C-Reactive Protein and Other Circulating Markers of Inflammation in the Prediction of Coronary Heart Disease
http://www.100md.com
《新英格兰医药杂志》
ABSTRACT
Background C-reactive protein is an inflammatory marker believed to be of value in the prediction of coronary events. We report data from a large study of C-reactive protein and other circulating inflammatory markers, as well as updated meta-analyses, to evaluate their relevance to the prediction of coronary heart disease.
Methods Measurements were made in samples obtained at base line from up to 2459 patients who had a nonfatal myocardial infarction or died of coronary heart disease during the study and from up to 3969 controls without a coronary heart disease event in the Reykjavik prospective study of 18,569 participants. Measurements were made in paired samples obtained an average of 12 years apart from 379 of these participants in order to quantify within-person fluctuations in inflammatory marker levels.
Results The long-term stability of C-reactive protein values (within-person correlation coefficient, 0.59; 95 percent confidence interval, 0.52 to 0.66) was similar to that of both blood pressure and total serum cholesterol. After adjustment for base-line values for established risk factors, the odds ratio for coronary heart disease was 1.45 (95 percent confidence interval, 1.25 to 1.68) in a comparison of participants in the top third of the group with respect to base-line C-reactive protein values with those in the bottom third, and similar overall findings were observed in an updated meta-analysis involving a total of 7068 patients with coronary heart disease. By comparison, the odds ratios in the Reykjavik Study for coronary heart disease were somewhat weaker for the erythrocyte sedimentation rate (1.30; 95 percent confidence interval, 1.13 to 1.51) and the von Willebrand factor concentration (1.11; 95 percent confidence interval, 0.97 to 1.27) but generally stronger for established risk factors, such as an increased total cholesterol concentration (2.35; 95 percent confidence interval, 2.03 to 2.74) and cigarette smoking (1.87; 95 percent confidence interval, 1.62 to 2.16).
Conclusions C-reactive protein is a relatively moderate predictor of coronary heart disease. Recommendations regarding its use in predicting the likelihood of coronary heart disease may need to be reviewed.
Since atherosclerosis may, in part, be an inflammatory disease,1 circulating factors related to inflammation may be predictors of cardiovascular disease in general populations.2 A recent statement from the Centers for Disease Control and Prevention and the American Heart Association concluded that it is reasonable to measure C-reactive protein, a sensitive circulating marker of inflammation, as an adjunct to the measurement of established risk factors in order to assess the risk of coronary heart disease.3 The report acknowledged, however, that the epidemiologic data to support this view were not entirely consistent and recommended that larger prospective studies be conducted to improve the reliability of the evidence.
We measured C-reactive protein concentrations in approximately 2400 patients with coronary heart disease diagnosed since their enrollment in the cohort and approximately 4000 controls nested within the Reykjavik Study, a prospective cohort study of about 19,000 middle-aged men and women without a history of myocardial infarction. The number of cases of coronary heart disease in this cohort was about four times as great as in the largest previous study4 and should reduce the scope for random error in our estimates. We also assessed the effect of within-person variation in the concentrations of inflammatory markers5 in serial blood samples obtained over a period of several years in several hundred participants. To compare the predictive value of the C-reactive protein concentration with that of some other inflammatory markers studied in coronary heart disease, we also analyzed the erythrocyte sedimentation rate and circulating concentrations of von Willebrand factor, each of which can also fluctuate considerably in acute-phase inflammatory responses.6,7 To help put the new data in context, we updated meta-analyses of previous relevant studies of each of these inflammatory markers.
Methods
Patients and Controls
The Reykjavik Study, initiated in 1967 as a prospective study of cardiovascular disease, has been described in detail previously.8 All men born between 1907 and 1934 and all women born between 1908 and 1935 who were residents of Reykjavik, Iceland, and its adjacent communities on December 1, 1966, were identified in the national population register and then invited to participate in the study during five stages of recruitment between 1967 and 1991. A total of 8888 men and 9681 women without a history of myocardial infarction were enrolled, reflecting a response rate of 72 percent.9
Nurses administered questionnaires, made physical measurements, performed spirometry and electrocardiography, and collected venous blood samples after an overnight fast for the measurement of the erythrocyte sedimentation rate and to prepare aliquots of serum, which were stored at –20°C for subsequent analysis. All participants have subsequently been monitored with respect to death from any cause and the occurrence of major cardiovascular conditions, with a total loss to follow-up of only about 0.6 percent of participants.9
A total of 2459 men and women with available serum samples had major coronary events between the beginning of follow-up and December 31, 1995, for a mean (±SD) duration of follow-up of 17.5±8.7 years, as compared with 20.6±8.2 years among controls. Among the men, 1073 deaths from coronary heart disease and 701 nonfatal myocardial infarctions were recorded (564 confirmed and 137 possible myocardial infarctions), and among the women, 385 died of coronary heart disease and 300 had a nonfatal myocardial infarction (237 confirmed and 63 possible myocardial infarctions). Deaths from coronary heart disease were ascertained from central registers on the basis of a death certificate listing an International Classification of Diseases code of 410 through 414, and the diagnosis of nonfatal myocardial infarction was based on the criteria of the Monitoring Trends and Determinants in Cardiovascular Disease study.
We selected 3969 control subjects from among the participants who had survived to the end of the study period without having a myocardial infarction. The controls were frequency-matched to the patients with respect to the calendar year of recruitment, sex, and age (in five-year increments).10
The National Bioethics Committee and the Data Protection Authority of Iceland approved the study protocol. All participants provided informed consent.
Laboratory Methods
Laboratory measurements were made without knowledge of the participants' disease status, and thus samples from patients and controls were randomly distributed among assay plates. Concentrations of C-reactive protein were measured by latex-enhanced immunoturbidimetry, with a lower limit of detection of 0.02 mg per liter (Roche Diagnostics).11 The variation in C-reactive protein values within runs was less than 1 percent, and the between-day variability was 1 percent at a concentration of 14 mg per liter and 3.7 percent at a concentration of 3.8 mg per liter. The concentration of von Willebrand factor was determined by means of a sensitive enzyme immunoassay. We also determined the concentration of von Willebrand factor in paired plasma and serum samples from 56 healthy persons from another study and found close agreement between plasma and serum values (correlation coefficient, 0.94).7 The Wintrobe method was used to measure the erythrocyte sedimentation rate in fresh blood samples obtained at the time of base-line venesection.6 Other biochemical and hematologic measurements involved the use of standard assays, as previously described.8 Measurements were made in pairs of samples obtained from 379 participants a mean of about 12 years apart. Data on erythrocyte sedimentation rate from the Reykjavik Study have been reported previously.12
Statistical Analysis
Comparisons between patients and controls were made by means of unmatched stratified logistic regression fitted according to the unconditional maximum likelihood (Stata software, version 7). To maximize the ability to compare our results with those of previous reports, primary analyses of values of C-reactive protein, erythrocyte sedimentation rate, and von Willebrand factor were prespecified to compare extreme thirds of patients and controls with respect to the distribution of values in the controls. Subsidiary analyses involved other cutoff values. Odds ratios were sequentially adjusted for the following variables: age, sex, calendar year of enrollment, smoking status, systolic blood pressure, total cholesterol level, triglyceride level, body-mass index (the weight in kilograms divided by the square of the height in meters), forced expiratory volume in one second, presence or absence of diabetes, socioeconomic status, and the concentrations of other markers of inflammation.
