study highlights that the clinical utility of a multiple marker panels is largely determined by the specific biomarkers that are selected for inclusion, the patient population studied, and how well the baseline risk factor model performs in that population. For example, the panels perform less well in populations with broad age ranges than in those with more narrow age ranges. In addition, this study illustrates that measurement of predictive utility can vary depending on the statistical metrics used to assess the value of the biomarkers.27

Most recently, in a large-scale effort by Blankenberg and colleagues, 30 novel biomarkers were evaluated for association with cardiovascular risk in 10,466 men and women in European cohorts (Figure 4).29 There were 538 fatal and nonfatal cardiovascular events noted after a 10-year follow-up period. Using multivariable regression, a score was derived from biomarkers with the strongest association with CVD, including NT-proBNP (HR for 1-SD increase, 1.23), CRP (HR, 1.23), and troponin I (HR, 1.18). The addition of the biomarker score to a conventional risk factor model improved 10-year risk estimation for cardiovascular events, as evidenced by several metrics, including the c-statistic, and more important from a clinical standpoint, correct reclassification of individuals to higher- or lower-risk categories. The NRI for the biomarker panel in this study was estimated at 0.11 (P<.001).

image

Figure 4.  Biomarkers evaluated in the MORGAM cohorts. Reproduced from reference 29.

Download figure to PowerPoint

Imaging Biomarkers: Coronary Artery Calcium Scoring

  1. Top of page
  2. Abstract
  3. Limitations of the FRS
  4. Individual Biomarkers
  5. Multiple Biomarker Panels
  6. Imaging Biomarkers: Coronary Artery Calcium Scoring
  7. Conclusions
  8. References

In addition to circulating proteins, biomarkers of CHD may also include imaging modalities, such as cardiac computed tomography (CT) and magnetic resonance imaging (MRI). Indeed, clinical applications of imaging have progressed much more rapidly than genomics and proteomics. One such “imaging biomarker” is coronary artery calcium (CAC) scoring using CT. Initial studies evaluating CAC screening were limited by referral bias,30 small sample size31 and soft endpoints,32 although more recent studies have evaluated CAC scoring in larger, nonreferral population-based cohorts. Detrano and associates33 published data from the Multi-Ethnic Study of Atherosclerosis (MESA), describing the results of coronary calcium scanning in a population-based sample of 6722 individuals without clinical cardiovascular disease, who were followed for a mean of 3.8 years. A doubling of the calcium score was found to increase the risk of major coronary events by 15% to 35% and to increase the risk of any coronary event by 18% to 39%. The c-statistic was significantly higher when the calcium score was added to traditional risk factors for the prediction of both major coronary events (0.79 vs 0.83; P=.006) and for any coronary event (0.77 vs 0.82; P<.001) (Figure 5).

image

Figure 5.  Coronary artery calcium score and incidence of coronary events. Adapted from reference 33.

Download figure to PowerPoint

Polonsky and colleagues further evaluated the predictive value of CAC scoring, from the same MESA cohort, using contemporary metrics of net reclassification improvement. The NRI for the overall population was 0.25 (95% CI, 0.16–0.34; P=.001) and for the intermediate risk subgroup was 0.55 (95% CI; 0.41–0.69; P<.001).34 Moreover, most of the correct reclassification observed in the overall population was upward reclassification to higher risk categories rather than downward reclassification to lower risk categories. In contrast, much of the reclassification observed for other biomarkers has been downward reclassification. This is important since it is not yet clear that it is safe to defer preventive therapies for individuals classified downward to a lower risk category on the basis of any of these novel risk assessment tools. The NRI for CAC screening (0.25) compares favorably to the NRIs reported for CRP, which have ranged from 0.047 to 0.09 in different cohorts.35,36 Consequently, this imaging modality may show greater promise for risk prediction compared to the nonimaging biomarkers.

Conclusions

  1. Top of page
  2. Abstract
  3. Limitations of the FRS
  4. Individual Biomarkers
  5. Multiple Biomarker Panels
  6. Imaging Biomarkers: Coronary Artery Calcium Scoring
  7. Conclusions
  8. References

In summary, traditional risk factors used alone or in combination do not predict risk well enough, leaving a significant number of “unpredicted” CVD events occurring in individuals classified as low-risk by standard classification methods. Simply put, the status quo is not sufficient and the need for novel risk assessment tools is substantial. We must be cautious not to become complacent with currently imperfect risk prediction tools and also to guard against nihilism during the “growing pains” of evaluation of novel tools.

