[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.158.119.60. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Download PDF
Table 1. 
Number of CVD Events During 8 Years and Participants at Risk*
Number of CVD Events During 8 Years and Participants at Risk*
Table 2. 
Characteristics of the Study Participants*
Characteristics of the Study Participants*
Table 3. 
Relative Risks for New Major CHD and Major and Total CVD Associated With Traditional Risk Factors*
Relative Risks for New Major CHD and Major and Total CVD Associated With Traditional Risk Factors*
Table 4. 
Age- and Sex-Adjusted CRP Level and Relative Risk for Major CHD and Major and Total CVD Incidence
Age- and Sex-Adjusted CRP Level and Relative Risk for Major CHD and Major and Total CVD Incidence
Table 5. 
CRP Level and Multivariable Adjusted Relative Risks for New Major CHD and Major and Total CVD*
CRP Level and Multivariable Adjusted Relative Risks for New Major CHD and Major and Total CVD*
Table 6. 
C Statistics for Discrimination of CHD and CVD Outcomes With Various Predictive Models
C Statistics for Discrimination of CHD and CVD Outcomes With Various Predictive Models
Table 7. 
CRP Tertile and Calculated 10-Year Risk of Vascular Events According to CHD Risk Level Determined With Traditional Risk Factors*
CRP Tertile and Calculated 10-Year Risk of Vascular Events According to CHD Risk Level Determined With Traditional Risk Factors*
1.
Wilson  PWD’Agostino  RBLevy  DBelanger  AMSilbershatz  HKannel  WB Prediction of coronary heart disease using risk factor categories. Circulation 1998;971837- 1847
PubMedArticle
2.
D’Agostino  RB  SrGrundy  SSullivan  LMWilson  PCHD Risk Prediction Group, Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA 2001;286180- 187
PubMedArticle
3.
Assmann  GCullen  PSchulte  H Simple scoring scheme for calculating the risk of acute coronary events based on the 10-year follow-up of the prospective cardiovascular Munster (PROCAM) Study. Circulation 2002;105310- 315
PubMedArticle
4.
Assmann  GSchulte  HOberwittler  W New aspects in the prediction of coronary artery disease: the Prospective Cardiovascular Munster Study. Fidge  NHNestel  PJedsAtherosclerosis VII Amsterdam, the Netherlands Elsevier Science Publishers1986;19- 24
5.
Haffner  SMLehto  SRonnemaa  TPyorala  KLaakso  M Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med 1998;339229- 234
PubMedArticle
6.
Greenland  PKnoll  MDStamler  J  et al.  Major risk factors as antecedents of fatal and nonfatal coronary heart disease events. JAMA 2003;290891- 897
PubMedArticle
7.
Pearson  TAMensah  GAAlexander  RW  et al.  Markers of inflammation and cardiovascular disease: application to clinical and public health practice: a statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation 2003;107499- 511
PubMedArticle
8.
Wilson  PW Metabolic risk factors for coronary heart disease: current and future prospects. Curr Opin Cardiol 1999;14176- 185
PubMedArticle
9.
Mosca  L C-reactive protein—to screen or not to screen [editorial]? N Engl J Med 2002;3471615- 1617
PubMedArticle
10.
Lipid Research Clinics Program, Manual of Laboratory Operation.  Bethesda, Md National Institutes of Health1974;
11.
Warnick  GRBenderson  JAlbers  JJ Dextran sulfate-Mg2+ precipitation procedure for quantitation of high-density-lipoprotein cholesterol. Clin Chem 1982;281379- 1382
PubMed
12.
Rost  NSWolf  PAKase  CS  et al.  Plasma concentration of C-reactive protein and risk of ischemic stroke and transient ischemic attack: the Framingham Study. Stroke 2001;322575- 2579
PubMedArticle
13.
Wang  TJNam  BHWilson  PW  et al.  Association of C-reactive protein with carotid atherosclerosis in men and women: the Framingham Heart Study. Arterioscler Thromb Vasc Biol 2002;221662- 1667
PubMedArticle
14.
Cupples  LAD’Agostino  RBKiely  D The Framingham Heart Study, Section 35: An Epidemiological Investigation of Cardiovascular Disease: Survival Following Cardiovascular Events: 30-Year Follow-up.  Bethesda, Md National Heart Lung and Blood Institute1988;
15.
Harrell  FE  JrLee  KLMark  DB Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15361- 387
PubMedArticle
16.
Pencina  MJD’Agostino  RB Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;232109- 2123
PubMedArticle
17.
Redberg  RFRifai  NGee  LRidker  PM Lack of association of C-reactive protein and coronary calcium by electron beam computed tomography in postmenopausal women: implications for coronary artery disease screening. J Am Coll Cardiol 2000;3639- 43
PubMedArticle
18.
Folsom  ARAleksic  NCatellier  DJuneja  HSWu  KK C-reactive protein and incident coronary heart disease in the Atherosclerosis Risk In Communities (ARIC) study. Am Heart J 2002;144233- 238
PubMedArticle
19.
Danesh  JCollins  RAppleby  PPeto  R Association of fibrinogen, C-reactive protein, albumin, or leukocyte count with coronary heart disease: meta-analyses of prospective studies. JAMA 1998;2791477- 1482
PubMedArticle
20.
Danesh  JWheeler  JGHirschfield  GM  et al.  C-reactive protein and other circulating markers of inflammation in the prediction of coronary heart disease. N Engl J Med 2004;3501387- 1397
