Background
Despite recent improvements in the management of cardiovascular risk factors [
1‐
4], cardiovascular disease (CVD) remains the leading cause of death in Europe [
5,
6]. Well-established risk factors for CVD include older age, male sex, smoking, elevated cholesterol levels, hypertension and diabetes mellitus [
7]. Assessment of these factors can be used to estimate patients’ global cardiovascular risk. Two widely used cardiovascular risk estimation systems are the Systematic Coronary Risk Evaluation (SCORE), developed in Europe to evaluate 10-year risk of cardiovascular mortality [
8,
9], and the Framingham Risk Score (FRS), developed in the USA to estimate 10-year risk of any cardiovascular event [
10].
Global cardiovascular risk estimation is critical for selecting appropriate management options in apparently healthy, asymptomatic individuals according to current clinical guidelines. For example, the European guidelines on CVD prevention recommend prescription of blood pressure medication or lipid-lowering drugs for patients with hypertension or elevated serum cholesterol levels, respectively, but only for those estimated to be at high 10-year cardiovascular risk (cardiovascular mortality of ≥5% according to SCORE [
7] or risk of major CVD event of ≥20% according to the FRS). Exceptions are patients who are already known to be at high cardiovascular risk owing to having a history of CVD, diabetes mellitus or chronic kidney disease, or having very high levels of single risk factors [
7].
In addition to traditional cardiovascular risk factors, slight elevations of the inflammatory marker C-reactive protein (CRP) may also indicate increased cardiovascular risk [
11‐
13]. Measurement of CRP levels may help to identify patients who are at lower or greater cardiovascular risk than is currently appreciated [
13,
14]. A recent clinical trial of rosuvastatin in patients who have low levels of low-density lipoprotein cholesterol (LDL-C) (<130 mg/dL [<3.4 mmol/L]) but elevated levels of CRP (≥2.0 mg/L) showed a 44% reduction in cardiovascular event rates in the rosuvastatin treatment group compared with placebo (the Justification for the Use of Statins in Prevention: an Interventional Trial Evaluating Rosuvastatin [JUPITER], ClinicalTrials.gov Identifier: NCT00239681), translating to an absolute risk reduction of 1.2%, and a number needed to treat of 95 for 2 years [
15]. The earlier Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS) had shown very low event rates and no evidence of beneficial effects of statin treatment among lower-risk primary prevention patients who had low levels of both LDL-C and CRP [
16]. Elevated CRP levels may thus distinguish a group of patients who are at higher cardiovascular risk than predicted according to conventional risk assessment tools, and who might therefore derive a greater than expected benefit from statin therapy.
The European Study on Cardiovascular Risk Prevention and Management in Usual Daily Practice (EURIKA; ClinicalTrials.gov Identifier: NCT00882336) was recently conducted to assess the management of cardiovascular risk factors in primary care in Europe [
17‐
19]. We have carried out a
post hoc analysis of this study to assess the prevalence of elevated CRP levels in patients with one or more traditional cardiovascular risk factors, across a range of levels of conventionally estimated global cardiovascular risk. In particular, patients were considered who were estimated to be at intermediate levels of cardiovascular risk according to SCORE and FRS, who did not have diabetes mellitus (a marker of high cardiovascular risk) and who were not already receiving statins (for whom a treatment decision had already been made).
Methods
Study design and participants
The EURIKA study (ClinicalTrials.gov Identifier: NCT00882336) was conducted in 12 European countries (Austria, Belgium, France, Germany, Greece, Norway, Russia, Spain, Sweden, Switzerland, Turkey and the UK), selected to represent a broad spectrum of CVD risk across Europe, as well as a variety of different healthcare systems. Data collection started in May 2009 and ended in January 2010, with a 3-month data-collection period for each country. The study protocol was approved by the appropriate clinical research ethics committees in each participating country (lead ethics committee for host institution and sponsor: Comité Ético de Investigación Clínica (CEIC) del Hospital Universitario La Paz [Spain]), and all participating patients provided signed informed consent.
