NHLBI Workshop Report
C-Reactive Protein: Basic and Clinical Research Needs
July 10-11, 2006
Workshop Report (in-page links)
Introduction and Objectives
Background and Rationale
Conclusions and Recommendations
Introduction and Objectives
The National Heart, Lung, and Blood Institute convened a workshop on July 10-11, 2006 to examine the state of the science and to determine what additional basic, clinical, population, and cross-cutting research is needed on the role of C-Reactive Protein (CRP) as a biomarker, a risk factor, and/or treatment target for cardiovascular disease (CVD), in the broader context of the general role of lipids, inflammation, and other mechanisms for atherosclerosis and CVD.
The objective of the workshop was to bring together basic, clinical, and population scientists to examine the state of science regarding the relationship between CRP and CVD and to provide recommendations for research needed to delineate the importance of CRP in development and progression of CVD and its potential for use in clinical practice. The agenda aimed to balance various viewpoints, to allow time for discussion, to aim for a cross-cutting perspective, and to obtain direction for future research from the expert community.
Several questions were posed: What is the quality and consistency of epidemiologic data linking CRP to incident cardiovascular events? Through what basic biological mechanisms may CRP be related to CVD? Is CRP a causal factor for CVD, a marker of CVD risk, or a marker of forms of inflammation that are causal for CVD? What additional research is needed to determine the proper role of CRP measurement in CVD prevention and management?
Back to top
Background and Rationale for the Workshop
A sizable series of prospective epidemiologic studies indicates that baseline levels of CRP, especially when measured with high-sensitivity methods (hs-CRP), are associated with increased risk of developing future cardiovascular disease (CVD) events. The relationship is seen even when controlling for usual CVD risk factors. CRP levels are also predictive of incident type 2 diabetes and increase risk of both diabetes and vascular disease when used as an adjunct to definitions of metabolic syndrome. It remains uncertain, however, whether the major role of CRP is as a contributing cause of CVD events, a marker of CVD risk, or a marker of inflammation that is causal for CVD. In the first case, one would look to test treatments that could reduce risk of CVD by lowering CRP levels. In the second case, one would focus on determining whether CRP measurements can help identify individuals for whom more intensive treatment of standard risk factors, like blood lipids, should be delivered. In the third case, one would focus on approaches to reducing inflammation.
Current AHA/CDC guidelines, which are based on data available through 2002, state that elevated hs-CRP is an independent marker of risk and, in those judged at intermediate risk by global risk assessment (10% to 20% risk of CHD over 10 years), hs-CRP level may, at the discretion of the physician, help direct further evaluation and therapy for primary prevention of CVD. As heart-healthy diets, weight loss, and physical activity all reduce CRP levels as well as other CVD risk factors, the AHA/CDC guidelines suggest that a finding of elevated CRP can be used to reinforce basic messages for lifestyle change. The guidelines do not recommend that CRP be a direct target of therapy but rather be an indicator of risk that could assist in decisions about intensity of standard risk-reduction therapies. The guidelines further state, however, that benefits of therapies based on this strategy are uncertain. Although questions remain about the value of measuring CRP as a clinical tool, some physicians are using CRP as an adjunct to traditional risk factors to guide patient therapy.
CRP was discovered in 1930 and subsequently examined in both basic and clinical studies. CRP is an acute-phase protein with a rapid synthesis stimulated by acute and chronic infections or tissue injury. CRP is elevated to varying degrees in numerous conditions, including autoimmune disorders, trauma, acute bacterial infections, chronic infections, malignancies, and other conditions and is somewhat elevated in sleep deprivation. Under certain conditions, CRP can activate the complement pathway, may stimulate phagocytosis, and has been said by some to bind to immunoglobulin Fc receptors – actions that indicate a potential role in protecting the body against infectious agents. CRP, via its capacity to activate complement, can be pro-inflammatory and exacerbate tissue damage, as has been seen in models of ischemic injury. There is evidence that CRP is present in atheromatous plaques. Although there are a variety of potential biological mechanisms, pathways are unclear and findings differ across studies.