To estimate the discriminative value of predictive models, we calculated the areas under the receiver-operating-characteristic curve, in order to determine whether the sequential addition of data on inflammatory markers increased the predictive value of major established coronary risk factors, as described previously.13 We performed meta-analyses of studies published before January 2003 that included essentially general populations (i.e., cohorts not selected on the basis of preexisting disease) with more than a year of follow-up, using search, abstraction, and data-synthesis methods that have been described previously and using nonfatal myocardial infarction or death from coronary heart disease as end points.6,7,14 We combined the results of the studies by using inverse variance-weighted averages of logarithmic odds ratios. Heterogeneity was assessed by means of standard 2 tests. Odds ratios are given with 95 percent confidence intervals, and two-sided P values are reported. Since previous studies have reported on the predictive values of single base-line measurements of inflammatory markers with respect to coronary heart disease, odds ratios have not been corrected for regression dilution in the present study, so as to allow direct comparisons with previous work.5
Results
The mean age at the time of the coronary heart disease event was 70.2±9.7 years. There were significant differences between patients and controls with respect to established coronary risk factors, such as smoking status, body-mass index, blood pressure, and serum lipid concentrations (Table 1).
Table 1. Base-Line Characteristics of the Patients with Coronary Heart Disease and Controls.
Base-Line Associations and Long-Term Stability of Inflammatory Markers
The partial correlation coefficients (adjusted for age, sex, calendar year of recruitment, and smoking status) for C-reactive protein, on the one hand, and the erythrocyte sedimentation rate and von Willebrand factor, on the other, were 0.38 and 0.18, respectively (P<0.001 for each comparison), and the partial correlation coefficient for the erythrocyte sedimentation rate and the von Willebrand factor concentration was 0.17 (P<0.001). A higher C-reactive protein concentration was significantly associated with cigarette smoking (P<0.001), an increased body-mass index (P<0.001), a low forced expiratory volume in one second (P<0.001), and an increased triglyceride concentration (P<0.001) (data not shown). Higher values for the erythrocyte sedimentation rate were significantly associated with older age (P<0.001), female sex (P<0.001), a low hemoglobin value (P<0.001), a low hematocrit (P<0.001), an elevated serum uric acid concentration (P<0.001), a low forced expiratory volume in one second (P<0.001), and smoking (P<0.001). A higher von Willebrand factor concentration was significantly associated with older age (P<0.001) and smoking (P<0.001).
Among 379 participants who provided paired blood samples, the within-person correlation coefficients for C-reactive protein, erythrocyte sedimentation rate, and von Willebrand factor were 0.59 (95 percent confidence interval, 0.52 to 0.66), 0.67 (95 percent confidence interval, 0.61 to 0.73), and 0.57 (95 percent confidence interval, 0.50 to 0.64), respectively. These values were similar with respect to long-term consistency to the values for systolic blood pressure (correlation coefficient, 0.66; 95 percent confidence interval, 0.60 to 0.72), diastolic blood pressure (correlation coefficient, 0.53; 95 percent confidence interval, 0.46 to 0.60), and total serum cholesterol (correlation coefficient, 0.60; 95 percent confidence interval, 0.54 to 0.66).
Inflammatory Markers and Incident Coronary Heart Disease
The odds ratio for coronary heart disease was 1.92 (95 percent confidence interval, 1.68 to 2.18; 2=105, with 1 df) among patients with values in the top third (cutoff value, 2.0 mg per liter), as compared with the bottom third (cutoff value, 0.78 mg per liter), of base-line C-reactive protein concentrations in the control group. The odds ratio fell to 1.45 (95 percent confidence interval, 1.25 to 1.68; 2=28, with 1 df) after adjustment for smoking status, other established coronary risk factors, and indicators of socioeconomic status (Table 2). Comparisons between the top and bottom thirds of patients and controls with respect to the other markers gave the following adjusted odds ratios for coronary heart disease: for erythrocyte sedimentation rate (cutoff value of 10 mm in first hour of measurement for the top third and 4 mm in first hour for the bottom third), 1.30 (95 percent confidence interval, 1.13 to 1.51; 2=13, with 1 df), and for von Willebrand factor (cutoff value of 124 IU per deciliter for the top third and 88 IU per deciliter for the bottom third), 1.11 (95 percent confidence interval, 0.97 to 1.27; 2=26, with 1 df) (Table 2 and Figure 1). The calculated areas under receiver-operating-characteristic curves indicate that information on the C-reactive protein concentration (and the other inflammatory markers that were assessed) provided comparatively little additional predictive value over that provided by assessment of major established risk factors (Figure 1).
Table 2. Relative Odds of Coronary Heart Disease (CHD) among Patients Who Had Levels of Inflammatory Markers in the Top Third of the Distribution of Values for Controls, as Compared with Those Who Had Values in the Bottom Third of This Distribution.
Figure 1. Odds Ratios for Coronary Heart Disease among 2459 Patients with Coronary Heart Disease and 3969 Controls.
Comparisons are between patients and controls with values in the top third and those in the bottom third of the distribution of values for controls, except for comparisons involving smoking status. Squares denote odds ratios, and horizontal lines represent 95 percent confidence intervals. The information plotted in this figure is based on odds ratios listed in the next-to-last column of Table 2. Logistic-regression analysis was used to calculate the areas under the receiver-operating-characteristic (ROC) curve after adjustment for age, sex, and period, with data on major established risk factors and inflammatory markers added to the model in the order of the strength of each variable's association with coronary heart disease.
These findings were not materially changed in analyses restricted to the 2083 patients without evidence of coronary heart disease at base line (Table 2), to the 2206 patients with C-reactive protein values who had a confirmed myocardial infarction or died of coronary heart disease, or to the participants without evidence of acute-phase reactions at the base-line examination (i.e., this analysis excluded 132 patients and 152 controls with a C-reactive protein concentration of more than 10 mg per liter15 or an erythrocyte sedimentation rate of more than 30 mm during the first hour). The findings were also unaffected by changes in the cutoff values (e.g., analyses of quarters or fifths, or according to increases of 1 SD) (Table 2).
Associations between the C-reactive protein concentration and the risk of coronary heart disease did not vary significantly according to established risk factors, such as smoking or increased blood lipid concentrations, blood pressure, or body-mass index (data not shown). An exploratory analysis suggested the possibility of more extreme odds ratios among the 1049 patients who died of coronary heart disease or had a nonfatal myocardial infarction within 10 years after enrollment (odds ratio, 1.84; 95 percent confidence interval, 1.49 to 2.28), as compared with the 1357 patients who had such an event after the first decade (odds ratio, 1.26; 95 percent confidence interval, 1.05 to 1.51). Such a trend, however, was not observed in the updated meta-analysis, described below, which was based on published data from 22 studies2,4,13,14,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33 (Figure 2). Therefore, it requires further examination involving larger numbers of participants with individual data. Such analysis is also required for a reliable characterization of the shape of the association between C-reactive protein and coronary heart disease.
Figure 2. Twenty-Two Prospective Studies of the Association of C-Reactive Protein Concentrations with the Risk of Coronary Heart Disease (CHD) in Essentially General Populations, Grouped According to Several Study Characteristics.
One of the 11 studies published before 2000 was updated in 200213,16; hence, data on 85 cases from this study contributed to two subtotals, but we did not double-count these cases in estimating the overall odds ratio. Two studies17,18 published in 1999 (comprising a total of 98 cases) were not included in a previous meta-analysis of studies published before March 200014; they have been included in the 11 studies published between 2000 and 2002. Although three studies published after 2000,17,18,19 involving a total of 245 cases of coronary heart disease, reported results for deaths from cardiovascular causes rather than specifically from coronary heart disease, the majority of these deaths were likely to have been due to coronary heart disease. It was not possible to separate results for 77 cases of coronary revascularization from results for nonfatal myocardial infarction and death from coronary heart disease in another study.20 The odds ratios used were those reported in studies that had adjusted for age, sex, smoking status, and other established risk factors for coronary heart disease (such as blood lipid levels, blood pressure, body-mass index, and diabetes status). The "Other" category in "Sample" includes participants selected according to various criteria (e.g., the absence of a history of coronary disease in randomized trials). The Reykjavik Study provided separate estimates for men (732 cases with C-reactive protein values) and women (674 cases with C-reactive protein values). Information on the storage temperature used for samples was unavailable for two studies involving a total of 316 cases.26,31 Odds ratios involve comparisons of patients in the top third versus those in the bottom third of C-reactive protein concentrations. The horizontal lines represent 99 percent confidence intervals.
