Clinical Utility of Inflammatory Markers

Kiran Musunuru

Samia Mora

High-sensitivity C-reactive protein (hsCRP) testing is inexpensive and has been shown to (a) predict cardiovascular risk in multiple patient groups; (b) add prognostic information to global coronary risk scores; (c) predict incident diabetes and cardiovascular disease (CVD) in patients with the metabolic syndrome; and (d) predict clinical outcomes in high-risk patients treated with statin therapy. The JUPITER (Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin) trial demonstrated that hsCRP can be used to target high-risk patients with normal low-density lipoprotein cholesterol (LDL-C) levels but without known vascular disease or diabetes who would benefit from statin use. Although evidence is not yet present that lowering hsCRP per se reduces cardiovascular risk, hsCRP has value as a screening tool to improve cardiovascular risk stratification and to more accurately target preventive therapies to patients. In addition, among high-risk patients treated with statin therapy, achieving low levels of hsCRP may be a clinically relevant therapeutic goal along with achieving low levels of LDL-C. There is a growing body of evidence to support recommendations for measurement of hsCRP in (a) selected asymptomatic individuals deemed to be at intermediate risk of CVD according to traditional risk factor assessment and who do not already warrant treatment with chronic aspirin or statin therapy and (b) selected secondary CVD prevention patients for further risk stratification in combination with LDL-C.

Current prevention guidelines recommend targeting the intensity of preventive cardiovascular interventions to the level of the patient’s risk. Coronary risk assessment is usually obtained from a number of traditional risk factors and a global risk score such as that used in the third report of the Adult Treatment Panel of the National Cholesterol Education Program (NCEP ATP III) for “hard” coronary heart disease (CHD) events (myocardial infarction and coronaryrelated death) (1) or the Framingham risk score equation for total CHD events (2). These global risk scores give an absolute risk or the percentage chance that a hard coronary event (in the case of ATP III) or any coronary event (in the case of the Framingham risk score) will occur over the next 10 years in the patient being evaluated based on his or her traditional risk factors. The intensity of treatment of dyslipidemia or high blood pressure and the use of other preventive therapies such as aspirin are then tailored to these risk categories, with low risk being generally defined as a 10-year absolute risk of a hard coronary event <10%, intermediate risk 10% to 20%, and high risk >20% (1). An alternative intermediate-risk classification of 5% to 20% has also been proposed by some experts.

Although one in two women and two in three men are affected by CVD after the age of 40 years, only a small proportion of asymptomatic U.S. adults (<1% of women and approximately 5% of men) are classified as “high risk” for CVD using contemporary risk assessment scores. This discrepancy has been coined the “detection gap” in coronary risk assessment, with the recognition that there is a need for better risk assessment in asymptomatic individuals, particularly those in the “intermediate risk” category. In the United States, 10% of asymptomatic women (∼7 million) and 40% of asymptomatic men (∼26 million) are considered to be at “intermediate risk,” as defined by an absolute hard cardiac event rate of 5% to 20% over the next decade (3). Since current guidelines recommend targeting the intensity of preventive therapies to the level of a patient’s risk, a substantial proportion of men and women could benefit from more accurate coronary risk assessment. The challenge has been to identify such individuals more accurately, especially those considered to be at intermediate risk by standard assessment methods.

DESIRABLE CHARACTERISTICS OF A BIOMARKER

Numerous biomarkers have been proposed as potentially useful for improving CVD risk prediction. A biomarker is felt to be useful if it (a) adds to clinical knowledge, (b) provides risk information that is independent of established predictors, (c) is easy to obtain and interpret in a primary care setting, (d) is accurate, reproducible, and internationally standardized, and (e) has a favorable cost-benefit ratio (4). As it relates to global risk scores, screening biomarkers should also contribute to patient management, particularly through more accurate risk classification and guidance in choice of therapy (4).