Epidemiologic studies have found associations between level of hs-CRP and CVD events. In many analyses of epidemiologic cohort data, particularly studies with large numbers of subjects, the CRP level predicts risk while controlling for the established risk factors. Other inflammatory markers – IL-6, serum amyloid A, fibrinogen, white cell count – are also associated with CVD risk; however, these markers are less stable than CRP and hence are less reliable indicators. Furthermore, in many analyses that have simultaneously controlled for other CVD risk factors – such as age, blood pressure, cholesterol, diabetes, and smoking – an effect of CRP on CVD risk remains. On the other hand, when CRP has been entered as a risk factor after traditional risk factors to Receiver Operator Curve (ROC) analyses, a few studies have reported that CRP levels do not add significantly to the area under the ROC curve (AUC) or C-statistic. This situation has led some investigators to conclude that CRP has little utility as a predictor in risk assessment. Other investigators have found incremental predictive power in ROC analysis as well as substantial reclassification of subjects, which would support its utility. Further, the issue has been raised whether ROC analysis can be taken as the “gold standard” for making decisions about clinical utility of risk markers. For example, the last risk factor to be entered into ROC analysis always has less predictive power than the first ones; hence the order of introduction of risk factors into ROC analysis can determine their predictive power.
Another way to evaluate the clinical utility of CRP may be in retrospective analysis of data from controlled clinical trials. Three randomized controlled trials of LDL-cholesterol reduction – PROVE-IT, REVERSAL, and A to Z – examined the association of CRP levels on outcomes in analyses combining the randomized groups and controlling for potential confounders. PROVE-IT and A to Z both found that CRP and LDL-C were strong predictors of CVD events after occurrence of acute coronary syndrome, and attainment of a lower CRP was associated with lower risk of recurrent CVD events. REVERSAL found that changes in both CRP and LDL-C were significantly correlated with changes in plaque burden, that study’s primary outcome. However, these are observational analyses of trial data relating to statin treatment, which reduces CRP as well as LDL-C; no clinical trials testing effects on CVD events of interventions targeting CRP reduction have been completed.
The JUPITER trial is currently testing effects of rosuvastatin on CVD events in persons without CHD but with elevated CRP and LDL-cholesterol level <130 mg/dL – subjects are selected to represent a higher risk cohort though already at their LDL-cholesterol goal. The trial will determine whether statins – which lower both CRP and LDL-C – will significantly reduce CVD events in the absence of elevated LDL-C at baseline.
Back to top
An introductory talk summarized the history of CRP discovery and its association with inflammation, noting that most people have low levels, that there is substantial ethnic variability as well as inter-individual variability over time, that values can be extremely high in the setting of acute inflammation, that ideal animal models are lacking, and that there are many unresolved questions about the role of CRP in normal physiology as well as pathophysiology.
A series of talks on findings from observational epidemiologic studies and clinical trials reviewed the substantial body of evidence that CRP level is associated with CVD incidence and mortality and with all-cause mortality. Prospective data have been reported in diverse populations including several ten-thousands of subjects from various ethnic groups. About two dozen observational epidemiologic studies have found associations between level of hs-CRP and CVD events, including stroke, myocardial infarction, other coronary heart disease events such as sudden death, and composite CVD outcomes including more than one of these. In many analyses of epidemiologic cohort data, particularly those with large numbers of subjects, CRP level predicts risk while controlling for multiple established risk factors, and in some studies the magnitude of relative risk associated with CRP elevation is comparable to well-accepted classical CVD risk factors, such as LDL-cholesterol.
The very strong relationship between adiposity and CRP was reviewed; it is supported by observational studies as well as randomized trials of weight loss effects. It was noted that many epidemiologic studies have inadequately controlled for lifestyle factors or obesity as potential confounders of the relationship between CRP and CVD, and that CRP levels appear to be influenced by physical activity, smoking, and diet. It was pointed out that in several epidemiologic studies the effect of CRP on CVD risk does remain significant even after controlling for BMI or other measures of adiposity. Other studies have shown that at all levels of both “fitness” and “fatness”, CRP is associated significantly with CVD risk.
Obstructive sleep apnea (OSA) is being increasingly associated with risk for CVD. It is often accompanied by cardiovascular risk factors, including elevated CRP. The relationship of sleep deprivation/apnea with CRP was reviewed, including the observation that CRP rises in persons deprived of sleep. The mechanisms whereby OSA is related to CRP have not yet been determined.