Updated Meta-Analysis
Twenty-two prospective studies of C-reactive protein (including the present study) have involved a total of 7068 patients, with a weighted mean age at entry of 57 years and a weighted mean follow-up of 12 years2,4,13,14,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33 (Table 3). All studies used high-sensitivity assays, and all but two19,32 reported adjustment for at least smoking status and some other established risk factors for coronary heart disease. There was evidence of heterogeneity between these studies (2=46, with 21 df; P=0.001), but with the exception of the date of publication (2=15, with 2 df; P<0.001), characteristics such as sample size (2=4.0, with 1 df; P=0.04), location (2=0.3, with 1 df; P=0.58), sampling method (2=5.2, with 1 df; P=0.02), sex of participants (2=3.4, with 2 df; P=0.18), mean duration of follow-up (2=1.6, with 1 df; P=0.20), and sample storage temperature (2=0.1, with 1 df; P=0.77) did not account for much of the overall heterogeneity (Figure 2).
Table 3. Comparison of Characteristics of Prospective Studies of C-Reactive Protein and Coronary Heart Disease (CHD) in Essentially General Populations.
The tendency toward more extreme findings in studies published before 2000 is consistent with the preferential publication of positive results in earlier studies. Restriction of analyses to the four studies involving more than 500 patients,4,14,20 comprising 4107 cases of coronary heart disease, should limit any such bias, and yielded a combined odds ratio of 1.49 (95 percent confidence interval, 1.37 to 1.62; 2=10.6, with 3 df; P=0.01). This value is somewhat smaller than the overall odds ratio of 1.58 (95 percent confidence interval, 1.48 to 1.68) derived from combining all 22 studies.
A previous meta-analysis6 of prospective studies of the effect of the erythrocyte sedimentation rate (based on 1703 cases of coronary heart disease) reported an odds ratio for coronary heart disease of about 1.3 (95 percent confidence interval, 1.2 to 1.5), and this estimate is reinforced by the odds ratio of 1.33 (95 percent confidence interval, 1.22 to 1.44) that we calculated in our updated meta-analysis (which involved an additional 2683 cases from a further two studies34). The present updated meta-analysis of prospective studies of von Willebrand factor (which adds 2445 cases of coronary heart disease to the previous total of 1524 cases) yielded an odds ratio of 1.23 (95 percent confidence interval, 1.14 to 1.33), which is probably weaker than the previous estimate of about 1.5 (95 percent confidence interval, 1.1 to 2.0).7
Discussion
We found that the decade-to-decade consistency of values for C-reactive protein, the erythrocyte sedimentation rate, and von Willebrand factor is similar to that of values for blood pressure and total serum cholesterol concentration, suggesting that these inflammatory markers are sufficiently stable for potential use in the long-term prediction of coronary heart disease. Our findings — reinforced by an updated meta-analysis — indicate, however, that the odds ratio for coronary heart disease in people with elevated C-reactive protein values is lower than that reported recently. Whereas a previous meta-analysis14 of studies published before 2000 (based on 1953 cases of coronary heart disease) reported an odds ratio for coronary heart disease of about 2.0 (95 percent confidence interval, 1.6 to 2.5), our updated meta-analysis, which adds 5115 cases of coronary heart disease from a further 12 studies, yielded an odds ratio of about 1.5 in a comparison of people with base-line values in the top third with those with base-line values in the bottom third for the population. Moreover, in comparison with major established risk factors (such as an increased total serum cholesterol concentration and cigarette smoking), the C-reactive protein concentration was a relatively moderate predictor of the risk of coronary heart disease and added only marginally to the predictive value of established risk factors for coronary heart disease. These findings suggest that recent recommendations regarding the use of measurements of C-reactive protein in the prediction of coronary heart disease may need to be reviewed.3
The potential limitations of our study merit careful consideration. The validity of our measurements is demonstrated by the reasonably high decade-to-decade consistency of C-reactive protein values recorded in paired samples from 379 participants (a level of stability that was at least as high as those recorded in previous studies with sampling intervals of just one to five years35,36,37,38). Further validation is suggested by the finding of the expected base-line associations of C-reactive protein with other inflammatory markers and with established coronary risk factors.
The mean values and the distributions of several established coronary risk factors (and the strength of their associations with the risk of coronary heart disease) in our study were generally similar to those reported in other western European populations.8 Therefore, although the relative homogeneity of the Reykjavik population should have minimized certain residual biases (such as that due to differences in socioeconomic status), the present findings should have wider relevance. Only total serum cholesterol concentrations were measured in the present study (rather than those of its subfractions, which have opposing effects on the risk of coronary heart disease), thereby underestimating the predictive ability of lipid concentrations (and potentially overestimating the adjusted predictive value of the C-reactive protein concentration).
No information was recorded on the use of aspirin and statins, which, like hormone-replacement treatment, may alter C-reactive protein values. However, fewer than 5 percent of the women in this study reported the use of such hormonal treatment during recruitment, and the use of aspirin and of statins was similarly uncommon in the general middle-aged population of Reykjavik between 1967 and 1991. We did not address the separate issues of the predictive value of inflammatory markers with respect to the risk of cardiac complications among patients recently hospitalized for acute coronary syndromes39 or the long-term risk of coronary heart disease in patients with a history of cardiovascular disease.14
As suggested by the statement of the Centers for Disease Control and Prevention and the American Heart Association,3 further clarification of the predictive value of C-reactive protein in coronary heart disease in general populations will require the pooling of studies on the basis of data for individual participants from each of the available prospective studies. Such a strategy will permit more complete adjustment for other risk factors and for within-person fluctuations of C-reactive protein levels, more precise quantification of the associations in particular subgroups (such as age-, sex-, and duration-specific associations as well as assessments of combinations of inflammatory markers), more reliable characterization of the shape of any dose–response relation, and more detailed investigation of potential sources of heterogeneity.
Supported by program grants from the British Heart Foundation (to Profs. Danesh and Lowe) and the Medical Research Council (to Prof. Pepys) and by the Raymond and Beverly Sackler Research Award in the Medical Sciences (to Prof. Danesh). Dr. Hirschfield was supported by a Medical Research Council Clinical Training Fellowship.
We are indebted to Prof. Simon Thompson and Dr. Gary Whitlock for helpful comments, to Kelsey Juzwishin for epidemiologic support, to the laboratory staff of the Icelandic Heart Association, to Drs. M. Thomas and D. Goodier of the Clinical Chemistry Department at the Royal Free Hospital, to Fiona Key and Karen Craig at the University Department of Medicine at the Glasgow Royal Infirmary, and to Roche Diagnostics for donating the C-reactive protein assay kits.
Source Information
From the Department of Public Health and Primary Care, University of Cambridge, Cambridge, United Kingdom (J.D., J.G.W.); the Centre for Amyloidosis and Acute Phase Proteins, Department of Medicine, Royal Free and University College Medical School, Royal Free Campus, London (G.M.H., M.B.P.); Roche Diagnostics, Tokyo, Japan (S.E.); the Icelandic Heart Association, Kopavogur, Iceland (G.E., V.G.); and the University Department of Medicine, Royal Infirmary, Glasgow, Scotland (A.R., G.D.O.L.).