PubMedArticle
21.
Koenig  WSund  MFrohlich  M  et al.  C-reactive protein, a sensitive marker of inflammation, predicts future risk of coronary heart disease in initially healthy middle-aged men: results from the MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) Augsburg Cohort Study, 1984 to 1992. Circulation 1999;99237- 242
PubMedArticle
22.
Ridker  PMWilson  PWGrundy  SM Should C-reactive protein be added to metabolic syndrome and to assessment of global cardiovascular risk? Circulation 2004;1092818- 2825
PubMedArticle
23.
Pradhan  ADManson  JERifai  NBuring  JERidker  PM C-reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA 2001;286327- 334
PubMedArticle
24.
Folsom  ARPankow  JSTracy  RP  et al.  Association of C-reactive protein with markers of prevalent atherosclerotic disease. Am J Cardiol 2001;88112- 117
PubMedArticle
25.
Cushman  MMeilahn  ENPsaty  BMKuller  LHDobs  ASTracy  RP Hormone replacement therapy, inflammation, and hemostasis in elderly women. Arterioscler Thromb Vasc Biol 1999;19893- 899
PubMedArticle
26.
Visser  MBouter  LMMcQuillan  GMWener  MHHarris  TB Elevated C-reactive protein levels in overweight and obese adults. JAMA 1999;2822131- 2135
PubMedArticle
27.
Albert  MADanielson  ERifai  NRidker  PMPRINCE Investigators, Effect of statin therapy on C-reactive protein levels: the Pravastatin Inflammation/CRP Evaluation (PRINCE): a randomized trial and cohort study. JAMA 2001;28664- 70
PubMedArticle
28.
Albert  MAGlynn  RJRidker  PM Alcohol consumption and plasma concentration of C-reactive protein. Circulation 2003;107443- 447
PubMedArticle
29.
Esposito  KPontillo  ADi Palo  C  et al.  Effect of weight loss and lifestyle changes on vascular inflammatory markers in obese women: a randomized trial. JAMA 2003;2891799- 1804
PubMedArticle
30.
Ockene  ISMatthews  CERifai  NRidker  PMReed  GStanek  E Variability and classification accuracy of serial high-sensitivity C-reactive protein measurements in healthy adults. Clin Chem 2001;47444- 450
PubMed
31.
Ridker  PMGlynn  RJHennekens  CH C-reactive protein adds to the predictive value of total and HDL cholesterol in determining risk of first myocardial infarction. Circulation 1998;972007- 2011
PubMedArticle
32.
Ridker  PMHennekens  CHBuring  JERifai  N C-reactive protein and other markers of inflammation in the prediction of cardiovascular disease in women. N Engl J Med 2000;342836- 843
PubMedArticle
33.
Ridker  PM High-sensitivity C-reactive protein: potential adjunct for global risk assessment in the primary prevention of cardiovascular disease. Circulation 2001;1031813- 1818
PubMedArticle
34.
Albert  MAGlynn  RJRidker  PM Plasma concentration of C-reactive protein and the calculated Framingham Coronary Heart Disease Risk Score. Circulation 2003;108161- 165
PubMedArticle
35.
Blake  GJRifai  NBuring  JERidker  PM Blood pressure, C-reactive protein, and risk of future cardiovascular events. Circulation 2003;1082993- 2999
PubMedArticle
36.
Ridker  PMBuring  JECook  NRRifai  N C-reactive protein, the metabolic syndrome, and risk of incident cardiovascular events: an 8-year follow-up of 14 719 initially healthy American women. Circulation 2003;107391- 397
PubMedArticle
37.
Ballantyne  CMHoogeveen  RCBang  H  et al.  Lipoprotein-associated phospholipase A2, high-sensitivity C-reactive protein, and risk for incident coronary heart disease in middle-aged men and women in the Atherosclerosis Risk in Communities (ARIC) study. Circulation 2004;109837- 842
PubMedArticle
38.
Greenland  PSmith  JS  JrGrundy  SM Improving coronary heart disease risk assessment in asymptomatic people: role of traditional risk factors and noninvasive cardiovascular tests. Circulation 2001;1041863- 1867
PubMedArticle
39.
Wilson  PWSmith  SC  JrBlumenthal  RSBurke  GLWong  ND 34th Bethesda Conference, Task Force #4—how do we select patients for atherosclerosis imaging? J Am Coll Cardiol 2003;411898- 1906
PubMedArticle
40.
Ridker  PMCook  N Clinical usefulness of very high and very low levels of C-reactive protein across the full range of Framingham Risk Scores. Circulation 2004;1091955- 1959
PubMedArticle
41.
Brennan  MLPenn  MSvan Lente  F  et al.  Prognostic value of myeloperoxidase in patients with chest pain. N Engl J Med 2003;3491595- 1604
PubMedArticle
42.
Rosamond  WDChambless  LEFolsom  AR  et al.  Trends in the incidence of myocardial infarction and in mortality due to coronary heart disease, 1987 to 1994. N Engl J Med 1998;339861- 867
PubMedArticle
43.
Chambless  LEFolsom  ARSharrett  AR  et al.  Coronary heart disease risk prediction in the Atherosclerosis Risk in Communities (ARIC) study. J Clin Epidemiol 2003;56880- 890
PubMedArticle
44.
Greenland  PLaBree  LAzen  SPDoherty  TMDetrano  RC Coronary artery calcium score combined with Framingham score for risk prediction in asymptomatic individuals. JAMA 2004;291210- 215[published correction appears in JAMA. 2004;291:563]
PubMedArticle
45.
Raggi  PCooil  BCallister  TQ Use of electron beam tomography data to develop models for prediction of hard coronary events. Am Heart J 2001;141375- 382
PubMedArticle
Original Investigation
November 28, 2005