The methods for the study have been reported elsewhere [
17]. Briefly, the study sample was selected in a two-stage process that involved the random selection of both physicians and their patients [
17,
19]. In the first stage, primary care practitioners (PCPs) and specialists involved in CVD prevention (including cardiologists, endocrinologists, and internal medicine specialists) were randomly selected for invitation to participate using the OneKey database (Cegedim Dendrite, Boulogne-Billancourt, France) [
20]. In total, 809 physicians (approximately 60 per country) agreed to participate in EURIKA, 64% of whom were PCPs [
19]. In the second stage, participating physicians invited patients consecutively visiting the clinic who met the selection criteria (age 50 years or older, free of CVD but having at least one major cardiovascular risk factor; see ‘Patient Characteristics’) [
18]. Approximately 600 patients were included per country, with a final population size of 7641. For the present analysis, only patients for whom CRP measurements were available were considered (n = 7565). All patients provided signed informed consent forms. In each country, a random sample of 10% of all study centres underwent a site visit for data monitoring and audit to ensure data quality.
Assessment of cardiovascular disease risk factors
Information on participating patients was collected via their medical records, clinical anamnesis, physical examination and a 12-hour fasting blood sample collected within 1 day of the outpatient consultation [
17]. Blood samples were sent to a central laboratory in Belgium for analysis (the Bio Analytical Research Corporation NV, Ghent, Belgium). CRP levels were measured by a high-sensitivity immunoturbidimetry method (Roche P-Modular analyzer, Roche, Germany), high-density lipoprotein cholesterol (HDL-C) levels were measured by a modified enzymatic method (Roche P-Modular analyzer, Roche, Germany), total cholesterol levels were measured by the CHOD-PAP method (Roche P-Modular analyzer, Roche, Germany) and triglyceride levels were measured by the GPO-PAP method (Roche P-Modular analyzer, Roche, Germany). LDL-C levels were calculated by the Friedewald formula [
21] and glycated haemoglobin (HbA
1c) levels were measured by ion-exchange chromatography (Menarini 8160, Menarini Diagnostics, Netherlands).
Metabolic syndrome markers that were considered were: low HDL-C levels (<1.0 mmol/L in men or <1.3 mmol/L in women), high triglyceride levels (≥1.7 mmol/L), high HbA
1c levels (≥6%), large waist circumference (cut-off dependent on ethnicity [European Caucasian, Sub-Saharan, Middle East/North African and Afro-American: males ≥94 cm and females ≥80 cm; Asian, South American and Caribbean: males ≥90 cm and females ≥80 cm; Native American: males ≥102 cm and females ≥88 cm], in line with clinical guidelines [
22]) and high blood pressure (systolic blood pressure ≥130 mmHg and/or diastolic blood pressure ≥85 mmHg, or taking antihypertensive medication). Global cardiovascular risk was estimated according to the SCORE and FRS systems [
8,
10].
Statistical analysis
Statistical analyses were carried out using SAS (V9.2, SAS Institute Inc., Cary, NC, USA). Associations between demographic factors and log CRP values were assessed using Pearson correlation coefficients for continuous variables, analysis of variance for categorical values in three or more categories, and Student’s t-tests for binary data. A multivariate linear regression model was developed, adjusted for all factors associated with log CRP values in univariate analysis. Statistical significance was defined as two-sided P < 0.05.