Address reprint requests to Prof. Danesh at the Department of Public Health and Primary Care, Strangeways Site, Institute of Public Health, University of Cambridge, Cambridge CB1 8RN, United Kingdom.
References
Ross R. Atherosclerosis -- an inflammatory disease. N Engl J Med 1999;340:115-126.
Kuller LH, Tracy RP, Shaten J, Meilahn EN. Relation of C-reactive protein and coronary heart disease in the MRFIT nested case-control study. Am J Epidemiol 1996;144:537-547.
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.
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.
Clarke R, Shipley M, Lewington S, et al. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol 1999;150:341-353.
Danesh J, Collins R, Peto R, Lowe GDO. Haematocrit, viscosity, erythrocyte sedimentation rate: meta-analyses of prospective studies of coronary heart disease. Eur Heart J 2000;21:515-520.
Whincup PH, Danesh J, Walker M, et al. von Willebrand factor and coronary heart disease: prospective study and meta-analysis. Eur Heart J 2002;23:1764-1770.
Jónsdóttir LS, Sigfússon N, Gudnason V, Sigvaldason H, Thorgeirsson G. Do lipids, blood pressure, diabetes, and smoking confer equal risk of myocardial infarction in women as in men? The Reykjavik Study. J Cardiovasc Risk 2002;9:67-76.
Tulinius H, Sigfusson N, Sigvaldason H, Bjarnadottir K, Tryggvadottir L. Risk factors for malignant diseases: a cohort study on a population of 22,946 Icelanders. Cancer Epidemiol Biomarkers Prev 1997;6:863-873.
Rothman KJ. Modern epidemiology. Boston: Little, Brown, 1986:237.
Roberts WL, Moulton L, Law TC, et al. Evaluation of nine automated high-sensitivity C-reactive protein methods: implications for clinical and epidemiological applications. Clin Chem 2001;47:418-425.
Andresdottir MB, Sigfusson N, Sigvaldason H, Gudnason V. Erythrocyte sedimentation rate, an independent predictor of coronary heart disease in men and women: the Reykjavik Study. Am J Epidemiol 2003;158:844-851.
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.
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.
Shine B, de Beer FC, Pepys MB. Solid phase radioimmunoassays for human C-reactive protein. Clin Chim Acta 1981;117:13-23.
Ridker PM, Buring JE, Shih J, Matias M, Hennekens CH. Prospective study of C-reactive protein and the risk of future cardiovascular events among apparently healthy women. Circulation 1998;98:731-733.
Harris TB, Ferrucci L, Tracy RP, et al. Associations of elevated interleukin-6 and C-reactive protein levels with mortality in the elderly. Am J Med 1999;106:506-512.
Jager A, van Hinsbergh VWM, Kostense PJ, et al. von Willebrand factor, C-reactive protein, and 5-year mortality in diabetic and nondiabetic subjects: the Hoorn Study. Arterioscler Thromb Vasc Biol 1999;19:3071-3078.
Strandberg TE, Tilvis RS. C-reactive protein, cardiovascular risk factors, and mortality in a prospective study in the elderly. Arterioscler Thromb Vasc Biol 2000;20:1057-1060.
Packard CJ, O'Reilly DSJ, Caslake MJ, et al. Lipoprotein-associated phospholipase A2 as an independent predictor of coronary heart disease. N Engl J Med 2000;343:1148-1155.
Agewall S, Wikstrand J, Fagerberg B. Prothrombin fragment 1+2 is a risk factor for myocardial infarction in treated hypertensive men. J Hypertens 1998;16:537-541.
Ridker PM, Cushman M, Stampfer MJ, Tracy RP, Hennekens CH. Inflammation, aspirin, and the risk of cardiovascular disease in apparently healthy men. N Engl J Med 1997;336:973-979.
Lowe GDO, Yarnell JWG, Rumley A, Bainton D, Sweetnam PM. C-reactive protein, fibrin D-dimer, and incident ischemic heart disease in the Speedwell study: are inflammation and fibrin turnover linked in pathogenesis? Arterioscler Thromb Vasc Biol 2001;21:603-610.
Gram J, Bladbjerg EM, M?ller L, Sj?l A, Jespersen J. Tissue-type plasminogen activator and C-reactive protein in acute coronary heart disease: a nested case-control study. J Intern Med 2000;247:205-212.
Koenig W, Sund M, Fr?hlich 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;99:237-242.
Ridker PM, Rifai N, Clearfield M, et al. Measurement of C-reactive protein for the targeting of statin therapy in the primary prevention of acute coronary events. N Engl J Med 2001;344:1959-1965.
Pradhan AD, Manson JE, Rossouw JE, et al. Inflammatory biomarkers, hormone replacement therapy, and incident coronary heart disease: prospective analysis from the Women's Health Initiative observational study. JAMA 2002;288:980-987.
Mendall MA, Strachan DP, Butland BK, et al. C-reactive protein: relation to total mortality, cardiovascular mortality and cardiovascular risk factors in men. Eur Heart J 2000;21:1584-1590.
Pirro M, Bergeron J, Dagenais GR, et al. Age and duration of follow-up as modulators of the risk for ischemic heart disease associated with high plasma C-reactive protein levels in men. Arch Intern Med 2001;161:2474-2480.
Lowe GDO, Rumley A, Sweetnam PM, Yarnell JWG. Fibrin D-dimer, C-reactive protein and risk of ischaemic heart disease: the Speedwell study. Blood Coagul Fibrinolysis 1999;10:Suppl 1:S92-S93.
Witherell H, Smith KL, Ley C, et al. Helicobacter pylori infection, C-reactive protein and risk for myocardial infarction: a prospective study. Gastroenterology 1999;116:A355-A355. abstract.
Roivainen M, Viik-Kajander M, Palosuo T, et al. Infections, inflammation, and the risk of coronary heart disease. Circulation 2000;101:252-257.
Tracy RP, Lemaitre RN, Psaty BM, et al. Relationship of C-reactive protein to risk of cardiovascular disease in the elderly: results from the Cardiovascular Health Study and the Rural Health Promotion Project. Arterioscler Thromb Vasc Biol 1997;17:1121-1127.
Erikssen G, Liest?l K, Bj?rnholt JV, Stormorken H, Thaulow E, Erikssen J. Erythrocyte sedimentation rate: a possible marker of atherosclerosis and a strong predictor of coronary heart disease mortality. Eur Heart J 2000;21:1614-1620.
Ridker PM, Rifai N, Pfeffer MA, et al. Inflammation, pravastatin, and the risk of coronary events after myocardial infarction in patients with average cholesterol levels. Circulation 1998;98:839-844.
Ockene IS, Matthews CE, Rifai N, Ridker PM, Reed G, Stanek E. Variability and classification accuracy of serial high-sensitivity C-reactive protein measurements in healthy adults. Clin Chem 2001;47:444-450.
Kayaba K, Ishikawa S, Gotoh T, et al. Five-year intra-individual variability in C-reactive protein levels in a Japanese population-based study: the Jichi Medical School Cohort Study at Yamato, 1993-1998. Jpn Circ J 2000;64:303-308.
Koenig W, Sund M, Fr?hlich M, L?wel H, Hutchinson WL, Pepys MB. Refinement of the association of serum C-reactive protein concentration and coronary heart disease risk by correction for within-subject variation over time: the MONICA Augsburg studies, 1984 and 1987. Am J Epidemiol 2003;158:357-364.