C-Reactive Protein and Risk of Cardiovascular Disease in Men and Women From the Framingham Heart Study

Author Affiliations

Author Affiliations: General Clinical Research Center, Medical University of South Carolina, Charleston (Dr Wilson); Department of Mathematics (Drs Nam, Pencina, and D’Agostino) and Department of Cardiology, School of Medicine (Dr Benjamin), Boston University, Boston, Mass; and the National Heart, Lung, and Blood Institute’s Framingham Heart Study, National Heart, Lung, and Blood Institute, Framingham, Mass (Dr O’Donnell).

Arch Intern Med. 2005;165(21):2473-2478. doi:10.1001/archinte.165.21.2473
Abstract

Background  Determination of C-reactive protein (CRP) level has been suggested to improve cardiovascular disease (CVD) risk assessment. This study examines the utility of CRP levels to assess CVD risk in a community setting.

Methods  We performed a prospective observational cohort study on a community population sample. A total of 1949 men and 2497 women without CVD from the Framingham Heart Study underwent CVD risk factor assessment. Initial CVD events during 8 years of follow-up were recorded.

Results  There were 283 major CVD and 160 major coronary heart disease incident events. Age-, sex-, and multivariable-adjusted analyses generally used CRP level categories of less than 1, 1 to 3, and greater than 3 mg/L. In age- and sex-adjusted models, the traditional risk factors and elevated CRP levels indicated increased risk. The age- and sex-adjusted relative risk (RR) and 95% confidence interval (CI) of CRP level greater than 3 mg/L for major CVD was elevated (RR, 1.60; 95% CI, 1.19-2.14), with evidence of attenuation (RR, 1.22; 95% CI, 0.90-1.66) in multivariable models. The C statistic, a measure of the discriminatory capability of the prediction models, was 0.74 for prediction of major CVD with age and CRP level. In multivariable models that included traditional risk factors, the C statistic was 0.78, a value that was unchanged with the addition of CRP to the multivariable model. Similar relations were noted for major coronary heart disease events.

Conclusion  Elevated CRP level provided no further prognostic information beyond traditional office risk factor assessment to predict future major CVD and major coronary heart disease in this population sample.

Traditional risk factors for coronary heart disease (CHD) and cardiovascular disease (CVD) have been shown to effectively predict vascular disease events.13 Age, sex, blood pressure, total cholesterol level, high-density lipoprotein cholesterol (HDL-C) level, and cigarette smoking have been considered key factors to assess risk of vascular disease, using continuous or categorical variables. In addition, type 2 diabetes mellitus is highly related to CVD risk, and disease management guidelines in the United States now consider this diagnosis a coronary disease risk equivalent because it augments vascular disease risk so markedly.4,5

Considerable interest exists in moving beyond traditional risk factors to predict CHD events. Although traditional risk factors account for much of the risk for CHD events, and at least 1 risk factor precedes 87% to 100% of CHD deaths, not all CHD risk is explained by the combined effect of traditional risk factors.6 The situation that stirs the greatest interest is screening of middle-aged and older individuals who have yet to experience clinical CVD. The factors and procedures that might improve CHD risk assessment in this situation are being scrutinized in observational studies that have follow-up for incident events.