Novel biomarkers should continue to be explored to improve risk prediction, but we believe the standards should be set extremely high before integrating any biomarkers into clinical practice.37 None of the discussed individual biomarkers are yet ready for widespread use in clinical practice for population screening. CRP provides only small improvements in risk prediction and is clearly not involved in the pathophysiologic processes leading to CHD.38 NT-proBNP and troponin (with high-sensitivity assays) deserve further investigation, as preliminary data suggest that they outperform CRP for risk assessment in the population. The stronger data for these biomarkers, as well as coronary calcium screening, argue that biomarkers of existing cardiac disease are likely to prove of greater predictive value than are nonspecific inflammatory markers. Combining multiple biomarkers may improve risk prediction, provided that each of the individual biomarkers provides unique pathophysiologic information and substantial individual value, but studies to date clearly show that when multiple mediocre biomarkers are combined the composite product is also mediocre. Moreover, these multimarker panels will likely not be ready for clinical application for years. CAC scoring shows promise compared to protein biomarkers, and combinations of protein biomarkers together with CAC may be even more useful.

Moving forward, it is imperative that the research agenda continue the exploration for novel markers of CVD that will improve the identification of those at risk. In the pursuit of discovering these markers, many more failures than successes will occur; nevertheless, these failures must not leave us resigned to accept the inadequate present state in which persons at risk remain undertreated based upon misclassification of clinical risk.