Pepys MB, Hirschfield GM. C-reactive protein: a critical update. J Clin Invest 2003;111:1805-1812.(John Danesh, M.B., Ch.B.,)
Background C-reactive protein is an inflammatory marker believed to be of value in the prediction of coronary events. We report data from a large study of C-reactive protein and other circulating inflammatory markers, as well as updated meta-analyses, to evaluate their relevance to the prediction of coronary heart disease.
Methods Measurements were made in samples obtained at base line from up to 2459 patients who had a nonfatal myocardial infarction or died of coronary heart disease during the study and from up to 3969 controls without a coronary heart disease event in the Reykjavik prospective study of 18,569 participants. Measurements were made in paired samples obtained an average of 12 years apart from 379 of these participants in order to quantify within-person fluctuations in inflammatory marker levels.
Results The long-term stability of C-reactive protein values (within-person correlation coefficient, 0.59; 95 percent confidence interval, 0.52 to 0.66) was similar to that of both blood pressure and total serum cholesterol. After adjustment for base-line values for established risk factors, the odds ratio for coronary heart disease was 1.45 (95 percent confidence interval, 1.25 to 1.68) in a comparison of participants in the top third of the group with respect to base-line C-reactive protein values with those in the bottom third, and similar overall findings were observed in an updated meta-analysis involving a total of 7068 patients with coronary heart disease. By comparison, the odds ratios in the Reykjavik Study for coronary heart disease were somewhat weaker for the erythrocyte sedimentation rate (1.30; 95 percent confidence interval, 1.13 to 1.51) and the von Willebrand factor concentration (1.11; 95 percent confidence interval, 0.97 to 1.27) but generally stronger for established risk factors, such as an increased total cholesterol concentration (2.35; 95 percent confidence interval, 2.03 to 2.74) and cigarette smoking (1.87; 95 percent confidence interval, 1.62 to 2.16).
Conclusions C-reactive protein is a relatively moderate predictor of coronary heart disease. Recommendations regarding its use in predicting the likelihood of coronary heart disease may need to be reviewed.
Since atherosclerosis may, in part, be an inflammatory disease,1 circulating factors related to inflammation may be predictors of cardiovascular disease in general populations.2 A recent statement from the Centers for Disease Control and Prevention and the American Heart Association concluded that it is reasonable to measure C-reactive protein, a sensitive circulating marker of inflammation, as an adjunct to the measurement of established risk factors in order to assess the risk of coronary heart disease.3 The report acknowledged, however, that the epidemiologic data to support this view were not entirely consistent and recommended that larger prospective studies be conducted to improve the reliability of the evidence.
We measured C-reactive protein concentrations in approximately 2400 patients with coronary heart disease diagnosed since their enrollment in the cohort and approximately 4000 controls nested within the Reykjavik Study, a prospective cohort study of about 19,000 middle-aged men and women without a history of myocardial infarction. The number of cases of coronary heart disease in this cohort was about four times as great as in the largest previous study4 and should reduce the scope for random error in our estimates. We also assessed the effect of within-person variation in the concentrations of inflammatory markers5 in serial blood samples obtained over a period of several years in several hundred participants. To compare the predictive value of the C-reactive protein concentration with that of some other inflammatory markers studied in coronary heart disease, we also analyzed the erythrocyte sedimentation rate and circulating concentrations of von Willebrand factor, each of which can also fluctuate considerably in acute-phase inflammatory responses.6,7 To help put the new data in context, we updated meta-analyses of previous relevant studies of each of these inflammatory markers.
Methods
Patients and Controls
The Reykjavik Study, initiated in 1967 as a prospective study of cardiovascular disease, has been described in detail previously.8 All men born between 1907 and 1934 and all women born between 1908 and 1935 who were residents of Reykjavik, Iceland, and its adjacent communities on December 1, 1966, were identified in the national population register and then invited to participate in the study during five stages of recruitment between 1967 and 1991. A total of 8888 men and 9681 women without a history of myocardial infarction were enrolled, reflecting a response rate of 72 percent.9
Nurses administered questionnaires, made physical measurements, performed spirometry and electrocardiography, and collected venous blood samples after an overnight fast for the measurement of the erythrocyte sedimentation rate and to prepare aliquots of serum, which were stored at –20°C for subsequent analysis. All participants have subsequently been monitored with respect to death from any cause and the occurrence of major cardiovascular conditions, with a total loss to follow-up of only about 0.6 percent of participants.9
A total of 2459 men and women with available serum samples had major coronary events between the beginning of follow-up and December 31, 1995, for a mean (±SD) duration of follow-up of 17.5±8.7 years, as compared with 20.6±8.2 years among controls. Among the men, 1073 deaths from coronary heart disease and 701 nonfatal myocardial infarctions were recorded (564 confirmed and 137 possible myocardial infarctions), and among the women, 385 died of coronary heart disease and 300 had a nonfatal myocardial infarction (237 confirmed and 63 possible myocardial infarctions). Deaths from coronary heart disease were ascertained from central registers on the basis of a death certificate listing an International Classification of Diseases code of 410 through 414, and the diagnosis of nonfatal myocardial infarction was based on the criteria of the Monitoring Trends and Determinants in Cardiovascular Disease study.
We selected 3969 control subjects from among the participants who had survived to the end of the study period without having a myocardial infarction. The controls were frequency-matched to the patients with respect to the calendar year of recruitment, sex, and age (in five-year increments).10
The National Bioethics Committee and the Data Protection Authority of Iceland approved the study protocol. All participants provided informed consent.
Laboratory Methods
Laboratory measurements were made without knowledge of the participants' disease status, and thus samples from patients and controls were randomly distributed among assay plates. Concentrations of C-reactive protein were measured by latex-enhanced immunoturbidimetry, with a lower limit of detection of 0.02 mg per liter (Roche Diagnostics).11 The variation in C-reactive protein values within runs was less than 1 percent, and the between-day variability was 1 percent at a concentration of 14 mg per liter and 3.7 percent at a concentration of 3.8 mg per liter. The concentration of von Willebrand factor was determined by means of a sensitive enzyme immunoassay. We also determined the concentration of von Willebrand factor in paired plasma and serum samples from 56 healthy persons from another study and found close agreement between plasma and serum values (correlation coefficient, 0.94).7 The Wintrobe method was used to measure the erythrocyte sedimentation rate in fresh blood samples obtained at the time of base-line venesection.6 Other biochemical and hematologic measurements involved the use of standard assays, as previously described.8 Measurements were made in pairs of samples obtained from 379 participants a mean of about 12 years apart. Data on erythrocyte sedimentation rate from the Reykjavik Study have been reported previously.12
Statistical Analysis
Comparisons between patients and controls were made by means of unmatched stratified logistic regression fitted according to the unconditional maximum likelihood (Stata software, version 7). To maximize the ability to compare our results with those of previous reports, primary analyses of values of C-reactive protein, erythrocyte sedimentation rate, and von Willebrand factor were prespecified to compare extreme thirds of patients and controls with respect to the distribution of values in the controls. Subsidiary analyses involved other cutoff values. Odds ratios were sequentially adjusted for the following variables: age, sex, calendar year of enrollment, smoking status, systolic blood pressure, total cholesterol level, triglyceride level, body-mass index (the weight in kilograms divided by the square of the height in meters), forced expiratory volume in one second, presence or absence of diabetes, socioeconomic status, and the concentrations of other markers of inflammation.
