C-Reactive Protein and Coronary Heart Disease
http://www.100md.com
《新英格兰医药杂志》
To the Editor: The analyses presented by Danesh and colleagues (April 1 issue)1 do not support their conclusion that C-reactive protein is only a moderate risk factor for coronary heart disease. First, their comparison of risk factors relies on the odds ratio for the comparison of the top and bottom thirds of observed distributions in their population. Such strategies yield arbitrary results that can vary greatly among populations with different risk-factor distributions.2 Cholesterol levels in Reykjavik, Iceland, in 1967 were far higher than current U.S. levels of cholesterol,3 whereas the levels of C-reactive protein were lower than current U.S. levels,4 thus raising the risk associated with cholesterol while reducing that associated with C-reactive protein in their analyses. Second, both the odds ratio and the incremental area under the receiver-operating-characteristic curve are virtually identical for C-reactive protein and systolic blood pressure, despite the fact that C-reactive protein was adjusted for blood pressure, but not vice versa. Third, the odds ratio with respect to C-reactive protein is much higher for the 10-year risk, which forms the basis for the risk models in the Framingham Heart Study.5 More statistically principled consideration of these case–control data would lead to the conclusion that C-reactive protein remains an important predictor of risk.
Robert J. Glynn, Sc.D.
Nancy R. Cook, Sc.D.
Harvard Medical School
Boston, MA 02115
References
Danesh J, Wheeler JG, Hirshfield GM, et al. C-reactive protein and other circulatory markers of inflammation in the prediction of coronary heart disease. N Engl J Med 2004;350:1387-1397.
Greenland S, Schlesselman JJ, Criqui MH. The fallacy of employing standardized regression coefficients and correlations as measures of effects. Am J Epidemiol 1986;123:203-208.
Ford ES, Mokdad AH, Giles WH, Mensah GA. Serum total cholesterol concentrations and awareness, treatment, and control of hypercholesterolemia among US adults: findings from the National Health and Nutrition Examination Survey, 1999 to 2000. Circulation 2003;107:2185-2189.
Folsom AR, Aleksic N, Catellier D, Juneja HS, Wu KK. C-reactive protein and incident coronary heart disease in the Atherosclerosis Risk In Communities (ARIC) study. Am Heart J 2002;144:233-238.
Wilson PW, D'Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation 1998;97:1837-1847.
To the Editor: By using a cutoff point of 2.0 mg per liter, rather than 3.0 mg per liter, as recommended by the American Heart Association (AHA) and the Centers for Disease Control and Prevention (CDC),1 Danesh et al. may have underestimated the risk associated with C-reactive protein. As can be surmised from a footnote to Table 2 of their article, the fully adjusted odds ratio associated with C-reactive protein when cutoff points closer to 3.0 mg per liter are used resembles those previously reported. Additional adjustment for diabetes, triglycerides, obesity, pulmonary function, social status, and other inflammatory markers also tends to reduce the predictive value of C-reactive protein, as does inclusion of the 20 percent of study end points that failed to meet confirmation criteria for cardiovascular events.
There have now been 22 prospective studies of C-reactive protein and the risk of heart disease. In all the large studies, including the study from Iceland, C-reactive protein was found to be a significant predictor of the risk of heart disease, even after adjustment for the covariates in the Framingham Heart Study risk algorithm. Given these findings, we concur that recommendations regarding the use of C-reactive protein may need to be reviewed.1 We believe, however, that, if anything, there should be a broadening, rather than a curtailing, of the use of C-reactive protein in screening selected populations.
Peter Libby, M.D.
Brigham and Women's Hospital
Boston, MA 02115
James T. Willerson, M.D.
Texas Heart Institute
Houston, TX 77030
Eugene Braunwald, M.D.
Brigham and Women's Hospital
Boston, MA 02115
References
Pearson TA, Mensah GA, Alexander 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;107:499-511.
To the Editor: Each year, 1.3 million Americans have an acute myocardial infarction. Only half of these patients would have been identified on the basis of traditional risk factors alone. Additional tools are needed to screen for the risk of cardiovascular disease. High-sensitivity C-reactive protein is a powerful, inexpensive tool for this purpose.1,2,3,4 Whereas 22 prospective studies have demonstrated the predictive power of high-sensitivity C-reactive protein, its clinical role has been challenged by Danesh et al. Their study of a homogeneous Icelandic population was based on a cutoff point of 2.0 mg per liter, rather than 3.0 mg per liter, as recommended by the AHA and the CDC,2 and underestimates the value of C-reactive protein assessment. Methodology aside, even the authors' conservative estimate of the increment in risk prediction — 50 percent — translates into a meaningful risk assessment at the population level.