INFLAMMATION AND ATHEROSCLEROSIS

High-sensitivity C-reactive protein (hsCRP) is an easily measured and widely investigated biomarker of inflammation. Recently, it has been recognized that inflammation is a key element of the atherosclerotic process, contributing to all of its stages (plaque initiation, progression, and rupture) (5). Macrophages, T cells, and mast cells are important players in the evolution of the initial fatty streak to an atheromatous lesion, and these inflammatory cells produce cytokines and proteolytic enzymes that may transform a stable atherosclerotic plaque into a “vulnerable” one whose fibrous cap is prone to rupture. While the link between inflammation and atherosclerosis is well established, hsCRP levels correlate only moderately with the extent of atherosclerosis. Thus, hsCRP levels may improve prediction of CHD and stroke risk in patient populations as a reflection of the propensity of plaque rupture more than of the extent of underlying atherosclerosis.

It is unclear whether CRP itself directly contributes to the pathophysiology of CVD. Supporting data for a role for CRP in atherothrombosis remain limited to in vitro studies and experiments in animal models (6,7). The presence of CRP in atherosclerotic lesions is suggestive but does not establish a casual role for the protein. Recent data from Pepys et al. (8) show that specific inhibition of CRP is feasible with a small-molecule synthetic compound [1,6-bis(phosphocholine)-hexane], resulting in smaller infarct size in rats.

EVIDENCE FOR THE USE OF hsCRP IN CARDIOVASCULAR RISK ASSESSMENT

Primary Prevention

Cardiovascular Disease

In 2003, the Centers for Disease Control and Prevention (CDC) and the American Heart Association (AHA) issued recommendations regarding the use of inflammatory biomarkers for CVD detection, prevention, and treatment (9). At the time, the body of evidence regarding the clinical use of hsCRP measurements was modest, and most recommendations were made at an American College of Cardiology and American Heart Association (ACC/AHA) Class II level of support, indicating that the weight of evidence was favorable but that more data were needed before general consensus could be reached.

Since 2003, a significant body of data has been published regarding the use of hsCRP levels in improving the assessment of cardiovascular risk in primary prevention patients. More than 20 prospective studies from distinct cohorts have demonstrated that elevated hsCRP levels are associated with elevated risk of future coronary events after multivariate adjustment for at least four traditional risk factors, including Framingham risk factors and/or diabetes and obesity (Table 15.1) (10). This finding applied both to men and women across a wide age range (e.g., middle-aged and elderly). Some of these studies stratified patients by hsCRP levels of <1, 1 to 3, and >3 mg/L and showed that these cutoffs partition individuals into lower-, moderate-, and higher-risk groups, although the risk was fairly linear across a wide range of hsCRP levels. A smaller number of studies reported a positive association between hsCRP and coronary events but did not reach statistical significance after adjustment for at least four other risk factors (Table 15.1) (10). An initial analysis of data from the Framingham Heart Study did not support a clear incremental value of this biomarker over the Framingham risk score. However, the investigators did not use a “high-sensitivity” assay to measure CRP in that study, and when the analysis was repeated using an appropriate high-sensitivity test, a positive result consistent with other studies was found, with hsCRP levels >3 mg/L significantly associated with increased incident CVD, after multivariate adjustment (Table 15.1).

A meta-analysis of 22 of the above-mentioned prospective studies found that individuals in the top tertile of hsCRP levels have an odds ratio of 1.45 for major cardiac events (95% confidence interval [CI] 1.25 to 1.68) compared with those in the lowest tertile, after adjustment for traditional risk factors (11). In the studies that analyzed traditional risk factors in a similar way to hsCRP (i.e., multivariate analysis to assess the strength of association with CVD risk), the magnitude of the association with incident CVD of hsCRP levels was comparable to that of the LDL-C, systolic blood pressure, or smoking behavior.

With respect to this well-substantiated and moderately strong association of elevated hsCRP levels with higher CVD risk, which is similar in magnitude to that of traditional risk factors, routine clinical use of the biomarker requires demonstration that the addition of hsCRP measurement into CVD risk assessment strategies has clinically meaningful results. Recent data from the Women’s Health Study suggest that the addition of hsCRP levels and family history of premature CHD to the ATP III global risk score provide a more accurate assessment of CVD risk. In this large prospective cohort of asymptomatic middle-aged women, the addition of hsCRP levels and family history to the ATP III global risk score—creating a new score termed the Reynolds Risk Score—reclassified approximately 30% of intermediate risk women into a higher- or lower-risk category (12). The addition of hsCRP and family history also improved risk classification in asymptomatic men (13).