Discussion
In this analysis of data from a large, multinational European study of the control of cardiovascular risk factors, we have shown that, among patients aged 50 years or older with at least one traditional cardiovascular risk factor who do not have diabetes mellitus and are not receiving statin treatment, more than one-third have CRP levels ≥3 mg/L, and approximately half have CRP levels ≥2 mg/L. This is the case irrespective of the global cardiovascular risk score, estimated using the two most widely used conventional risk estimation systems (SCORE and FRS). In particular, 34.0 to 38.2% of patients without diabetes mellitus who were not receiving statin therapy, and at an estimated intermediate 10-year cardiovascular risk according to SCORE or FRS, had CRP levels ≥3 mg/L, and 49.2 to 54.1% had CRP levels ≥2 mg/L. Elevated baseline CRP levels are associated with increased cardiovascular risk [
11,
12]; hence, it may be more appropriate to consider classifying such patients in the high-risk category, for whom pharmaceutical intervention for risk factor management may be appropriate [
7,
23]. The high prevalence of elevated CRP levels in patients considered to be at intermediate risk indicates that assessment of CRP levels in these individuals is likely to identify a considerable number of patients who are at higher risk than would be expected on the basis of their conventional risk factors alone. Patients with diabetes mellitus were excluded from the analysis because they were already considered to be at high risk [
7], and patients receiving statins were excluded because a positive treatment decision had already been made. Thus, almost a quarter (23.3%) of the EURIKA study patients without diabetes mellitus and not already taking a statin were found to be at intermediate risk by SCORE and had CRP levels ≥2 mg/L (17% had CRP levels ≥3 mg/L). These patients represent approximately 10% (and 8% respectively) of the entire EURIKA study cohort.
Data from the JUPITER study support the hypothesis that measurement of CRP levels could be used to identify individuals who are likely to benefit from statin therapy, although this approach is not currently endorsed by European guidelines. In this study, ‘high-risk’ cardiovascular event rates were observed among patients receiving placebo treatment with CRP levels ≥2 mg/L, but considered to be at intermediate cardiovascular risk according to SCORE and FRS [
15]. Statin treatment in this population resulted in a greater absolute risk reduction than would have been predicted on the basis of reductions in levels of LDL-C alone [
15,
24]. These data from the EURIKA study show that such individuals are very common in primary care in Europe. We have also shown that, in the overall population, CRP levels are positively associated with BMI, HbA
1c levels and the number of components of the metabolic syndrome present, and are negatively associated with HDL-C levels. We observed that CRP levels were higher in women than in men, and higher in those predicted to be at high cardiovascular risk than in those predicted to be at low cardiovascular risk according to SCORE.
Although there is considerable and consistent evidence for an association between CRP levels and cardiovascular risk, the use of CRP measurement in cardiovascular risk assessment remains controversial [
25,
26]. A meta-analysis published in 2004 of 22 population-based prospective studies, including a total of 7068 incident cases of coronary heart disease, found an adjusted odds ratio for the incidence of coronary heart disease of 1.6 (95% confidence interval 1.5 to 1.7) in patients with baseline CRP levels in the top third of the population analysed (approximately ≥2.4 mg/L), compared with patients with baseline CRP levels in the bottom third (approximately 1.0 mg/L) [
11]. However, as our data support, CRP levels are positively associated with several established cardiovascular risk factors, including high blood pressure, atherogenic dyslipidaemia, high BMI, diabetes mellitus, metabolic syndrome and smoking, and are low in individuals with protective factors, including high levels of physical activity, high HDL-C and apolipoprotein A1 levels, and consumption of fruits and vegetables [
27‐
32]. Incomplete adjustment for confounding factors in multivariate analysis may therefore lead to an overestimation of the strength of association between CRP levels and risk. In support of this, as the number of established cardiovascular risk factors adjusted for in multivariate analysis models increases, the correlation of CRP levels with cardiovascular risk decreases [
33]. Although our multivariate analysis model was adjusted for all factors found to be associated with CRP levels in univariate analysis, it is possible that the results could still be confounded by CVD risk factors that were not assessed and which are associated with CRP levels, such as dietary habits, levels of physical activity, and low-grade infection such as periodontal disease. The weak independence of CRP levels as a predictor of cardiovascular risk makes it unlikely that increased CRP is a major causal factor for CVD. Moreover, there is no association between cardiovascular event rates and genetic factors that raise CRP levels but have no effect on other cardiovascular risk factors [
33]. Such an association would be expected if elevated CRP levels are genuinely causally linked to CVD.