Given the high societal costs of cardiovascular disease and the current suboptimal screening paradigms, high-sensitivity C-reactive protein is an inexpensive, valid tool to predict the risk of this disease. We strongly support the AHA–CDC recommendations with respect to C-reactive protein and believe that high-sensitivity C-reactive protein should be more fully incorporated into clinical practice, rather than restricted, as Danesh et al. suggest.
JoAnne M. Foody, M.D.
Yale University School of Medicine
New Haven, CT 06520
Antonio M. Gotto, M.D., D.Phil.
Weill Medical College of Cornell University
New York, NY 10021
Nanette Wenger, M.D.
Emory University School of Medicine
Atlanta, GA 30303
References
Ridker PM, Glynn RJ, Hennekens CH. C-reactive protein adds to the predictive value of total and HDL cholesterol in determining risk of first myocardial infarction. Circulation 1998;97:2007-2011.
Pearson TA, Mensah GA, Alexander 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;107:499-511.
Ballantyne CM, Hoogeveen RC, Bang 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;109:837-842.
Koenig W, Lowel H, Baumert J, Meisinger C. C-reactive protein modulates risk prediction based on the Framingham Score: implications for future risk assessment: results from a large cohort study in Southern Germany. Circulation 2004;109:1349-1353.
To the Editor: Danesh et al. report that the decade-to-decade consistency of C-reactive protein is equal to that of cholesterol, an observation that strongly supports the use of C-reactive protein as a new means for predicting cardiovascular disease. However, the authors also state that the risk associated with C-reactive protein is a "relatively moderate predictor," a misleading conclusion that is, unfortunately, echoed by Tall in his accompanying editorial.1 In fact, the odds ratio with respect to C-reactive protein as reported by Danesh et al., after adjustment for age, sex, period, and risk factors for coronary heart disease (1.5; 95 percent confidence interval, 1.3 to 1.7), is the same as the odds ratio associated with hypertension and is statistically similar to that associated with smoking. Thus, according to their own analysis, if the authors do not conclude that C-reactive protein has clinical value for risk prediction, then they must also believe that hypertension and smoking are unimportant predictors. We also note that the 10-year risk estimate reported by Danesh et al. for C-reactive protein (adjusted odds ratio, 1.84) is fully in line with data from the Women's Health Study,2 the MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) study,3 and the ARIC (Atherosclerosis Risk in Communities) study.4
Paul M. Ridker, M.D.
Brigham and Women's Hospital
Boston, MA 02115
Wolfgang Koenig, M.D.
University of Ulm Medical Center
D-89081 Ulm, Germany
Valentin Fuster, M.D.
Mount Sinai Cardiovascular Institute
New York, NY 10029
Editor's note: Dr. Ridker is listed as a coinventor on patents held by the Brigham and Women's Hospital that relate to the use of inflammatory biomarkers in cardiovascular disease.
References
Tall AR. C-reactive protein reassessed. N Engl J Med 2004;350:1450-1452.
Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. Comparison of C-reactive protein and low-density lipoprotein cholesterol levels in the prediction of first cardiovascular events. N Engl J Med 2002;347:1557-1565.
Koenig W, Lowel H, Baumert J, Meisinger C. C-reactive protein modulates risk prediction based on the Framingham Score: implications for future risk assessment: results from a large cohort study in Southern Germany. Circulation 2004;109:1349-1353.
Ballantyne CM, Hoogeveen RC, Bang 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;109:837-842.
The authors reply: New data from our large prospective study (involving 2406 patients who had nonfatal myocardial infarction or who died from coronary heart disease) and from an updated meta-analysis of 21 previous reports (including a total of 4662 such patients) collectively indicate that the value of C-reactive protein measurement in the prediction of the risk of coronary disease was previously overestimated. In a previous meta-analysis, the odds ratio for nonfatal myocardial infarction or death from coronary heart disease was about 2.0 (95 percent confidence interval, 1.6 to 2.5) in a comparison of extreme thirds of C-reactive protein values,1 and in a few, relatively small studies, the odds ratio was even greater.2 By contrast, the best current evidence suggests that the odds ratio is about 1.5 (95 percent confidence interval, 1.4 to 1.6).