In models of relative risk for CVD, CRP level appears to be an independent risk factor for CVD events after statistical adjustment for other known CVD risk factors (age, blood pressure, blood lipoproteins, demographics, etc.). A critical question addressed in this workshop was whether adding CRP to clinical risk assessment can be useful for modifying the type and intensity of risk-reducing therapies. Different views were expressed. A widely used method to evaluate the utility of newer risk factors is receiver operating curve (ROC) analysis. According to one view, ROC is the current “gold standard” method and the Area Under the Curve (AUC) and resultant C-statistic should be used to determine the usefulness of adding factors to the equation. According to the other view, the commonly used C-statistic (area under the ROC curve), while useful for classification, has a limited role in evaluating new additions to risk prediction. ROC analysis erects a high barrier for newer risk factors, such as CRP, when added to traditional risk factors – a barrier that even traditional risk factors would not have met were they required to do so today. A view expressed was that the degree of patient reclassification to another risk category is more important than significant improvement in an ROC, as it’s the reclassification that will drive treatment approach for the individual patient. Some data show that models including CRP have the ability to more accurately classify individuals into risk strata. The role of age in various analysis approaches is also crucial, as age plays a dominant role in risk prediction. It remains an unresolved question whether risk scoring based on populations across a wide age span are reliable risk predictors for specific individuals. One of the major outcomes of the discussion was the need to develop theoretically sound statistical approaches to the evaluation of newer risk factors.
One strategy that received favorable attention was to employ newer risk factors to adjudicate risk within risk categories first identified with traditional risk factors. For example, CRP may prove to be a useful adjuster of risk in persons identified to be at intermediate risk. If this approach can be proven to be theoretically sound, it could lead to a more targeted way to employ CRP in clinical practice. Even so, the magnitude of reclassification of risk category that can be achieved by use of CRP in intermediate-risk individuals remains to be determined.
The workshop examined data from clinical trials of cholesterol-lowering therapy in which CRP has been measured. PROVE-IT, REVERSAL, and A to Z, examined effects of CRP on outcomes in analyses combining the randomized groups. PROVE-IT found that CRP and LDL-C were strong predictors of CVD events after occurrence of acute coronary syndrome, although multivariable-adjustment was not performed. In a post-hoc analysis of data from this trial, attainment of a lower CRP was associated with lower risk of recurrent CVD events, regardless of attained LDL-C level. In this study, the best outcomes were observed among those who achieved both LDL-C < 70 mg/dL and hs-CRP levels < 2 mg/L. Similar findings have now been reported from the A to Z trial. Finally, REVERSAL found that changes in both CRP and LDL-C were significantly correlated with changes in atheroma burden measured by intravascular ultrasound.
The JUPITER trial is currently testing the effects of rosuvastatin compared to placebo in the primary prevention of CVD events in persons without CHD but with elevated CRP and LDL-cholesterol level <130 mg/dL (the currently recommended goal for primary prevention). In this trial of 18,000 participants, it is estimated that both LDL-C and CRP levels will be reduced 40 to 50 percent. Since participants in JUPITER are not candidates for LDL-lowering drugs by current guidelines, the trial is designed to address whether statin therapy can prevent major vascular events among those at high risk due to elevated CRP levels. Whether selection of subjects on the basis of CRP for use of statin therapy in primary prevention will be cost-effective will remain to be determined.
Several talks on basic physiology and mechanisms of CRP were presented. In humans, plasma levels of CRP may rise rapidly and markedly, by as much as 1000-fold or more, after an acute inflammatory stimulus. CRP induction is part of a larger picture of the acute phase response, in which synthesis of many plasma proteins is increased and synthesis of a smaller number, notably albumin, is decreased. There are many acute phase plasma proteins, including clotting proteins, complement factors, anti-proteases, and transport proteins. Changes in circulating levels of these presumably contribute to defensive or adaptive capabilities. CRP has been shown to play a protective role in a variety of inflammatory conditions. However, like many mediators of inflammatory processes, both "pro-inflammatory" and "anti-inflammatory" activities of CRP have been described. It is likely that the function of CRP is context-dependent and that it can either enhance or dampen inflammatory responses depending on the circumstance. It was noted that the CVD association with CRP is frequently observed in studies also demonstrating similar associations with other inflammatory markers, suggesting non-specificity of the CRP role.