Assessment of a novel risk factor requires an appreciation of several issues.7 The first is whether the measurement is standardized, accurate, precise, and reproducible in a population setting. A second issue is whether the new factor provides added information over and above what is available with existing assessments. Finally, the existence of a sound biological basis for the measurement, consistent results in observational studies, the potential for interventions to alter the factor, and cost are concerns that need to be addressed before including a new factor in CHD risk assessment.8,9

The Framingham Heart Study provides an opportune cohort in which to test the usefulness of novel risk factors such as inflammatory markers to improve our prediction of initial CHD. Combining the experience of the original Framingham cohort and the second-generation Framingham offspring with 5 years of follow-up for new CVD events allowed this analytic approach to be tested in a community setting.

METHODS

Participants in the original Framingham cohort and the second-generation offspring were considered for this study. The original cohort examination took place in 1980 and served as the baseline for this investigation. The second-generation offspring baseline examination included participants who attended their fifth examination, which took place in 1991-1995. Only persons free of CVD at the baseline examination were considered for inclusion in this study, and all participants were followed up for 8 years for the development of new vascular disease events. A total of 1136 participants in the original cohort and 3310 from the offspring sample attended the baseline examination, had their C-reactive protein (CRP) level measured, and were free of prevalent CVD. The study was approved by the Boston University Medical Center Investigational Review Board, and all participants signed written informed consent.

Baseline risk factor information was available from the baseline examination cycle for age, cigarette smoking during the past year, presence of diabetes mellitus, systolic and diastolic blood pressure measurements obtained in the sitting position, use of blood pressure medications, and body mass index (calculated as weight in kilograms divided by the square of height in meters). Total cholesterol and HDL-C levels were measured at the time of the examinations,10,11 and the laboratory participated in the quality control administered by the Lipid Research Clinics Program and the Centers for Disease Control and Prevention.10 Persons with a fasting glucose level of 126 mg/dL or higher (≥6.99 mmol/L) or who were being treated with insulin or oral hypoglycemic agents were considered diabetic. Serum frozen at −20°C and thawed in 1999 was used to determine CRP level (Hemagen Diagnostics Inc, Columbia, Md).12 Blood samples were collected from an antecubital vein between 8 and 9 AM, with participants in a supine position after an overnight fast. The coefficient of variation for the Hemagen CRP was 8% to 12% for a CRP level less than 1.0 mg/L and 5% to 8% for a CRP level of 1.0 mg/L or greater. On split specimens the correlation coefficient for the Hemagen CRP was 0.86, as previously reported.13 The Pearson product moment correlation between the Hemagen and Dade Behring Inc (Deerfield, Ill) CRP methods for 80 specimens obtained at a later Framingham examination was 0.98, and assignment to CRP quartiles for those split specimens was identical for the Hemagen and Dade Behring techniques. The coefficient of variations for glucose determinations made for this article was 2.7%.

The participants were followed up for the development of initial major CHD (recognized myocardial infarction or CHD death), major CVD (major CHD, stroke, or stroke death), and total CVD (major CVD, angina pectoris, coronary insufficiency, intermittent claudication, or congestive heart failure) using previously described methods.14 Surveillance for CVD events occurred during each examination cycle at the Framingham Heart Study clinic and included review of medical records from physician office visits and hospitalizations for heart and cerebrovascular disease. All outcomes were adjudicated by an end point committee that consisted of 3 trained physicians (P.W.F.W., E.J.B., and C.J.O.).14

Sex-pooled Cox proportional hazards regression was used to examine the risk factors that predicted 3 outcomes: major CHD, major CVD, and total CVD. The relative risks (RRs) were calculated for each risk factor in age- and sex-adjusted and multivariable-adjusted models. Variables included in the CVD prediction analyses were predetermined and based on the published experience for this cohort age. They included age, systolic blood pressure, total cholesterol/HDL-C ratio, diabetes mellitus, cigarette smoking, and blood pressure treatment. Obesity has been shown not to be statistically associated with CVD outcomes in this database, and body mass index and waist girth were not used in the analyses.

For age and systolic blood pressure, the RR and its 95% confidence interval (CI) are based on a 10-U increase. Discrimination, a model’s ability to correctly distinguish events and nonevents or predict the order of events on the the basis of the baseline risk factor profile, was calculated using the overall C statistic.15,16 It can be described as the probability that for any 2 people the model assigns a higher probability of event-free survival to the 1 who survives longer. Reported C statistic levels range from 0.50 (no discrimination) up to a maximum of 1.0 (perfect discrimination). Comparisons were made between the models with CRP and without CRP. Statistical methods included Cox proportional hazards regression and discrimination analyses that included C statistic comparisons with CIs around the estimates.14,16 The Kaplan-Meier and log-rank procedures were used to test for effects of CRP tertiles on risk of vascular outcomes in persons with CHD estimated using a previously published equation that estimates risk of major CHD events with traditional variables. We used SAS statistical software, version 8.2 (SAS Institute Inc, Cary, NC) to perform all statistical analyses. All tests were 2 sided, and P<.05 was considered statistically significant.