In a cohort of middle-aged asymptomatic men, additional information regarding risk was provided by hsCRP levels beyond that obtained from the Framingham risk score, particularly in those at intermediate risk (14). Other studies show that hsCRP levels provide incremental risk information to the Framingham risk score in elderly men at intermediate risk and elderly women at high risk (10). Thus, there are several prospective cohorts for which the addition of hsCRP measurement to CVD risk prediction reclassifies a significant proportion (approximately 20% to 30%) of intermediate-risk individuals to a different risk category and may have important implications for preventive pharmacotherapy for these individuals.

However, it remains to be demonstrated that this degree of reclassification will result in improved patient outcomes. In the absence of long-term prospective studies, statistical criteria are being used to evaluate the incremental utility of hsCRP measurement. In a recent publication from the Framingham Offspring Study, baseline levels of hsCRP were associated with higher overall mortality during a 7-year follow-up period (15). Despite the higher mortality, the c-statistic (derived from the receiver-operator curve, or ROC, where a value of 0.5 signifies a test of no utility and a value of 1.0 signifies a test
with perfect discrimination) of the risk prediction model did not change with the addition of hsCRP. Indeed, most studies have not found the inclusion of hsCRP in models to increase the c-statistic significantly, primarily because any new risk factor would need to increase risk dramatically—at least severalfold—to affect the c-statistic of a model that already includes several risk factors.