References

  1. Top of page
  2. Abstract
  3. Limitations of the FRS
  4. Individual Biomarkers
  5. Multiple Biomarker Panels
  6. Imaging Biomarkers: Coronary Artery Calcium Scoring
  7. Conclusions
  8. References
  • 1
    Wilson PW, D’Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories. Circulation. 1998;97:18371847.
  • 2
    Conroy RM, Pyorala K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J. 2003;24:9871003.
  • 3
    Hippisley-Cox J, Coupland C, Vinogradova Y, et al. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: prospective open cohort study. BMJ. 2007;335:136.
  • 4
    Redberg RF, Benjamin EJ, Bittner V, et al. ACCF/AHA 2009 performance measures for primary prevention of cardiovascular disease in adults: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Performance Measures (Writing Committee to Develop Performance Measures for Primary Prevention of Cardiovascular Disease) developed in collaboration with the American Academy of Family Physicians; American Association of Cardiovascular and Pulmonary Rehabilitation; and Preventive Cardiovascular Nurses Association: endorsed by the American College of Preventive Medicine, American College of Sports Medicine, and Society for Women’s Health Research. J Am Coll Cardiol. 2009;54:13641405.
  • 5
    Grundy SM, Cleeman JI, Merz CN, et al. Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III guidelines. Circulation. 2004;110:227239.
  • 6
    Orford JL, Sesso HD, Stedman M, et al. A comparison of the Framingham and European Society of Cardiology coronary heart disease risk prediction models in the normative aging study. Am Heart J. 2002;144:95100.
  • 7
    Ford ES, Giles WH, Mokdad AH. The distribution of 10-year risk for coronary heart disease among US adults: findings from the National Health and Nutrition Examination Survey III. J Am Coll Cardiol. 2004;43:17911796.
  • 8
    Centers for Disease Control and Prevention (CDC). National Center for Health Statistics (NCHS). National Health and Nutrition Examination Survey Data. Hyattsville, MD: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention; 2010.
  • 9
    Vasan RS, Sullivan LM, Wilson PW, et al. Relative importance of borderline and elevated levels of coronary heart disease risk factors. Ann Intern Med. 2005;142:393402.
  • 10
    Lloyd-Jones DM, Wilson PW, Larson MG, et al. Framingham risk score and prediction of lifetime risk for coronary heart disease. Am J Cardiol. 2004;94:2024.
  • 11
    Berry JD, Garside DB, Cai X, et al. Remaining lifetime risks for cardiovascular disease death by risk factor burden at selected ages in black and white men and women. Circulation. 2007;116:832.
  • 12
    Cook NR. Use and misuse of the receiver operating characteristic curve in risk prediction. Circulation. 2007;115:928935.
  • 13
    Danesh J, Wheeler JG, Hirschfield GM, et al. C-reactive protein and other circulating markers of inflammation in the prediction of coronary heart disease. N Engl J Med. 2004;350:13871397.
  • 14
    Kaptoge S, Di Angelantonio E, Lowe G, et al. C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet. 2010;375:132140.
  • 15
    Kistorp C, Raymond I, Pedersen F, et al. N-terminal pro-brain natriuretic peptide, C-reactive protein, and urinary albumin levels as predictors of mortality and cardiovascular events in older adults. JAMA. 2005;293:16091616.
  • 16
    Rutten JH, Mattace-Raso FU, Steyerberg EW, et al. Amino-terminal pro-B-type natriuretic peptide improves cardiovascular and cerebrovascular risk prediction in the population: the Rotterdam study. Hypertension. 2010;55:785791.
  • 17
    Di Angelantonio E, Chowdhury R, Sarwar N, et al. B-type natriuretic peptides and cardiovascular risk: systematic review and meta-analysis of 40 prospective studies. Circulation. 2009;120:21772187.
  • 18
    Wallace TW, Abdullah SM, Drazner MH, et al. Prevalence and determinants of troponin T elevation in the general population. Circulation. 2006;113:19581965.
  • 19
    Zethelius B, Johnston N, Venge P. Troponin I as a predictor of coronary heart disease and mortality in 70-year-old men: a community-based cohort study. Circulation. 2006;113:10711078.
  • 20
    Daniels LB, Laughlin GA, Clopton P, et al. Minimally elevated cardiac troponin T and elevated N-terminal pro-B-type natriuretic peptide predict mortality in older adults: results from the Rancho Bernardo Study. J Am Coll Cardiol. 2008;52:450459.
  • 21
    Latini R, Masson S, Anand IS, et al. Prognostic value of very low plasma concentrations of troponin T in patients with stable chronic heart failure. Circulation. 2007;116:12421249.
  • 22
    Omland T, De Lemos JA, Sabatine MS, et al. A sensitive cardiac troponin T assay in stable coronary artery disease. N Engl J Med. 2009;361:25382547.
  • 23
    Blankenberg S, McQueen MJ, Smieja M, et al. Comparative impact of multiple biomarkers and N-Terminal pro-brain natriuretic peptide in the context of conventional risk factors for the prediction of recurrent cardiovascular events in the Heart Outcomes Prevention Evaluation (HOPE) Study. Circulation. 2006;114:201208.
  • 24
    Wang TJ, Gona P, Larson MG, et al. Multiple biomarkers for the prediction of first major cardiovascular events and death. N Engl J Med. 2006;355:26312639.
  • 25
    Cook NR. Statistical evaluation of prognostic versus diagnostic models: beyond the ROC curve. Clin Chem. 2008;54:1723.
  • 26
    Zethelius B, Berglund L, Sundstrom J, et al. Use of multiple biomarkers to improve the prediction of death from cardiovascular causes. N Engl J Med. 2008;358:21072116.
  • 27
    De Lemos JA, Rohatgi A. Separating the contenders from the pretenders: competitive high-throughput biomarker screening in large population-based studies. Circulation. 2010;121:23812383.
  • 28
    Melander O, Newton-Cheh C, Almgren P, et al. Novel and conventional biomarkers for prediction of incident cardiovascular events in the community. JAMA. 2009;302:4957.
  • 29
    Blankenberg S, Zeller T, Saarela O, et al. Contribution of 30 biomarkers to 10-year cardiovascular risk estimation in 2 population cohorts: the MONICA, risk, genetics, archiving, and monograph (MORGAM) biomarker project. Circulation. 2010;121:23882397.
  • 30
    Wong ND, Detrano RC, Diamond G, et al. Does coronary artery screening by electron beam computed tomography motivate potentially beneficial lifestyle behaviors? Am J Cardiol. 1996;78:12201223.
  • 31
    Janowitz WR, Agatston AS, Viamonte M Jr. Comparison of serial quantitative evaluation of calcified coronary artery plaque by ultrafast computed tomography in persons with and without obstructive coronary artery disease. Am J Cardiol. 1991;68:16.
  • 32
    Budoff MJ, Georgiou D, Brody A, et al. Ultrafast computed tomography as a diagnostic modality in the detection of coronary artery disease: a multicenter study. Circulation. 1996;93:898904.
  • 33
    Detrano R, Guerci AD, Carr JJ, et al. Coronary calcium as a predictor of coronary events in four racial or ethnic groups. N Engl J Med. 2008;358:13361345.
  • 34
    Polonsky TS, McClelland RL, Jorgensen NW, et al. Coronary artery calcium score and risk classification for coronary heart disease prediction. JAMA. 2010;303:16101616.
  • 35
    Wilson PW, Pencina M, Jacques P, et al. C-reactive protein and reclassification of cardiovascular risk in the Framingham Heart Study. Circ Cardiovasc Qual Outcomes. 2008;1:9297.
  • 36
    Cook NR. Comments on ‘Evaluating the added predictive ability of a new marker: From area under the ROC curve to reclassification and beyond’ by MJ Pencina et al., Statistics in Medicine (DOI: 10.1002/sim.2929). Stat Med. 2008;27:191195.
  • 37
    Morrow DA, De Lemos JA. Benchmarks for the assessment of novel cardiovascular biomarkers. Circulation. 2007;115:949952.
  • 38
    Shah SH, De Lemos JA. Biomarkers and cardiovascular disease: determining causality and quantifying contribution to risk assessment. JAMA. 2009;302:9293.