RESULTS

There were 1136 participants from the original cohort, including 433 men and 703 women. Correspondingly, there were 3310 offspring participants, including 1516 men and 1794 women. A total of 283 new major CVD events, 160 new major CHD events, and 466 new total CVD events occurred during follow-up (Table 1). Mean levels and frequencies of traditional risk factors are given in Table 2 for men and women. The average age was 57 years in men and 59 years in women. The total cholesterol level was in the 205 to 215 mg/dL (5.31-5.57 mmol/L) range for most participants, current or previous cigarette smoking was reported in approximately 20%, blood pressure treatment was reported in 20% of the sample, and approximately 7% had type 2 diabetes mellitus.

Table 3 gives the RRs for development of new major CHD, major CVD, and total CVD events after adjustment for age and sex and each of the single traditional risk factor variables. Initial multivariable analyses using the traditional variables did not show different effects for the variables in men or women, and there was no evidence of differential effects in the first- and second-generation samples (data not shown), so we have presented the analyses using combined data for both sexes and including the offspring and first-generation cohort experience. Increasing age, systolic blood pressure, total cholesterol/HDL-C ratio, diabetes mellitus, cigarette smoking, and blood pressure drug treatment were each significantly and independently related to the development of new CVD events in both sexes, with a few exceptions.

The age- and sex-adjusted RRs for the development of new major CHD, major CVD, and total CVD using the lowest category of CRP as the referent group are given in Table 4. The age- and sex-adjusted RRs for major CHD and major CVD in the participants with a CRP level greater than 3.0 mg/L were 1.68 (95% CI, 1.14-2.49) and 1.60 (95% CI, 1.19-2.14), respectively. There was evidence of a sex interaction with CRP level only for the outcome total CVD (P = .05). Statistically significant effects of CRP were observed only when the level exceeded 3.0 mg/L; effect sizes for the intermediate category of CRP level of 1.00 to 3.00 mg/L were more modest and not statistically significant.

The multivariable RRs for new CVD according to traditional variables and the 3 clinically relevant CRP categories are given in Table 5. Statistical significance for the traditional variables in Table 5 generally paralleled the individual risk factor analyses presented in Table 3, but the RRs for CRP were attenuated in the multivariable analyses compared with the age-adjusted CRP results (Table 4). Furthermore, after multivariable adjustment, none of the RR estimates for the CRP categories of 1.00 to 3.00 mg/L or greater than 3.00 mg/L were statistically significant compared with the referent level of CRP less than 1.00 mg/L.

The ability to discriminate CVD cases was evaluated by calculation of the C statistic for men and women (Table 6). For age- plus sex-adjusted CRP levels alone, the C statistics (measures of discrimination of cases vs noncases) of major CHD and major CVD were 0.75 and 0.74, respectively. This result implies that given any 2 individuals we have a 75% chance of correctly predicting the individual who will develop the event vs the individual who will not develop the event. For the age- plus sex-adjusted total cholesterol/HDL-C ratio, the comparable C statistics were 0.77 and 0.76, respectively. A model that incorporated all the traditional CVD risk factors was superior to either the model incorporating age- plus sex-adjusted CRP levels or age- and sex-adjusted total cholesterol/HDL-C ratio in its ability to discriminate major CHD (C = 0.80) and major CVD (C = 0.78) cases vs noncases during follow-up. A full predictive model, including traditional risk factors and the CRP categories, provided no added discriminatory capability over and above the traditional model in both sexes (C = 0.80 for major CHD, C = 0.78 for major CVD).

Table 7 gives the estimated 10-year major CVD rates according to CRP levels after prior stratification of persons according to their 10-year CHD risk estimate using a major CHD risk score model developed by D’Agostino et al2 that takes into consideration sex, age, blood pressure, total cholesterol level, HDL-C level, smoking, and diabetes mellitus. The 8-year follow-up information was used to generate the table, and the results were extrapolated to 10 years to aid the comparisons between estimates and actual experience. Tertiles of CRP were used in these analyses to ensure an adequate representation of CRP levels for all strata under investigation. The CRP tertiles generally had no effect at low (<10.00%), intermediate (10.00%-19.99%), or high (≥20.00%) 10-year risk level. The CRP tertiles showed an effect at low (<10.00%), but not at intermediate (10.00%-19.99%) or high (<20.00%) 10-year risk level.

COMMENT

We found no additional predictive value of CRP measurements in estimating risk of new cardiovascular events in this study. The blood biomarker CRP is thought to be the best indicator of vascular inflammation that can be reliably measured,7 and knowing CRP levels in our study showed little benefit in discriminating future major CHD and major CVD cases from noncases. From the Framingham experience, inclusion of CRP testing at the time of screening for CVD and CHD risk does not appear to provide a great benefit for CVD risk assessment, especially because the C statistic results did not increase with the addition of CRP levels over and above traditional risk factors.