Moreover, contrary to some earlier suggestions, the current data indicate that established coronary risk factors are generally stronger predictors than are C-reactive protein values. For example, in our study, the adjusted odds ratios among people with raised total cholesterol concentrations and among smokers were approximately 2.4 (95 percent confidence interval, 2.0 to 2.7) and 1.9 (95 percent confidence interval, 1.6 to 2.2), respectively. Measurement of C-reactive protein added only marginally to the predictive value of established risk factors: the area under the receiver-operating-characteristic curve increased from 0.64 (95 percent confidence interval, 0.63 to 0.65) to only 0.65 (95 percent confidence interval, 0.64 to 0.67) when it was included.
About 92 percent of the cases in our study (2206 of 2406) involved confirmed nonfatal myocardial infarction or death from coronary causes (not 80 percent of the cases, as implied by Libby and colleagues), and, as stated previously, the predictive value of C-reactive protein was not materially altered by exclusion of the 8 percent of cases involving possible myocardial infarction. Our main analyses were adjusted appropriately for possible confounders (including obesity and diabetes); contrary to the suggestion of Libby and colleagues, further adjustments for socioeconomic status and other inflammatory markers were included only in subsidiary analyses. In our article, we state that the contrast between the current findings and the substantially greater odds ratios reported in the previous meta-analysis was not altered by comparisons among extreme quarters or extreme fifths rather than among extreme thirds. The use of any particular cutoff point (such as 3 mg per liter), consistently applied among studies, would also yield similar contrasts with previous estimates, provided that the shape of the relationship between C-reactive protein values and coronary risk were approximately log-linear. Reliable characterization of the shape of this relationship and investigation of the use of C-reactive protein for risk prediction in particular subgroups (e.g., in those with different serum lipid concentrations and over different periods of follow-up) will require pooling of individual records from prospective studies. Such work is now in progress at the secretariat of the Emerging Risk Factors Collaboration,3 which has already received more than half of the available data.
John Danesh, M.B., Ch.B., D.Phil.
University of Cambridge
Cambridge CB1 8RN, United Kingdom
Mark Pepys, M.D., Ph.D.
Royal Free and University College Medical School
London NW3 2PF, United Kingdom
Vilmundur Gudnason, M.D., Ph.D.
Icelandic Heart Association
201 Kopavogur, Iceland
References
Danesh J, Whincup P, Walker M, et al. Low grade inflammation and coronary heart disease: prospective study and updated meta-analyses. BMJ 2000;321:199-204.
Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. Comparison of C-reactive protein and low-density lipoprotein cholesterol levels in the prediction of first cardiovascular events. N Engl J Med 2002;347:1557-1565.
Fibrinogen Studies Collaboration. Collaborative meta-analyses of prospective studies of plasma fibrinogen and cardiovascular disease. Eur J Cardiovasc Prev Rehabil 2004;11:9-17.
Robert J. Glynn, Sc.D.
Nancy R. Cook, Sc.D.
Harvard Medical School
Boston, MA 02115
References
Danesh J, Wheeler JG, Hirshfield GM, et al. C-reactive protein and other circulatory markers of inflammation in the prediction of coronary heart disease. N Engl J Med 2004;350:1387-1397.
Greenland S, Schlesselman JJ, Criqui MH. The fallacy of employing standardized regression coefficients and correlations as measures of effects. Am J Epidemiol 1986;123:203-208.
Ford ES, Mokdad AH, Giles WH, Mensah GA. Serum total cholesterol concentrations and awareness, treatment, and control of hypercholesterolemia among US adults: findings from the National Health and Nutrition Examination Survey, 1999 to 2000. Circulation 2003;107:2185-2189.
Folsom AR, Aleksic N, Catellier D, Juneja HS, Wu KK. C-reactive protein and incident coronary heart disease in the Atherosclerosis Risk In Communities (ARIC) study. Am Heart J 2002;144:233-238.
Wilson PW, D'Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation 1998;97:1837-1847.
To the Editor: By using a cutoff point of 2.0 mg per liter, rather than 3.0 mg per liter, as recommended by the American Heart Association (AHA) and the Centers for Disease Control and Prevention (CDC),1 Danesh et al. may have underestimated the risk associated with C-reactive protein. As can be surmised from a footnote to Table 2 of their article, the fully adjusted odds ratio associated with C-reactive protein when cutoff points closer to 3.0 mg per liter are used resembles those previously reported. Additional adjustment for diabetes, triglycerides, obesity, pulmonary function, social status, and other inflammatory markers also tends to reduce the predictive value of C-reactive protein, as does inclusion of the 20 percent of study end points that failed to meet confirmation criteria for cardiovascular events.