TABLE 15.1 ASSOCIATION OF C-REACTIVE PROTEIN WITH CORONARY HEART DISEASE IN PRIMARY PREVENTION POPULATIONS

Study	Population	Adjusted for	End point	Comparison	RR or OR (95% confidence interval)
Studies that show a significant association after multivariate adjustment (p < 0.05)
AFCAPS/TexCAPS	Men, women	Age, sex, smoking, HTN, parental history of CAD, lipid levels	MI, CHD death, UA	Quartiles (per one-quartile increase)	1.17 (1.03-1.33)
ARIC	Men, women	Age, sex, race, smoking, HTN, DM, LDL, HDL	Incident CHD	Tertiles (3 vs. 1)	1.72 (1.24-2.39)
BRHS	Men	Age, town, smoking, BP, TC, HDL, TG, BMI, occupation, housing tenure, marital status, car ownership, childhood socioeconomic factors	MI, CHD death	Tertiles (3 vs. 1)	2.13 (1.38-3.28)
Caerphilly + Speedwell	Men	Age, area, smoking, BMI, DBP, TC, evidence of ischemia at baseline	MI, CHD death	Quintiles (5 vs. 1)	1.72 (1.14-2.58)
CHS	Men, women aged >65	Age, sex, race, field center, HTN, DM, smoking, BMI, waist circumference, TC, HDL, aspirin use	MI, CHD death	CRP >3.0 mg/L vs. <1.0 mg/L	1.45 (1.14-1.86)
Edinburgh	Men, women	Age, sex, subclinical disease (ABI), pack-years smoking, DM, BMI, TC/HDL ratio, physical activity	MI, stroke, revascularization	Tertiles (3 vs. 1)	1.62 (1.11-2.38)
EPIC-Norfolk	Men, women	Age, sex, smoking, DM, BMI, SBP, LDL, HDL	CAD	Quartiles (4 vs. 1)	1.66 (1.31-2.12)
FHS (high-sensitivity CRP assay)	Men, women	Age, sex, smoking, TC/HDL ratio, DM, SBP, use of antihypertensives	CVD	CRP >3.0 mg/L vs. <1.0 mg/L	1.74 (1.15-2.63)
Honolulu	Men	Age, smoking, alcohol, TC, HTN, DM, BMI, physical activity index	MI	Quartiles (4 vs. 1)	1.6 (1.1-2.2)
HPFUS	Men	Age, smoking, month of blood sampling, parental history of CHD, alcohol, physical activity, TC/HDL ratio, BMI, DM, HTN	MI, CHD death	Quintiles (5 vs. 1)	2.55 (1.40-4.65)
Kuopio	Men	Age, year of exam, smoking, LDL, HDL, SBP, use of antihypertensives, diet, fasting insulin, fasting glucose, waist girth, exercise, alcohol, socioeconomic status	CVD death	Tertiles (3 vs. 1)	1.71 (1.16-2.54)
MONICA-Augsburg	Men	Age, survey, BMI, smoking, alcohol, physical activity, SBP, TC/HDL ratio, parental history of MI, history of DM	MI, CHD death	Tertiles (3 vs. 1)	1.89 (1.28-2.77)
PHS	Men	Age, BMI, DM, history of HTN, family history of CAD	MI	Quartiles (4 vs. 1)	2.6 (1.6-4.4)
PRIME	Men	Age, smoking, DM, HTN, LDL, HDL, TG	MI, CHD death	Tertiles (3 vs. 1)	2.16 (1.26-3.72)
PROSPER	Men, women aged >70	Age, sex, randomized treatment, country, current and past smoking, SBP, DBP, use of antihypertensives LDL, HDL, TG, DM, BMI	MI, CHD death, stroke	Tertiles (3 vs. 1)	1.51 (1.17-1.95)
Reykjavik	Men, women	Age, sex, year of enrollment, smoking, SBP, TC, TG, BMI, FEV₁, DM, socioeconomic status	MI, CHD death	Tertiles (3 vs. 1)	1.45 (1.25-1.68)
SOF	Women aged >65	Age, HTN, LDL, HDL, DM, smoking, BMI, estrogen use, education level, clinical site	CVD death	Quartiles (4 vs. 1)	8.0 (2.2-29)
WHI	Women	Age, ethnicity, smoking, length of follow-up, TC/HDL ratio, BMI, history of HTN, family history of premature CAD, DM, exercise frequency, alcohol, use of HRT	MI, CHD death	Quartiles (4 vs. 1)	2.1 (1.1-4.1)
WHS	Women	Age, smoking, DM, BP, use of HRT	MI, ischemic stroke, coronary revascularization, CVD death	Quintiles (5 vs. 1)	2.3 (1.6-3.4)
WOSCOP	Men	Age, statin treatment, DM, HTN, angina, BMI, SBP, TC, LDL, HDL, TG	MI, CHD death, revascularization	Quintiles (5 vs. 1)	1.49 (1.00-2.22)
Studies that do not show a significant association after multivariate adjustment (p > 0.05)
FHS (non-high-sensitivity assay)	Men, women	Age, sex, smoking, TC/HDL ratio, DM, SBP, use of antihypertensives	MI, CHD death	CRP >3.0 mg/L vs. <3.0 mg/L	1.22 (0.81-1.84)
MONICA-Augsburg	Women	Age, survey, BMI, smoking, alcohol, physical activity, SBP, TC/HDL ratio, parental history of MI, history of DM	MI, CHD death	Tertiles (3 vs. 1)	1.35 (0.64-2.84)
MRFIT	Men	Age, cigarettes smoked, DBP, HDL, LDL, TG	MI, CHD death	Quartiles (4 vs. 1)	1.54 (0.96-2.50)
Health ABC	Men, women aged >70	Age, sex, race, smoking, DM, HTN, BMI, HDL, TG, albumin	Incident CHD	Tertiles (3 vs. 1)	1.20 (0.83-1.75)
Hoorn	Men, women	Age, sex, impaired glucose tolerance, DM, HTN, smoking, TC, HDL, TG, IHD, PAD, obesity	CVD death	Tertiles (3 vs. 1)	1.32 (0.52-3.35)
Iowa 65+	Men, women aged >65	Age, sex, prevalent CVD, smoking, DM, BMI	CVD death	Quartiles (4 vs. 1)	1.8 (0.9-3.6)
NHS	Women	Age, smoking, month of blood sampling, fasting status, parental history of CHD, alcohol, physical activity, TC/HDL ratio, BMI, DM, HTN, use of HRT	MI, CHD death	Quintiles (5 vs. 1)	1.61 (0.84-3.07)
Quebec	Men	Age, smoking, history of DM, SBP, medication use at baseline, BMI, LDL, HDL, TC/HDL ratio	MI, CHD death, angina, coronary insufficiency	Halves (2 vs. 1)	1.1 (0.7-1.6)
Rotterdam	Men, women aged >55	Age, sex, current smoking, BMI, HTN, DM, family history of early MI, TC, HDL	MI	Quartiles (4 vs. 1)	1.2 (0.6-2.2)
SMILE	Men	Age, smoking, alcohol, DM, obesity, SBP, DBP, TC, HDL, TG	MI	Quintiles (5 vs. 1)	1.4 (0.9-2.1)
Studies: AFCAPS/TexCAPS, Air Force/Texas Coronary Atherosclerosis Prevention Study; ARIC, Atherosclerosis Risk in Communities; BRHS, British Regional Heart Study; Caerphilly, Caerphilly Heart Study; CHS, Cardiovascular Health Study; Edinburgh, Edinburgh Artery Study; EPIC-Norfolk, European Prospective Investigation into Cancer Norfolk Study; FHS, Framingham Heart Study; Health ABC, Dynamics of Health, Aging and Body Composition Study; Honolulu, Honolulu Heart Study; Hoorn, Hoorn Study; HPFUS, Health Professionals Follow Up Study; Iowa 65+, Iowa 65+ Rural Health Study; Kuopio, Kuopio Ischaemic Heart Disease Risk Factor Study; MONICA-Augsburg, Monitoring Cardiovascular Disease Augsburg Cohort Study; MRFIT, Multiple Risk Factor Intervention Trial; NHS, Nurses’ Health Study; PHS, Physicians’ Health Study; PRIME, PRIME (étude prospective du l’infarctus myocarde) Study; PROSPER, Prospective Study of Pravastatin in the Elderly at Risk; Quebec, Quebec Cardiovascular Study; Reykjavik, Reykjavik Study; Rotterdam, Rotterdam Study; SMILE, Study of Myocardial Infarctions Leiden; SOF, Study of Osteoporotic Fractures; Speedwell, Speedwell Prospective Study; WHI, Women’s Health Initiative; WHS, Women’s Health Study; WOSCOP, West of Scotland Coronary Protection Study.
AF, atrial fibrillation; BMI, body mass index; BP, blood pressure; CAD, coronary artery disease; CHD, coronary heart disease; CRP, C-reactive protein; CVD, cardiovascular disease; DBP, diastolic blood pressure; DM, diabetes mellitus; FEV₁, forced expiratory volume in 1 second; HDL, high-density lipoprotein; HRT, hormone replacement therapy; HTN, hypertension; IHD, ischemic heart disease; LDL, low-density lipoprotein; MI, myocardial infarction; OR, odds ratio; RR, risk ratio; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides; UA, unstable angina.
Adapted from Musunuru K, Kral BG, Blumenthal RS, et al. The use of high sensitivity assays for C-reactive protein in clinical practice. Nat Clin Pract Cardiovasc Med. 2008. In press.