Additionally, a narrative review published in 2006 questioned the value of the universal use of CRP measurements in global cardiovascular risk estimation, on the basis that a very high adjusted odds ratio would be required to provide a moderate improvement in the accuracy of current risk-estimation systems [
25]. However, this review did acknowledge that elevated CRP levels may help to distinguish patients who are currently classified as being at intermediate risk using conventional estimation systems, but who may have event rates comparable to those in high estimated risk groups [
25,
34]. It should also be noted that CRP levels differ considerably between individuals and may fluctuate over time within individuals, so developing a universal system for assessing cardiovascular risk based on single measurements and cut-off CRP values may be challenging [
35‐
40]. However, their variability appears no greater than that of blood pressure and lipid levels. After the initial measurement of CRP levels, repeat measurements can be made at follow-up visits to the PCP.
JUPITER reported that patients with low levels of LDL-C but high levels of CRP benefitted from statin therapy, while the earlier AFCAPS/TexCAPS study found that patients with low levels of both LDL-C and CRP did not [
15,
16]. In contrast to both of these findings, an analysis of the Heart Protection Study showed that patients benefitted from statin therapy regardless of either their LDL-C or CRP levels [
41]. Furthermore, an analysis of data from the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) showed that CRP levels did not predict the efficacy of statin treatment in patients with LDL-C levels ≤250 mg/dL (6.5 mmol/L) [
42]. However, the majority of patients recruited into these studies would already be considered to be at high cardiovascular risk on the basis of established CVD, diabetes mellitus, or severe or complicated hypertension. These studies are therefore less relevant to the main question considered here, specifically regarding how we might improve the identification of individuals from lower-risk groups who might derive greater than otherwise expected reductions in their absolute CVD risk with more intensive risk factor treatment. It should be noted that the AFCAPS/TexCAPS study used lovastatin, a less potent agent than rosuvastatin, which was used in the JUPITER study [
15,
16]. It is therefore possible that the intensity of treatment in the AFCAPS/TexCAPS study was not sufficient for an outcome benefit to be detected in the lower-risk subgroup of patients with low levels of both LDL-C and CRP.
Our study has the strength of centralized assessment of a large sample of patients from multiple countries according to standardized procedures. The participation acceptance rate among physicians was low (3.1 to 22.8% across all countries), but the random selection of patients and a relatively high patient acceptance rate of 62.1% within the participating physicians’ practices is likely to have reduced patient selection bias. It should also be noted that, because the data-collection period for each country was only 3 months, it is possible that frequent healthcare service users were over-represented in the study cohort. This may bias the patient population towards the inclusion of less healthy patients, who could have higher CRP levels than healthier individuals.
Acknowledgments
Writing support was provided by Dr Stephen Sweet from Oxford PharmaGenesis™ Ltd, Oxford, UK, and was funded by AstraZeneca. The authors thank Professor Dr Eliseo Guallar, Department of Cardiovascular Epidemiology and Population Genetics, National Center for Cardiovascular Research, Madrid, Spain, who participated as a member of the EURIKA study academic steering committee.
Funding
The EURIKA study was funded by AstraZeneca. The study was run by an independent academic steering committee. The authors had full access to all data and had final responsibility for the contents of the manuscript and the decision to submit it for publication.
Competing interests
JPJH and JD have received speaker and consulting fees from AstraZeneca; FT has received research funding from AstraZeneca; PGS has received research grants from Sanofi, Servier and the New York University School of Medicine, speaking or consulting fees from Amarin, Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo-Lilly, GlaxoSmithKline, Iroko, Medtronic, MSD, Novartis, Otsuka, Pfizer, Roche, Sanofi, Servier and The Medicines Company, and is a stockholder and cofounder of Aterovax. JM and OS are employees of AstraZeneca. The rest of the authors declare that they have no competing interests.
Authors’ contributions
All authors contributed to the design and conduct of the study, and to the drafting and final approval of the manuscript.