In other presentations, it was pointed out that variations in the CRP gene influencing plasma levels of CRP in adults are known to exist. The CRP gene is located on the short arm of chromosome 1, contains only one intron, which separates the region encoding the signal peptide and the first 2 amino acids of the mature protein from that encoding the remainder. Induction of CRP in hepatocytes is principally regulated at the transcriptional level by the cytokines interleukin-6 (IL-6) and interleukin-1 (IL-1). Both IL-6 and IL-1 control expression of many acute phase protein genes through activation of the transcription factors STAT3, C/EBP family members, and Rel proteins (NF-B). The unique regulation of each acute phase gene is due to cytokine-induced specific interactions of these and other transcription factors on their promoters. Thus, STAT3 is the major factor for the fibrinogen genes, NF-B is essential for the serum amyloid A genes, and the C/EBP family members C/EBP and C/EBP are critical for induction of CRP. The relation of variants in the CRP gene to future CVD is yet to be established, but given that CRP levels do predict future CVD and genetic alterations do affect CRP protein levels this is an intriguing possibility. In one recent analysis, polymorphism in the CRP gene set was associated both with lower CRP levels and lower event rates, data supporting a direct role of CRP in atherogenesis.
Regarding measurement, hs-CRP levels are currently measured in many clinical laboratories. The assay is reasonably simple using established reference materials with analytical coefficients of variation under 5%.
Workshop attendees discussed various issues regarding the relationship between CRP and CVD, with a focus on future areas of research needs, including the following:
- the need to better understand the mechanistic role of CRP in physiology and pathophysiology
- the role of CRP in inflammation and inflammatory consequences
- the need to clarify whether CRP is a causal risk factor or just a marker of CVD risk and the role of inflammation in this relationship
- the value of using CRP testing in clinical practice (particularly in persons considered to be at “intermediate risk” by the Framingham risk equation)
- cost-effectiveness of using CRP measurements for screening or clinical practice
- the potential need for clinical trials of interventions targeting CRP
- the need to determine whether reclassification of a patient’s risk in practice by adding CRP to traditional risk factors is valuable (both for patient outcomes and healthcare costs)
- the need to determine whether inflammation (however measured) is a cause of atherosclerosis and CVD events, and
- the possibility that genetic approaches may shed light on the role of CRP.
It was noted that the need to determine the utility of CRP in clinical practice should be the driving force behind determination of the most important research questions. Towards this end improvement of existing animal models and fostering the development of new and better ones should be supported.
Back to top
Conclusions and Recommendations
The working group concluded that there is a lot of evidence that CRP level is a predictor of CVD events. Although it has not been shown to be a causal risk factor due to lack of clinical trial evidence that reducing CRP will reduce CVD events, the predictive power of CRP is in the same range found with several established risk factors. However, there was not agreement on how to integrate CRP measurement into clinical practice. Whether CRP plays a pathophysiologic role in atherosclerosis remains controversial. No drugs are available that specifically lower CRP to test efficacy of reduction. Candidate drugs nevertheless have been identified and are currently being assessed. The most pressing research need is for studies that will establish more firmly the role of CRP measurement in CVD etiology and risk categorization in order to inform the screening of patients for risk of future CVD and in guiding preventive therapy. More research and analysis are needed to resolve the question of clinical utility of CRP.
Research recommendations were made in three major areas: (a) epidemiologic and clinical research (the major focus of the recommendations), (b) methods development, and (c) basic research. It was also recommended that research examining CRP be broadened to include circulating inflammatory mediators in general.
A. Epidemiologic and Clinical Research
1. Analyses of individual and pooled data from existing and new epidemiologic studies (or ancillary studies) to examine whether CRP measurement improves CVD risk prediction beyond traditional risk factors and can help target who should be treated. The following issues and questions were noted:
2. Studies to further delineate whether CRP is a causal risk factor for CVD and the utility of measuring CRP in clinical practice.
- Whether the association between CRP level and CVD is truly independent of other CVD risk factors.
- Whether adding CRP to basic risk assessment moves a sufficient proportion of patients from one treatment category to another (reclassification), making it worthwhile in practice. Evaluation of this approach likely will require methods beyond ROC analysis (e.g., new methods of decision analysis that deal with calibration and reclassification in prospective cohorts) and consideration of health economics, especially in light of changing costs of therapy.