Statistical Considerations

The clinical utility of a biomarker such as hsCRP is determined by the degree to which it would enhance the accuracy of existing cardiovascular risk assessment tools such as the ATP III global risk score. From a practical standpoint, the clinician who is seeing a patient in his or her office wants to know “What is the chance that this patient will develop CVD in the future?” Ideally, clinicians would provide an assessment of the patient’s chance of developing CVD that would match the observed outcome for that patient during follow-up. In statistical terms, this is referred to as accuracy, and it is assessed in two ways, calibration and discrimination.

Model calibration refers to how well the predicted probability from the model agrees with the actual risk observed for that patient if he or she were to be followed in time. Model discrimination (usually assessed with the c-statistic or the
receiver operating curve [ROC]) refers to the relative ranking of individuals such that those who develop CVD during the follow-up period have a higher predicted risk than those who do not develop CVD. Whether calibration or discrimination is more important depends on the clinical setting. In particular, when the aim is to accurately predict the percentage chance that a patient will develop a CVD event during follow-up, then calibration should be emphasized over discrimination, since the goal in this case is to predict the chance of a future event in that patient as accurately as possible compared with the actual event. If the goal is to compare that patient’s chance of an event relative to other patients, then discrimination should be emphasized over calibration. For risk prediction, the most important issue may be the extent of reclassification of risk level compared to ATP III and whether that reclassification is accurate.

Only gold members can continue reading. Log In or Register to continue