To estimate the discriminative value of predictive models, we calculated the areas under the receiver-operating-characteristic curve, in order to determine whether the sequential addition of data on inflammatory markers increased the predictive value of major established coronary risk factors, as described previously.13 We performed meta-analyses of studies published before January 2003 that included essentially general populations (i.e., cohorts not selected on the basis of preexisting disease) with more than a year of follow-up, using search, abstraction, and data-synthesis methods that have been described previously and using nonfatal myocardial infarction or death from coronary heart disease as end points.6,7,14 We combined the results of the studies by using inverse variance-weighted averages of logarithmic odds ratios. Heterogeneity was assessed by means of standard 2 tests. Odds ratios are given with 95 percent confidence intervals, and two-sided P values are reported. Since previous studies have reported on the predictive values of single base-line measurements of inflammatory markers with respect to coronary heart disease, odds ratios have not been corrected for regression dilution in the present study, so as to allow direct comparisons with previous work.5
Results
The mean age at the time of the coronary heart disease event was 70.2±9.7 years. There were significant differences between patients and controls with respect to established coronary risk factors, such as smoking status, body-mass index, blood pressure, and serum lipid concentrations (Table 1).
Table 1. Base-Line Characteristics of the Patients with Coronary Heart Disease and Controls.
Base-Line Associations and Long-Term Stability of Inflammatory Markers
The partial correlation coefficients (adjusted for age, sex, calendar year of recruitment, and smoking status) for C-reactive protein, on the one hand, and the erythrocyte sedimentation rate and von Willebrand factor, on the other, were 0.38 and 0.18, respectively (P<0.001 for each comparison), and the partial correlation coefficient for the erythrocyte sedimentation rate and the von Willebrand factor concentration was 0.17 (P<0.001). A higher C-reactive protein concentration was significantly associated with cigarette smoking (P<0.001), an increased body-mass index (P<0.001), a low forced expiratory volume in one second (P<0.001), and an increased triglyceride concentration (P<0.001) (data not shown). Higher values for the erythrocyte sedimentation rate were significantly associated with older age (P<0.001), female sex (P<0.001), a low hemoglobin value (P<0.001), a low hematocrit (P<0.001), an elevated serum uric acid concentration (P<0.001), a low forced expiratory volume in one second (P<0.001), and smoking (P<0.001). A higher von Willebrand factor concentration was significantly associated with older age (P<0.001) and smoking (P<0.001).
Among 379 participants who provided paired blood samples, the within-person correlation coefficients for C-reactive protein, erythrocyte sedimentation rate, and von Willebrand factor were 0.59 (95 percent confidence interval, 0.52 to 0.66), 0.67 (95 percent confidence interval, 0.61 to 0.73), and 0.57 (95 percent confidence interval, 0.50 to 0.64), respectively. These values were similar with respect to long-term consistency to the values for systolic blood pressure (correlation coefficient, 0.66; 95 percent confidence interval, 0.60 to 0.72), diastolic blood pressure (correlation coefficient, 0.53; 95 percent confidence interval, 0.46 to 0.60), and total serum cholesterol (correlation coefficient, 0.60; 95 percent confidence interval, 0.54 to 0.66).
Inflammatory Markers and Incident Coronary Heart Disease
The odds ratio for coronary heart disease was 1.92 (95 percent confidence interval, 1.68 to 2.18; 2=105, with 1 df) among patients with values in the top third (cutoff value, 2.0 mg per liter), as compared with the bottom third (cutoff value, 0.78 mg per liter), of base-line C-reactive protein concentrations in the control group. The odds ratio fell to 1.45 (95 percent confidence interval, 1.25 to 1.68; 2=28, with 1 df) after adjustment for smoking status, other established coronary risk factors, and indicators of socioeconomic status (Table 2). Comparisons between the top and bottom thirds of patients and controls with respect to the other markers gave the following adjusted odds ratios for coronary heart disease: for erythrocyte sedimentation rate (cutoff value of 10 mm in first hour of measurement for the top third and 4 mm in first hour for the bottom third), 1.30 (95 percent confidence interval, 1.13 to 1.51; 2=13, with 1 df), and for von Willebrand factor (cutoff value of 124 IU per deciliter for the top third and 88 IU per deciliter for the bottom third), 1.11 (95 percent confidence interval, 0.97 to 1.27; 2=26, with 1 df) (Table 2 and Figure 1). The calculated areas under receiver-operating-characteristic curves indicate that information on the C-reactive protein concentration (and the other inflammatory markers that were assessed) provided comparatively little additional predictive value over that provided by assessment of major established risk factors (Figure 1).
Table 2. Relative Odds of Coronary Heart Disease (CHD) among Patients Who Had Levels of Inflammatory Markers in the Top Third of the Distribution of Values for Controls, as Compared with Those Who Had Values in the Bottom Third of This Distribution.
Figure 1. Odds Ratios for Coronary Heart Disease among 2459 Patients with Coronary Heart Disease and 3969 Controls.
Comparisons are between patients and controls with values in the top third and those in the bottom third of the distribution of values for controls, except for comparisons involving smoking status. Squares denote odds ratios, and horizontal lines represent 95 percent confidence intervals. The information plotted in this figure is based on odds ratios listed in the next-to-last column of Table 2. Logistic-regression analysis was used to calculate the areas under the receiver-operating-characteristic (ROC) curve after adjustment for age, sex, and period, with data on major established risk factors and inflammatory markers added to the model in the order of the strength of each variable's association with coronary heart disease.
These findings were not materially changed in analyses restricted to the 2083 patients without evidence of coronary heart disease at base line (Table 2), to the 2206 patients with C-reactive protein values who had a confirmed myocardial infarction or died of coronary heart disease, or to the participants without evidence of acute-phase reactions at the base-line examination (i.e., this analysis excluded 132 patients and 152 controls with a C-reactive protein concentration of more than 10 mg per liter15 or an erythrocyte sedimentation rate of more than 30 mm during the first hour). The findings were also unaffected by changes in the cutoff values (e.g., analyses of quarters or fifths, or according to increases of 1 SD) (Table 2).
Associations between the C-reactive protein concentration and the risk of coronary heart disease did not vary significantly according to established risk factors, such as smoking or increased blood lipid concentrations, blood pressure, or body-mass index (data not shown). An exploratory analysis suggested the possibility of more extreme odds ratios among the 1049 patients who died of coronary heart disease or had a nonfatal myocardial infarction within 10 years after enrollment (odds ratio, 1.84; 95 percent confidence interval, 1.49 to 2.28), as compared with the 1357 patients who had such an event after the first decade (odds ratio, 1.26; 95 percent confidence interval, 1.05 to 1.51). Such a trend, however, was not observed in the updated meta-analysis, described below, which was based on published data from 22 studies2,4,13,14,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33 (Figure 2). Therefore, it requires further examination involving larger numbers of participants with individual data. Such analysis is also required for a reliable characterization of the shape of the association between C-reactive protein and coronary heart disease.
Figure 2. Twenty-Two Prospective Studies of the Association of C-Reactive Protein Concentrations with the Risk of Coronary Heart Disease (CHD) in Essentially General Populations, Grouped According to Several Study Characteristics.
One of the 11 studies published before 2000 was updated in 200213,16; hence, data on 85 cases from this study contributed to two subtotals, but we did not double-count these cases in estimating the overall odds ratio. Two studies17,18 published in 1999 (comprising a total of 98 cases) were not included in a previous meta-analysis of studies published before March 200014; they have been included in the 11 studies published between 2000 and 2002. Although three studies published after 2000,17,18,19 involving a total of 245 cases of coronary heart disease, reported results for deaths from cardiovascular causes rather than specifically from coronary heart disease, the majority of these deaths were likely to have been due to coronary heart disease. It was not possible to separate results for 77 cases of coronary revascularization from results for nonfatal myocardial infarction and death from coronary heart disease in another study.20 The odds ratios used were those reported in studies that had adjusted for age, sex, smoking status, and other established risk factors for coronary heart disease (such as blood lipid levels, blood pressure, body-mass index, and diabetes status). The "Other" category in "Sample" includes participants selected according to various criteria (e.g., the absence of a history of coronary disease in randomized trials). The Reykjavik Study provided separate estimates for men (732 cases with C-reactive protein values) and women (674 cases with C-reactive protein values). Information on the storage temperature used for samples was unavailable for two studies involving a total of 316 cases.26,31 Odds ratios involve comparisons of patients in the top third versus those in the bottom third of C-reactive protein concentrations. The horizontal lines represent 99 percent confidence intervals.