- Whether replacing a traditional risk factor(s) with CRP (or other newly discovered risk predictors) could improve CVD risk prediction approaches that are feasible for use in clinical practice.
- The shape of the dose-response curve.
- Possible heterogeneity by population subgroups (e.g., age, ethnicity, etc.), including prospective studies in minority groups to assess the association of CVD incidence with CRP and other inflammatory markers and to establish whether CRP cutpoints should vary by ethnicity or sex.
- Whether CRP is better than other inflammatory markers, novel biochemical measures or subclinical disease markers, or traditional CVD risk factors, singly or in combination, in cost-effectively predicting CVD.
3. Primary prevention studies
- Observational studies associating CVD with CRP genetic polymorphisms that influence CRP level. This approach, based on the assumption of “Mendelian randomization,” can help address whether CRP is a causal risk factor for CVD.
- Development and testing of CRP-inhibiting drugs for use in clinical trials that could determine the efficacy of such drugs in preventing CVD events.
- Clinical trials to test efficacy of CRP-lowering in reducing mortality and infarct size in the setting of acute coronary syndromes.
- Clinical trials to determine whether measuring CRP enables better targeting of therapy which can significantly lower CVD events. These trials could help establish whether a high or low CRP should influence the type of preventive therapy, be used to monitor response to therapy, or to provoke modifications of therapy.
4. Relationship of CRP to other conditions
- Clinical trials of various primary prevention approaches in which elevated CRP is used to identify subjects at high risk, for example, trials of hypertension or diabetes prevention in persons with high CRP. Analyses potentially could be done post-hoc using stored samples from completed trials.
- Molecular and cellular epidemiologic studies to identify and characterize the components of innate immunity pathways and to identify variations in pathway biology in the general population.
- Studies to develop effective individual or population-wide strategies to prevent development of elevated CVD risk in the first place, including elevated CRP and its key behavioral determinants, obesity and exercise.
- Clinical trials to determine whether measuring and using CRP levels can motivate patients or physicians to follow CVD prevention guidelines.
- Observational or experimental studies addressing the degree to which dietary composition or physical activity influence CRP levels independent of obesity.
5. Other clinical studies
- Studies of the interrelations of sleep deprivation or obstructive sleep apnea with obesity, CRP, and CVD.
- Prospective studies of inflammatory conditions (e.g., psoriasis, rheumatoid arthritis, etc.) and risk of subsequent CVD and whether CRP elevation may explain the associations.
- Studies to determine genetic polymorphisms in the CRP gene that affect levels and acute variation of CRP. Verification in young people would be particularly useful.
- Studies examining the relationship between CRP levels and degree of atherosclerosis.
- Studies to assess the value of measuring CRP in acute coronary syndromes.
- Analyses of clinical trial data to examine whether anti-inflammatory effects of treating blood pressure or lipids contribute to a difference in treatment efficacy.
B. Methods Development
- Development of improved biostatistical methods for clinical decision-making when the aim is risk prediction, which can be used for treatment targeting as well as for screening. Specific application to CRP and multiple inflammatory markers is needed.
- Studies to develop validated biomarkers that reflect pathology-associated changes in inflammation. Linking biomarker and imaging measures (e.g., PET, MRI, ultrasound, etc.) is potentially important. This might be facilitated by an inflammation-centered clinical trials network that could evaluate surrogate markers and new candidate targets and therapeutics.
- Development of data to establish acceptable laboratory error for CRP and measurement standards, and possible development of a CRP measurement standardization program.
C. Basic Research
- Development of appropriate animal models of CRP and atherosclerosis. For example, identify appropriate genes and develop and study knockout animals.
- Use animal models to examine mechanisms by which CRP may affect atherosclerosis and ischemic CVD; determine the role of CRP in atherogenesis. Additional basic research on the physiological functions of human CRP and its possible role in atherosclerosis and thrombosis. For example, study the impact of CRP inhibition.
- Use animal models to test potential therapeutic approaches prior to conducting CRP-targeted clinical trials in humans.
- Basic and clinical studies to identify specific markers of atherosclerosis and develop assays. These are needed to help characterize the role of the inflammatory process and its components in the development of atherosclerosis and subsequent CVD events.
- Investigations of the sources of CRP elevation and sources of stimuli for CRP production.
Back to top