In our study, CRP level alone was an effective discriminator of future major CHD and major CVD cases, a finding that is consistent with most clinical and subclinical vascular disease studies in which CRP levels have been highly related to CVD.17,18 Meta-analyses have also generally shown an increased risk for CHD in persons with higher CRP levels,19,20 and it has been reported that CRP levels may improve estimation of risk for CHD events in both sexes.2022 In addition, CRP and interleukin 6 may be increased in persons before the development of type 2 diabetes mellitus.17,19,23,24

A variety of environmental factors may be associated with higher CRP levels, including exogenous estrogen use and adiposity.25,26 On the other hand, lower levels of CRP may occur with statins, physical activity, and alcohol intake.2729 There has been some concern about the accuracy and precision of CRP testing,7 but the newer assays appear to categorize individuals consistently and the Centers for Disease Control and Prevention is enacting a quality control program.30

A critical issue for CRP testing is whether the assay improves prediction of clinical CVD events. A number of observational studies21,3137 have suggested that CHD prediction is improved for men and women, using cross-sectional and prospective designs. Some prospective studies20 have not confirmed a clinically important independent predictive ability of CRP over and above traditional risk factors. Reports7,3740 have also suggested that selective testing should be considered, and measuring CRP levels, or other novel markers, may provide benefit for individuals at an intermediate level of CVD or CHD risk.

Our findings suggest that the incremental benefit of CRP testing would be negligible. Some differences have occurred in some of the published reports that might help to explain the more modest incremental predictive value of CRP in our cohort. First, in the Framingham Heart Study all risk factors were determined in an identical manner as a typical office evaluation. Blood pressure was measured directly by a physician as a standardized procedure, and diabetes status and lipid levels were determined by blood testing. Other studies obtain some of the risk factor information by report in a questionnaire rather than direct measurement. Self-reported risk factors may lead to misclassification of risk factor status and a poorer performance of risk factors in risk algorithms. Second, different results might be obtained for prediction of more specific CVD outcomes. It is possible that CRP is more highly related to the occurrence of specific atherothrombotic CVD events, such as myocardial infarction or CHD death, especially in the setting of acute coronary syndromes.41 Our study had insufficient numbers of individual myocardial infarction or CHD death events to test this hypothesis. Finally, it is possible that high-sensitivity assays may differ in their ability to discriminate future CVD cases. In our study, CRP alone performs well in risk discrimination, in concordance with many other positive studies of CRP. The coefficient of variation for our CRP assay is acceptable at higher levels (CRP ≥1.00 mg/L), above which consensus groups have suggested that a CRP value is a potential risk marker, although the coefficient of variation for our assay is somewhat higher at low levels (CRP <1.00 mg/L).

Investigators from the Atherosclerosis Risk in Communities (ARIC) study42,43 have reported that a battery of novel risk factors added to the prediction of major CHD events in persons with and without type 2 diabetes mellitus, but single novel factors were generally unimpressive in their ability to improve the prediction of CHD events. The ARIC study results and those reported in the current investigation suggest that development of new CHD and CVD risk estimating equations should proceed with caution, bearing in mind that single factors may add little to the prediction of CVD in the community. It has been suggested that the same situation might not hold for prediction of CHD risk within specific subgroups. For example, coronary artery calcification scoring improved assessment of CHD risk for persons believed to be at intermediate risk using traditional risk factor assessment.44,45 In addition, the ARIC investigators have recently shown that the highest tertile of CRP, in the setting of low-density lipoprotein cholesterol levels less than 130 mg/dL (3.37 mmol/L) and elevated phospholipase A2 assay, identified persons at elevated risk for CHD events.37 Further research may be needed to identify the patient subgroups in which testing for CRP, other new biomarkers, and other imaging tests might offer clinically important incremental value in terms of both risk prediction and preventive treatment benefit.

Back to top
Article Information

Correspondence: Peter W. F. Wilson, MD, General Clinical Research Center, Medical University of South Carolina, 96 Jonathan Lucas St, Clinical Sciences Building, Suite 215, Charleston, SC 29425 (wilsonpw@musc.edu).

Accepted for Publication: July 24, 2005.

Financial Disclosure: Dr Wilson has received research support from GlaxoSmithKline and Wyeth; is a consultant for Eli Lilly and Company, GlaxoSmithKline, and Sanofi-Aventis; and has received honoraria for speaking engagements from Merck & Co Inc and Pfizer Inc.

Funding/Support: This study was supported in part by National Heart, Lung, and Blood Institute grant R01-HL 073272 (Medical University of South Carolina, Charleston); National Heart, Lung, and Blood Institute contract N01-HC-25195 (Boston University, Boston, Mass); and the Donald W. Reynolds Foundation Program in Clinical Cardiovascular Research (Medical University of South Carolina).

