Dr Myerson is Director of Preventive Cardiology, St. Luke’s-Roosevelt Hospital, Columbia University College of Physicians and Surgeons, New York, NY.
There has been tremendous progress in the treatment of acute cardiovascular (CV) disease (CVD), and there have also been developments in predicting who will have a CV event. But although drug-coated stents have taken intervention to a new level, it is not clear whether novel risk markers have improved our ability to identify those at greatest risk.
Much of our early knowledge about risk factors came from the Framingham Heart Study. Begun in the late 1940s, this epidemiologic trial recruited men and women without CVD and followed them over many years to see who developed CVD. In this way, factors that were associated with the development of disease, namely, “risk factors,” could be identified. These factors, which were entered into an algorithm known as the Framingham Risk Score, included age, gender, cigarette smoking, systolic blood pressure, and total cholesterol and high-density lipoprotein cholesterol levels. This score has held up well throughout the years, even as other markers are identified and new information is incorporated into guidelines. The Third Report of the National Cholesterol Education Program recommends including other components when assessing risk, such as low-density lipoprotein cholesterol (LDL-C) level, metabolic syndrome, diabetes, and triglycerides. Of note, treatment algorithms use LDL concentration as the basis for management decisions.
Over the past several years, studies have identified new factors that may confer risk for CVD. However, physicians are often uncertain about which are useful, and when and how to use those that may be of help.
Can the New Markers Add Anything?
If a patient is clearly at high risk of CVD by virtue of having multiple risk factors or diabetes, most physicians will be aggressive with their management. The presence or absence of novel risk factors would not do much to change the management of such a patient. Nor would patients stratified as being at low risk according to traditional risk factors have their management plan changed. It is those in the intermediate-risk category who may be helped by checking markers such as high-sensitivity C-reactive protein (CRP) or lipoprotein (Lp) (a). This added information may help physicians determine what further testing is needed and how intensive medical therapy should be.
One novel marker that has received much attention is CRP. Much of the original work was performed by Paul M. Ridker, MD, of Brigham and Women’s Hospital, Boston. Inflammation is believed to be important in all phases of atherothrombosis, from plaque formation to acute rupture. There are many inflammatory markers, but the best characterized has been CRP. Numerous studies have shown CRP to be an independent predictor of risk for myocardial infarction (MI), stroke, peripheral vascular disease, and sudden cardiac death. Recent statements by the American Heart Association have suggested that CRP may be most useful for predicting CV events in those at intermediate risk. CRP levels <1 mg/dL, 1 to 3 mg/dL, and >3 mg/dL are considered low, moderate, and high, respectively, in terms of CV risk prediction.
It is important to remember, however, that no study to date has shown that lowering CRP levels will lower disease risk; this is in contrast to the multitude of studies demonstrating that lowering LDL-C decreases the incidence of heart attacks and coronary heart disease (CHD) deaths.
Arch Intern Med
Not all studies have supported the independent role of CRP in predicting CVD. A prospective observational study (. 2005;165: 2473-2489) that involved >4000 men and women from the Framingham Heart Study without CVD at baseline who were followed for 8 years found that elevated CRP level provided no prognostic information beyond traditional risk factor assessment. In a more recent report (. 2006;113:38-43), CRP level was measured in 3373 patients, aged 30 to 65 years, who participated in the Dallas Heart Study. Electron beam computed tomography scans were used to measure coronary artery calcification, an indicator of subclinical atherosclerosis. Among patients with higher CRP levels, the prevalence of subclinical atherosclerosis was modestly increased, but the association was not independent of traditional CV risk factors.
Hemostatic and thrombotic factors have also been investigated for their ability to forecast CVD. Fibrinogen is the major coagulation protein in blood by mass, the precursor of fibrin, and an important determinant of blood viscosity and platelet aggregation. A meta-analysis of prospective studies of patients without known CVD at baseline (. 2005;294:1799-1809) showed a moderately strong association between usual plasma fibrinogen level and risk of CHD and stroke and of other vascular mortality. An earlier trial, the Bezafibrate Infarction Prevention study, was designed to look at how a fibrate drug could lower lipids; it also looked at fibrinogen levels but not specifically for the purpose of lowering them per se. Its findings did not support lowering fibrinogen levels for CVD prevention, but it did show “an overall trend in reduction of the incidence of primary end point,” which included fatal/nonfatal MI or sudden death.
Lp(a) consists of an LDL particle with its apolipoprotein (apo) B-100 component linked to an apo(a). Lp(a) has a sequence homology to plasminogen, thereby linking its function to an inhibitor of fibrinolysis. The exact biologic function of Lp(a) is still unclear, in part because there are many heritable forms of Lp(a). Investigators have looked at the association between Lp(a) and CVD for many decades, with early studies using different measurement techniques. More recent research has benefited from improvement and standardization of measurement techniques.
Most, but not all, prospective, retrospective, and cross-sectional studies have linked Lp(a) with risk for CVD. It is felt that Lp(a) has a strong hereditary component and is minimally, if at all, affected by lifestyle and environmental factors. The only medication other than estrogen that lowers Lp(a) is niacin (which is now covered under Medicare; see article in this issue).
Homocysteine, Tried but Failed
Homocysteine is an amino acid derived from dietary methionine. Many early studies showed a strong positive association between plasma levels and CVD risk. The pathophysiologic rationale was based on observations in individuals with a rare genetic disorder of methionine metabolism, in whom severe elevations of homocysteine led to premature atherosclerosis. A 2002 meta-analysis showed that a 25% lower level of homocysteine was associated with an approximate 11% lower risk of CHD. Because folate and vitamin B12 supplementation (the accepted treatment for lowering homocysteine levels) are safe, inexpensive, and readily available, many clinicians went ahead and treated accordingly.
However, as was reported in this journal last month (, April 2006), 2 new, large prospective studies did not find that folate and vitamins B6 and B12 reduced risk of major CVD events in patients with vascular disease or diabetes. In patients with acute MI who were randomized to different combinations of B vitamins and folate, homocysteine levels declined by 27% in those taking folate and B12, but treatment had no significant effect on recurrent MI, stroke, or sudden death from CHD. In the folate and vitamins B12 and B6 group, there was a trend toward an increased risk. As a result of these latest findings, screening for homocysteine levels is not recommended.
When Is a Marker Clinically Useful?
Clinicians should look for several key features when deciding whether to incorporate a marker when screening for risk. Prospective studies must show that it is associated with specific events. It should have adequate sensitivity and specificity. The measure should add independent information, above and beyond what can be determined by known risk factors, markers, and tests. The measure should be free from variation that cannot be accounted for by things such as diet, weight gain, or active disease. Finally, there should be evidence that treating a person based on that marker or that lowering levels of that marker will either reduce risk or prevent the occurrence of disease.
Much like advancements in technology, research continues to improve our ability to predict who will develop disease. However, unlike tests for new technology, the science of prediction must wait many years to determine whether a factor is truly a risk factor and indeed prognostic.