Updated Meta-Analysis
Twenty-two prospective studies of C-reactive protein (including the present study) have involved a total of 7068 patients, with a weighted mean age at entry of 57 years and a weighted mean follow-up of 12 years2,4,13,14,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33 (Table 3). All studies used high-sensitivity assays, and all but two19,32 reported adjustment for at least smoking status and some other established risk factors for coronary heart disease. There was evidence of heterogeneity between these studies (2=46, with 21 df; P=0.001), but with the exception of the date of publication (2=15, with 2 df; P<0.001), characteristics such as sample size (2=4.0, with 1 df; P=0.04), location (2=0.3, with 1 df; P=0.58), sampling method (2=5.2, with 1 df; P=0.02), sex of participants (2=3.4, with 2 df; P=0.18), mean duration of follow-up (2=1.6, with 1 df; P=0.20), and sample storage temperature (2=0.1, with 1 df; P=0.77) did not account for much of the overall heterogeneity (Figure 2).
Table 3. Comparison of Characteristics of Prospective Studies of C-Reactive Protein and Coronary Heart Disease (CHD) in Essentially General Populations.
The tendency toward more extreme findings in studies published before 2000 is consistent with the preferential publication of positive results in earlier studies. Restriction of analyses to the four studies involving more than 500 patients,4,14,20 comprising 4107 cases of coronary heart disease, should limit any such bias, and yielded a combined odds ratio of 1.49 (95 percent confidence interval, 1.37 to 1.62; 2=10.6, with 3 df; P=0.01). This value is somewhat smaller than the overall odds ratio of 1.58 (95 percent confidence interval, 1.48 to 1.68) derived from combining all 22 studies.
A previous meta-analysis6 of prospective studies of the effect of the erythrocyte sedimentation rate (based on 1703 cases of coronary heart disease) reported an odds ratio for coronary heart disease of about 1.3 (95 percent confidence interval, 1.2 to 1.5), and this estimate is reinforced by the odds ratio of 1.33 (95 percent confidence interval, 1.22 to 1.44) that we calculated in our updated meta-analysis (which involved an additional 2683 cases from a further two studies34). The present updated meta-analysis of prospective studies of von Willebrand factor (which adds 2445 cases of coronary heart disease to the previous total of 1524 cases) yielded an odds ratio of 1.23 (95 percent confidence interval, 1.14 to 1.33), which is probably weaker than the previous estimate of about 1.5 (95 percent confidence interval, 1.1 to 2.0).7
Discussion
We found that the decade-to-decade consistency of values for C-reactive protein, the erythrocyte sedimentation rate, and von Willebrand factor is similar to that of values for blood pressure and total serum cholesterol concentration, suggesting that these inflammatory markers are sufficiently stable for potential use in the long-term prediction of coronary heart disease. Our findings — reinforced by an updated meta-analysis — indicate, however, that the odds ratio for coronary heart disease in people with elevated C-reactive protein values is lower than that reported recently. Whereas a previous meta-analysis14 of studies published before 2000 (based on 1953 cases of coronary heart disease) reported an odds ratio for coronary heart disease of about 2.0 (95 percent confidence interval, 1.6 to 2.5), our updated meta-analysis, which adds 5115 cases of coronary heart disease from a further 12 studies, yielded an odds ratio of about 1.5 in a comparison of people with base-line values in the top third with those with base-line values in the bottom third for the population. Moreover, in comparison with major established risk factors (such as an increased total serum cholesterol concentration and cigarette smoking), the C-reactive protein concentration was a relatively moderate predictor of the risk of coronary heart disease and added only marginally to the predictive value of established risk factors for coronary heart disease. These findings suggest that recent recommendations regarding the use of measurements of C-reactive protein in the prediction of coronary heart disease may need to be reviewed.3
The potential limitations of our study merit careful consideration. The validity of our measurements is demonstrated by the reasonably high decade-to-decade consistency of C-reactive protein values recorded in paired samples from 379 participants (a level of stability that was at least as high as those recorded in previous studies with sampling intervals of just one to five years35,36,37,38). Further validation is suggested by the finding of the expected base-line associations of C-reactive protein with other inflammatory markers and with established coronary risk factors.
The mean values and the distributions of several established coronary risk factors (and the strength of their associations with the risk of coronary heart disease) in our study were generally similar to those reported in other western European populations.8 Therefore, although the relative homogeneity of the Reykjavik population should have minimized certain residual biases (such as that due to differences in socioeconomic status), the present findings should have wider relevance. Only total serum cholesterol concentrations were measured in the present study (rather than those of its subfractions, which have opposing effects on the risk of coronary heart disease), thereby underestimating the predictive ability of lipid concentrations (and potentially overestimating the adjusted predictive value of the C-reactive protein concentration).
No information was recorded on the use of aspirin and statins, which, like hormone-replacement treatment, may alter C-reactive protein values. However, fewer than 5 percent of the women in this study reported the use of such hormonal treatment during recruitment, and the use of aspirin and of statins was similarly uncommon in the general middle-aged population of Reykjavik between 1967 and 1991. We did not address the separate issues of the predictive value of inflammatory markers with respect to the risk of cardiac complications among patients recently hospitalized for acute coronary syndromes39 or the long-term risk of coronary heart disease in patients with a history of cardiovascular disease.14
As suggested by the statement of the Centers for Disease Control and Prevention and the American Heart Association,3 further clarification of the predictive value of C-reactive protein in coronary heart disease in general populations will require the pooling of studies on the basis of data for individual participants from each of the available prospective studies. Such a strategy will permit more complete adjustment for other risk factors and for within-person fluctuations of C-reactive protein levels, more precise quantification of the associations in particular subgroups (such as age-, sex-, and duration-specific associations as well as assessments of combinations of inflammatory markers), more reliable characterization of the shape of any dose–response relation, and more detailed investigation of potential sources of heterogeneity.
Supported by program grants from the British Heart Foundation (to Profs. Danesh and Lowe) and the Medical Research Council (to Prof. Pepys) and by the Raymond and Beverly Sackler Research Award in the Medical Sciences (to Prof. Danesh). Dr. Hirschfield was supported by a Medical Research Council Clinical Training Fellowship.
We are indebted to Prof. Simon Thompson and Dr. Gary Whitlock for helpful comments, to Kelsey Juzwishin for epidemiologic support, to the laboratory staff of the Icelandic Heart Association, to Drs. M. Thomas and D. Goodier of the Clinical Chemistry Department at the Royal Free Hospital, to Fiona Key and Karen Craig at the University Department of Medicine at the Glasgow Royal Infirmary, and to Roche Diagnostics for donating the C-reactive protein assay kits.
Source Information
From the Department of Public Health and Primary Care, University of Cambridge, Cambridge, United Kingdom (J.D., J.G.W.); the Centre for Amyloidosis and Acute Phase Proteins, Department of Medicine, Royal Free and University College Medical School, Royal Free Campus, London (G.M.H., M.B.P.); Roche Diagnostics, Tokyo, Japan (S.E.); the Icelandic Heart Association, Kopavogur, Iceland (G.E., V.G.); and the University Department of Medicine, Royal Infirmary, Glasgow, Scotland (A.R., G.D.O.L.).