There have now been 22 prospective studies of C-reactive protein and the risk of heart disease. In all the large studies, including the study from Iceland, C-reactive protein was found to be a significant predictor of the risk of heart disease, even after adjustment for the covariates in the Framingham Heart Study risk algorithm. Given these findings, we concur that recommendations regarding the use of C-reactive protein may need to be reviewed.1 We believe, however, that, if anything, there should be a broadening, rather than a curtailing, of the use of C-reactive protein in screening selected populations.
Peter Libby, M.D.
Brigham and Women's Hospital
Boston, MA 02115
James T. Willerson, M.D.
Texas Heart Institute
Houston, TX 77030
Eugene Braunwald, M.D.
Brigham and Women's Hospital
Boston, MA 02115
References
Pearson TA, Mensah GA, Alexander 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;107:499-511.
To the Editor: Each year, 1.3 million Americans have an acute myocardial infarction. Only half of these patients would have been identified on the basis of traditional risk factors alone. Additional tools are needed to screen for the risk of cardiovascular disease. High-sensitivity C-reactive protein is a powerful, inexpensive tool for this purpose.1,2,3,4 Whereas 22 prospective studies have demonstrated the predictive power of high-sensitivity C-reactive protein, its clinical role has been challenged by Danesh et al. Their study of a homogeneous Icelandic population was based on a cutoff point of 2.0 mg per liter, rather than 3.0 mg per liter, as recommended by the AHA and the CDC,2 and underestimates the value of C-reactive protein assessment. Methodology aside, even the authors' conservative estimate of the increment in risk prediction — 50 percent — translates into a meaningful risk assessment at the population level.
Given the high societal costs of cardiovascular disease and the current suboptimal screening paradigms, high-sensitivity C-reactive protein is an inexpensive, valid tool to predict the risk of this disease. We strongly support the AHA–CDC recommendations with respect to C-reactive protein and believe that high-sensitivity C-reactive protein should be more fully incorporated into clinical practice, rather than restricted, as Danesh et al. suggest.
JoAnne M. Foody, M.D.
Yale University School of Medicine
New Haven, CT 06520
Antonio M. Gotto, M.D., D.Phil.
Weill Medical College of Cornell University
New York, NY 10021
Nanette Wenger, M.D.
Emory University School of Medicine
Atlanta, GA 30303
References
Ridker PM, Glynn RJ, Hennekens CH. C-reactive protein adds to the predictive value of total and HDL cholesterol in determining risk of first myocardial infarction. Circulation 1998;97:2007-2011.
Pearson TA, Mensah GA, Alexander 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;107:499-511.
Ballantyne CM, Hoogeveen RC, Bang 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;109:837-842.
Koenig W, Lowel H, Baumert J, Meisinger C. C-reactive protein modulates risk prediction based on the Framingham Score: implications for future risk assessment: results from a large cohort study in Southern Germany. Circulation 2004;109:1349-1353.
To the Editor: Danesh et al. report that the decade-to-decade consistency of C-reactive protein is equal to that of cholesterol, an observation that strongly supports the use of C-reactive protein as a new means for predicting cardiovascular disease. However, the authors also state that the risk associated with C-reactive protein is a "relatively moderate predictor," a misleading conclusion that is, unfortunately, echoed by Tall in his accompanying editorial.1 In fact, the odds ratio with respect to C-reactive protein as reported by Danesh et al., after adjustment for age, sex, period, and risk factors for coronary heart disease (1.5; 95 percent confidence interval, 1.3 to 1.7), is the same as the odds ratio associated with hypertension and is statistically similar to that associated with smoking. Thus, according to their own analysis, if the authors do not conclude that C-reactive protein has clinical value for risk prediction, then they must also believe that hypertension and smoking are unimportant predictors. We also note that the 10-year risk estimate reported by Danesh et al. for C-reactive protein (adjusted odds ratio, 1.84) is fully in line with data from the Women's Health Study,2 the MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) study,3 and the ARIC (Atherosclerosis Risk in Communities) study.4
Paul M. Ridker, M.D.
Brigham and Women's Hospital
Boston, MA 02115
Wolfgang Koenig, M.D.
University of Ulm Medical Center
D-89081 Ulm, Germany
Valentin Fuster, M.D.