References
1.
Wilson  PWD’Agostino  RBLevy  DBelanger  AMSilbershatz  HKannel  WB Prediction of coronary heart disease using risk factor categories. Circulation 1998;971837- 1847
PubMedArticle
2.
D’Agostino  RB  SrGrundy  SSullivan  LMWilson  PCHD Risk Prediction Group, Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA 2001;286180- 187
PubMedArticle
3.
Assmann  GCullen  PSchulte  H Simple scoring scheme for calculating the risk of acute coronary events based on the 10-year follow-up of the prospective cardiovascular Munster (PROCAM) Study. Circulation 2002;105310- 315
PubMedArticle
4.
Assmann  GSchulte  HOberwittler  W New aspects in the prediction of coronary artery disease: the Prospective Cardiovascular Munster Study. Fidge  NHNestel  PJedsAtherosclerosis VII Amsterdam, the Netherlands Elsevier Science Publishers1986;19- 24
5.
Haffner  SMLehto  SRonnemaa  TPyorala  KLaakso  M Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med 1998;339229- 234
PubMedArticle
6.
Greenland  PKnoll  MDStamler  J  et al.  Major risk factors as antecedents of fatal and nonfatal coronary heart disease events. JAMA 2003;290891- 897
PubMedArticle
7.
Pearson  TAMensah  GAAlexander  RW  et al.  Markers of inflammation and cardiovascular disease: application to clinical and public health practice: a statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation 2003;107499- 511
PubMedArticle
8.
Wilson  PW Metabolic risk factors for coronary heart disease: current and future prospects. Curr Opin Cardiol 1999;14176- 185
PubMedArticle
9.
Mosca  L C-reactive protein—to screen or not to screen [editorial]? N Engl J Med 2002;3471615- 1617
PubMedArticle
10.
Lipid Research Clinics Program, Manual of Laboratory Operation.  Bethesda, Md National Institutes of Health1974;
11.
Warnick  GRBenderson  JAlbers  JJ Dextran sulfate-Mg2+ precipitation procedure for quantitation of high-density-lipoprotein cholesterol. Clin Chem 1982;281379- 1382
PubMed
12.
Rost  NSWolf  PAKase  CS  et al.  Plasma concentration of C-reactive protein and risk of ischemic stroke and transient ischemic attack: the Framingham Study. Stroke 2001;322575- 2579
PubMedArticle
13.
Wang  TJNam  BHWilson  PW  et al.  Association of C-reactive protein with carotid atherosclerosis in men and women: the Framingham Heart Study. Arterioscler Thromb Vasc Biol 2002;221662- 1667
PubMedArticle
14.
Cupples  LAD’Agostino  RBKiely  D The Framingham Heart Study, Section 35: An Epidemiological Investigation of Cardiovascular Disease: Survival Following Cardiovascular Events: 30-Year Follow-up.  Bethesda, Md National Heart Lung and Blood Institute1988;
15.
Harrell  FE  JrLee  KLMark  DB Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15361- 387
PubMedArticle
16.
Pencina  MJD’Agostino  RB Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;232109- 2123
PubMedArticle
17.
Redberg  RFRifai  NGee  LRidker  PM Lack of association of C-reactive protein and coronary calcium by electron beam computed tomography in postmenopausal women: implications for coronary artery disease screening. J Am Coll Cardiol 2000;3639- 43
PubMedArticle
18.
Folsom  ARAleksic  NCatellier  DJuneja  HSWu  KK C-reactive protein and incident coronary heart disease in the Atherosclerosis Risk In Communities (ARIC) study. Am Heart J 2002;144233- 238
PubMedArticle
19.
Danesh  JCollins  RAppleby  PPeto  R Association of fibrinogen, C-reactive protein, albumin, or leukocyte count with coronary heart disease: meta-analyses of prospective studies. JAMA 1998;2791477- 1482
PubMedArticle
20.
Danesh  JWheeler  JGHirschfield  GM  et al.  C-reactive protein and other circulating markers of inflammation in the prediction of coronary heart disease. N Engl J Med 2004;3501387- 1397
PubMedArticle
21.
Koenig  WSund  MFrohlich  M  et al.  C-reactive protein, a sensitive marker of inflammation, predicts future risk of coronary heart disease in initially healthy middle-aged men: results from the MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) Augsburg Cohort Study, 1984 to 1992. Circulation 1999;99237- 242
PubMedArticle
22.
Ridker  PMWilson  PWGrundy  SM Should C-reactive protein be added to metabolic syndrome and to assessment of global cardiovascular risk? Circulation 2004;1092818- 2825
PubMedArticle
23.
Pradhan  ADManson  JERifai  NBuring  JERidker  PM C-reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA 2001;286327- 334
PubMedArticle
24.