Address reprint requests to Prof. Danesh at the Department of Public Health and Primary Care, Strangeways Site, Institute of Public Health, University of Cambridge, Cambridge CB1 8RN, United Kingdom.
References
Ross R. Atherosclerosis -- an inflammatory disease. N Engl J Med 1999;340:115-126.
Kuller LH, Tracy RP, Shaten J, Meilahn EN. Relation of C-reactive protein and coronary heart disease in the MRFIT nested case-control study. Am J Epidemiol 1996;144:537-547.
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.
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.
Clarke R, Shipley M, Lewington S, et al. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol 1999;150:341-353.
Danesh J, Collins R, Peto R, Lowe GDO. Haematocrit, viscosity, erythrocyte sedimentation rate: meta-analyses of prospective studies of coronary heart disease. Eur Heart J 2000;21:515-520.
Whincup PH, Danesh J, Walker M, et al. von Willebrand factor and coronary heart disease: prospective study and meta-analysis. Eur Heart J 2002;23:1764-1770.
Jónsdóttir LS, Sigfússon N, Gudnason V, Sigvaldason H, Thorgeirsson G. Do lipids, blood pressure, diabetes, and smoking confer equal risk of myocardial infarction in women as in men? The Reykjavik Study. J Cardiovasc Risk 2002;9:67-76.
Tulinius H, Sigfusson N, Sigvaldason H, Bjarnadottir K, Tryggvadottir L. Risk factors for malignant diseases: a cohort study on a population of 22,946 Icelanders. Cancer Epidemiol Biomarkers Prev 1997;6:863-873.
Rothman KJ. Modern epidemiology. Boston: Little, Brown, 1986:237.
Roberts WL, Moulton L, Law TC, et al. Evaluation of nine automated high-sensitivity C-reactive protein methods: implications for clinical and epidemiological applications. Clin Chem 2001;47:418-425.
Andresdottir MB, Sigfusson N, Sigvaldason H, Gudnason V. Erythrocyte sedimentation rate, an independent predictor of coronary heart disease in men and women: the Reykjavik Study. Am J Epidemiol 2003;158:844-851.
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.
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.
Shine B, de Beer FC, Pepys MB. Solid phase radioimmunoassays for human C-reactive protein. Clin Chim Acta 1981;117:13-23.
Ridker PM, Buring JE, Shih J, Matias M, Hennekens CH. Prospective study of C-reactive protein and the risk of future cardiovascular events among apparently healthy women. Circulation 1998;98:731-733.
Harris TB, Ferrucci L, Tracy RP, et al. Associations of elevated interleukin-6 and C-reactive protein levels with mortality in the elderly. Am J Med 1999;106:506-512.
Jager A, van Hinsbergh VWM, Kostense PJ, et al. von Willebrand factor, C-reactive protein, and 5-year mortality in diabetic and nondiabetic subjects: the Hoorn Study. Arterioscler Thromb Vasc Biol 1999;19:3071-3078.
Strandberg TE, Tilvis RS. C-reactive protein, cardiovascular risk factors, and mortality in a prospective study in the elderly. Arterioscler Thromb Vasc Biol 2000;20:1057-1060.
Packard CJ, O'Reilly DSJ, Caslake MJ, et al. Lipoprotein-associated phospholipase A2 as an independent predictor of coronary heart disease. N Engl J Med 2000;343:1148-1155.
Agewall S, Wikstrand J, Fagerberg B. Prothrombin fragment 1+2 is a risk factor for myocardial infarction in treated hypertensive men. J Hypertens 1998;16:537-541.
Ridker PM, Cushman M, Stampfer MJ, Tracy RP, Hennekens CH. Inflammation, aspirin, and the risk of cardiovascular disease in apparently healthy men. N Engl J Med 1997;336:973-979.
Lowe GDO, Yarnell JWG, Rumley A, Bainton D, Sweetnam PM. C-reactive protein, fibrin D-dimer, and incident ischemic heart disease in the Speedwell study: are inflammation and fibrin turnover linked in pathogenesis? Arterioscler Thromb Vasc Biol 2001;21:603-610.
Gram J, Bladbjerg EM, M?ller L, Sj?l A, Jespersen J. Tissue-type plasminogen activator and C-reactive protein in acute coronary heart disease: a nested case-control study. J Intern Med 2000;247:205-212.
Koenig W, Sund M, Fr?hlich 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;99:237-242.
Ridker PM, Rifai N, Clearfield M, et al. Measurement of C-reactive protein for the targeting of statin therapy in the primary prevention of acute coronary events. N Engl J Med 2001;344:1959-1965.
Pradhan AD, Manson JE, Rossouw JE, et al. Inflammatory biomarkers, hormone replacement therapy, and incident coronary heart disease: prospective analysis from the Women's Health Initiative observational study. JAMA 2002;288:980-987.
Mendall MA, Strachan DP, Butland BK, et al. C-reactive protein: relation to total mortality, cardiovascular mortality and cardiovascular risk factors in men. Eur Heart J 2000;21:1584-1590.
Pirro M, Bergeron J, Dagenais GR, et al. Age and duration of follow-up as modulators of the risk for ischemic heart disease associated with high plasma C-reactive protein levels in men. Arch Intern Med 2001;161:2474-2480.
Lowe GDO, Rumley A, Sweetnam PM, Yarnell JWG. Fibrin D-dimer, C-reactive protein and risk of ischaemic heart disease: the Speedwell study. Blood Coagul Fibrinolysis 1999;10:Suppl 1:S92-S93.
Witherell H, Smith KL, Ley C, et al. Helicobacter pylori infection, C-reactive protein and risk for myocardial infarction: a prospective study. Gastroenterology 1999;116:A355-A355. abstract.
Roivainen M, Viik-Kajander M, Palosuo T, et al. Infections, inflammation, and the risk of coronary heart disease. Circulation 2000;101:252-257.
Tracy RP, Lemaitre RN, Psaty BM, et al. Relationship of C-reactive protein to risk of cardiovascular disease in the elderly: results from the Cardiovascular Health Study and the Rural Health Promotion Project. Arterioscler Thromb Vasc Biol 1997;17:1121-1127.
Erikssen G, Liest?l K, Bj?rnholt JV, Stormorken H, Thaulow E, Erikssen J. Erythrocyte sedimentation rate: a possible marker of atherosclerosis and a strong predictor of coronary heart disease mortality. Eur Heart J 2000;21:1614-1620.
Ridker PM, Rifai N, Pfeffer MA, et al. Inflammation, pravastatin, and the risk of coronary events after myocardial infarction in patients with average cholesterol levels. Circulation 1998;98:839-844.
Ockene IS, Matthews CE, Rifai N, Ridker PM, Reed G, Stanek E. Variability and classification accuracy of serial high-sensitivity C-reactive protein measurements in healthy adults. Clin Chem 2001;47:444-450.
Kayaba K, Ishikawa S, Gotoh T, et al. Five-year intra-individual variability in C-reactive protein levels in a Japanese population-based study: the Jichi Medical School Cohort Study at Yamato, 1993-1998. Jpn Circ J 2000;64:303-308.
Koenig W, Sund M, Fr?hlich M, L?wel H, Hutchinson WL, Pepys MB. Refinement of the association of serum C-reactive protein concentration and coronary heart disease risk by correction for within-subject variation over time: the MONICA Augsburg studies, 1984 and 1987. Am J Epidemiol 2003;158:357-364.
Pepys MB, Hirschfield GM. C-reactive protein: a critical update. J Clin Invest 2003;111:1805-1812.(John Danesh, M.B., Ch.B.,)