Mount Sinai Cardiovascular Institute
New York, NY 10029
Editor's note: Dr. Ridker is listed as a coinventor on patents held by the Brigham and Women's Hospital that relate to the use of inflammatory biomarkers in cardiovascular disease.
References
Tall AR. C-reactive protein reassessed. N Engl J Med 2004;350:1450-1452.
Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. Comparison of C-reactive protein and low-density lipoprotein cholesterol levels in the prediction of first cardiovascular events. N Engl J Med 2002;347:1557-1565.
Koenig W, Lowel H, Baumert J, Meisinger C. C-reactive protein modulates risk prediction based on the Framingham Score: implications for future risk assessment: results from a large cohort study in Southern Germany. Circulation 2004;109:1349-1353.
Ballantyne CM, Hoogeveen RC, Bang 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;109:837-842.
The authors reply: New data from our large prospective study (involving 2406 patients who had nonfatal myocardial infarction or who died from coronary heart disease) and from an updated meta-analysis of 21 previous reports (including a total of 4662 such patients) collectively indicate that the value of C-reactive protein measurement in the prediction of the risk of coronary disease was previously overestimated. In a previous meta-analysis, the odds ratio for nonfatal myocardial infarction or death from coronary heart disease was about 2.0 (95 percent confidence interval, 1.6 to 2.5) in a comparison of extreme thirds of C-reactive protein values,1 and in a few, relatively small studies, the odds ratio was even greater.2 By contrast, the best current evidence suggests that the odds ratio is about 1.5 (95 percent confidence interval, 1.4 to 1.6).
Moreover, contrary to some earlier suggestions, the current data indicate that established coronary risk factors are generally stronger predictors than are C-reactive protein values. For example, in our study, the adjusted odds ratios among people with raised total cholesterol concentrations and among smokers were approximately 2.4 (95 percent confidence interval, 2.0 to 2.7) and 1.9 (95 percent confidence interval, 1.6 to 2.2), respectively. Measurement of C-reactive protein added only marginally to the predictive value of established risk factors: the area under the receiver-operating-characteristic curve increased from 0.64 (95 percent confidence interval, 0.63 to 0.65) to only 0.65 (95 percent confidence interval, 0.64 to 0.67) when it was included.
About 92 percent of the cases in our study (2206 of 2406) involved confirmed nonfatal myocardial infarction or death from coronary causes (not 80 percent of the cases, as implied by Libby and colleagues), and, as stated previously, the predictive value of C-reactive protein was not materially altered by exclusion of the 8 percent of cases involving possible myocardial infarction. Our main analyses were adjusted appropriately for possible confounders (including obesity and diabetes); contrary to the suggestion of Libby and colleagues, further adjustments for socioeconomic status and other inflammatory markers were included only in subsidiary analyses. In our article, we state that the contrast between the current findings and the substantially greater odds ratios reported in the previous meta-analysis was not altered by comparisons among extreme quarters or extreme fifths rather than among extreme thirds. The use of any particular cutoff point (such as 3 mg per liter), consistently applied among studies, would also yield similar contrasts with previous estimates, provided that the shape of the relationship between C-reactive protein values and coronary risk were approximately log-linear. Reliable characterization of the shape of this relationship and investigation of the use of C-reactive protein for risk prediction in particular subgroups (e.g., in those with different serum lipid concentrations and over different periods of follow-up) will require pooling of individual records from prospective studies. Such work is now in progress at the secretariat of the Emerging Risk Factors Collaboration,3 which has already received more than half of the available data.
John Danesh, M.B., Ch.B., D.Phil.
University of Cambridge
Cambridge CB1 8RN, United Kingdom
Mark Pepys, M.D., Ph.D.
Royal Free and University College Medical School
London NW3 2PF, United Kingdom
Vilmundur Gudnason, M.D., Ph.D.
Icelandic Heart Association
201 Kopavogur, Iceland
References
Danesh J, Whincup P, Walker M, et al. Low grade inflammation and coronary heart disease: prospective study and updated meta-analyses. BMJ 2000;321:199-204.
Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. Comparison of C-reactive protein and low-density lipoprotein cholesterol levels in the prediction of first cardiovascular events. N Engl J Med 2002;347:1557-1565.
Fibrinogen Studies Collaboration. Collaborative meta-analyses of prospective studies of plasma fibrinogen and cardiovascular disease. Eur J Cardiovasc Prev Rehabil 2004;11:9-17.