Folsom  ARPankow  JSTracy  RP  et al.  Association of C-reactive protein with markers of prevalent atherosclerotic disease. Am J Cardiol 2001;88112- 117
PubMedArticle
25.
Cushman  MMeilahn  ENPsaty  BMKuller  LHDobs  ASTracy  RP Hormone replacement therapy, inflammation, and hemostasis in elderly women. Arterioscler Thromb Vasc Biol 1999;19893- 899
PubMedArticle
26.
Visser  MBouter  LMMcQuillan  GMWener  MHHarris  TB Elevated C-reactive protein levels in overweight and obese adults. JAMA 1999;2822131- 2135
PubMedArticle
27.
Albert  MADanielson  ERifai  NRidker  PMPRINCE Investigators, Effect of statin therapy on C-reactive protein levels: the Pravastatin Inflammation/CRP Evaluation (PRINCE): a randomized trial and cohort study. JAMA 2001;28664- 70
PubMedArticle
28.
Albert  MAGlynn  RJRidker  PM Alcohol consumption and plasma concentration of C-reactive protein. Circulation 2003;107443- 447
PubMedArticle
29.
Esposito  KPontillo  ADi Palo  C  et al.  Effect of weight loss and lifestyle changes on vascular inflammatory markers in obese women: a randomized trial. JAMA 2003;2891799- 1804
PubMedArticle
30.
Ockene  ISMatthews  CERifai  NRidker  PMReed  GStanek  E Variability and classification accuracy of serial high-sensitivity C-reactive protein measurements in healthy adults. Clin Chem 2001;47444- 450
PubMed
31.
Ridker  PMGlynn  RJHennekens  CH C-reactive protein adds to the predictive value of total and HDL cholesterol in determining risk of first myocardial infarction. Circulation 1998;972007- 2011
PubMedArticle
32.
Ridker  PMHennekens  CHBuring  JERifai  N C-reactive protein and other markers of inflammation in the prediction of cardiovascular disease in women. N Engl J Med 2000;342836- 843
PubMedArticle
33.
Ridker  PM High-sensitivity C-reactive protein: potential adjunct for global risk assessment in the primary prevention of cardiovascular disease. Circulation 2001;1031813- 1818
PubMedArticle
34.
Albert  MAGlynn  RJRidker  PM Plasma concentration of C-reactive protein and the calculated Framingham Coronary Heart Disease Risk Score. Circulation 2003;108161- 165
PubMedArticle
35.
Blake  GJRifai  NBuring  JERidker  PM Blood pressure, C-reactive protein, and risk of future cardiovascular events. Circulation 2003;1082993- 2999
PubMedArticle
36.
Ridker  PMBuring  JECook  NRRifai  N C-reactive protein, the metabolic syndrome, and risk of incident cardiovascular events: an 8-year follow-up of 14 719 initially healthy American women. Circulation 2003;107391- 397
PubMedArticle
37.
Ballantyne  CMHoogeveen  RCBang  H  et al.  Lipoprotein-associated phospholipase A2, high-sensitivity C-reactive protein, and risk for incident coronary heart disease in middle-aged men and women in the Atherosclerosis Risk in Communities (ARIC) study. Circulation 2004;109837- 842
PubMedArticle
38.
Greenland  PSmith  JS  JrGrundy  SM Improving coronary heart disease risk assessment in asymptomatic people: role of traditional risk factors and noninvasive cardiovascular tests. Circulation 2001;1041863- 1867
PubMedArticle
39.
Wilson  PWSmith  SC  JrBlumenthal  RSBurke  GLWong  ND 34th Bethesda Conference, Task Force #4—how do we select patients for atherosclerosis imaging? J Am Coll Cardiol 2003;411898- 1906
PubMedArticle
40.
Ridker  PMCook  N Clinical usefulness of very high and very low levels of C-reactive protein across the full range of Framingham Risk Scores. Circulation 2004;1091955- 1959
PubMedArticle
41.
Brennan  MLPenn  MSvan Lente  F  et al.  Prognostic value of myeloperoxidase in patients with chest pain. N Engl J Med 2003;3491595- 1604
PubMedArticle
42.
Rosamond  WDChambless  LEFolsom  AR  et al.  Trends in the incidence of myocardial infarction and in mortality due to coronary heart disease, 1987 to 1994. N Engl J Med 1998;339861- 867
PubMedArticle
43.
Chambless  LEFolsom  ARSharrett  AR  et al.  Coronary heart disease risk prediction in the Atherosclerosis Risk in Communities (ARIC) study. J Clin Epidemiol 2003;56880- 890
PubMedArticle
44.
Greenland  PLaBree  LAzen  SPDoherty  TMDetrano  RC Coronary artery calcium score combined with Framingham score for risk prediction in asymptomatic individuals. JAMA 2004;291210- 215[published correction appears in JAMA. 2004;291:563]
PubMedArticle
45.
Raggi  PCooil  BCallister  TQ Use of electron beam tomography data to develop models for prediction of hard coronary events. Am Heart J 2001;141375- 382
PubMedArticle
×