Minneapolis MN and Boston MA - Whether or not measuring CRP would improve coronary risk prediction is the subject of three papers and two editorials in the July 10, 2006 issue of the Archives of Internal Medicine and the July 4, 2006 issue of the Annals of Internal Medicine.

Although there is some difference of opinion, most of the authors/editorialists appear to suggest that adding CRP to risk-prediction models for routine use will not improve their accuracy much more than measuring traditional major risk factors alone. The exception to this view comes from a team including Dr Paul Ridker (Brigham and Women's Hospital, Boston, MA) that reports data from the Women's Health Study showing that CRP does improve CHD risk prediction, particularly in the intermediate-risk group [1].

In the July 10, 2006 issue of the Archives of Internal Medicine, a team led by Dr Aaron R Folsom (University of Minnesota, Minneapolis) report the results of the Atherosclerosis Risk in Communities (ARIC) study, which assessed the association of 19 novel risk markers with incident CHD in 15 792 adults followed up since 1987-1989 [2]. The participants underwent a physical examination, including assessment of major risk factors, at the beginning of the study and every three years afterward. At four times during the follow-up period, researchers collected blood and DNA samples for analysis. Patients continue to be tracked for the development of CHD.

Novel markers included measures of inflammation, endothelial function, fibrin formation, fibrinolysis, B vitamins, and antibodies to infectious agents. Change in the area under the receiver-operating-characteristic curve (AUC) was used to assess the additional contribution of novel risk markers to CHD prediction beyond that of traditional risk factors.

Results showed that the basic risk-factor model, which included traditional risk factors (age, race, sex, total and high-density lipoprotein-cholesterol levels, systolic blood pressure, antihypertensive medication use, smoking status, and diabetes), predicted CHD well, as evidenced by an AUC of approximately 0.8. While C-reactive level was significantly associated with CHD, it did not add significantly to the AUC (increase in AUC of 0.003), and neither did most other novel risk factors. Of the 19 markers studied, lipoprotein-associated phospholipase A₂ (LpPLA₂), vitamin B₆, interleukin 6, and soluble thrombomodulin added the most to the AUC (range 0.006-0.011).

Folsom et al conclude: "C-reactive protein level does not emerge as a clinically useful addition to basic risk-factor assessment for identifying patients at risk of a first CHD event. Routine screening is not warranted for any of the other 18 novel risk factors tested, either." But they add that their findings do reinforce the usefulness of measuring the traditional risk factors to identify individuals at risk for CHD for preventive action.

In an accompanying editorial [3], Drs Donald Lloyd-Jones and Lu Tian (Northwestern University Feinberg School of Medicine, Chicago, IL) describe the ARIC study as "one of the premier community-based epidemiologic studies in the US." They say that Folsom et al take an important step beyond merely publishing the relative risks for these novel markers by examining the change in the AUC when each novel risk marker is added individually to the established risk factors.

Can it really be that 19 of the most exciting new markers for CVD, about which there are thousands of published articles and on which entire careers have been based, do not add anything substantial to risk prediction for CVD?

"In this study, only LpPLA₂ increased the AUC significantly (with an increment of 0.006, from 0.774 to 0.780), and the greatest increment in AUC was only 0.011 (from 0.813 to 0.824 for log vitamin B₆), indicating minimal clinical improvement in risk discrimination," the editorialists point out. They also note that Folsom et al also used another method to establish whether these novel markers improved risk prediction—they plotted graphs of the predicted probability of events across risk deciles with the base model of established risk factors and the same model with a novel risk marker added. The curves were essentially superimposable for C-reactive protein and LpPLA₂.

Lloyd-Jones and Tian write: "On their face, these results appear astonishing. Can it really be that 19 of the most exciting new markers for CVD, about which there are thousands of published articles and on which entire careers have been based, do not add anything substantial to risk prediction for CVD?" But they add: "A close reading of the literature indicates that the study by Folsom et al validates that which has been evident all along"—that although some of these novel risk markers have age-adjusted hazard ratios that are similar to those of traditional risk factors such as blood pressure or cholesterol levels, these traditional risk factors have been proven to have a causal role in CHD and are the targets of therapy once high-risk status is determined and so must form the basic risk-prediction models. They add that a new marker is useful only if it corrects a substantial portion of misclassification by the existing risk score, which the 19 novel markers in the ARIC study did not do.

The editorialists say they agree with the conclusions of Folsom et al—that the established risk factors should remain the focus of CHD risk estimation and prevention for now and that routine measurement of any of these 19 novel risk markers for the entire population cannot be recommended. But they add: "We do not rule out the possibility that some markers could be used to better discriminate certain subpopulations, especially those patients near the boundary of high risk based on levels of traditional risk factors."

Lloyd-Jones and Tian are also two of the authors of one of the papers on CRP in the July 4, 2006 issue of the Annals of Internal Medicine [4]. They review the literature published before January 2006 and find no definitive evidence that, for most individuals, CRP adds substantial predictive value above that provided by risk estimation using traditional risk factors for CVD. "Use of CRP may add to risk estimation in a limited subset of individuals who are at intermediate predicted risk according to the Framingham risk score, but many questions must be addressed before CRP is incorporated into risk-prediction algorithms and before universal screening with CRP can be recommended," they conclude.

But the other Annals paper takes a different view. In this article, Ridker and Drs Nancy Cook and Julie Buring (Brigham and Women's Hospital) describe a study in which they compared the clinical utility of global cardiovascular risk-prediction models based on Framingham covariables with and without CRP among the more than 15 000 participants in the Women's Health Study who were followed for an average of 10 years [1].

They report that including CRP in the risk-prediction model improves cardiovascular risk classification in women, particularly among those with a 10-year risk of 5% to 20%. "As shown in our data, approximately 20% of individuals at "intermediate risk" had their risk estimates substantively increased or decreased with improved accuracy when hs-CRP was added to the Framingham covariables. Thus, the use of an hs-CRP-modified Framingham risk score also has the potential to help more patients and physicians to better direct the use of preventive statin therapy to appropriate risk groups," they write.

In an accompanying Annals editorial [5], a group led by Dr George Davey Smith (University of Bristol, UK) say that although they reach different conclusions, the two papers published in the Annals "actually discuss similar data, and the discrepancy between them relates to their assessment of the value of the small improvement in risk prediction provided by CRP." Davey Smith et al conclude: "There are many reasons for skepticism about the role of CRP as a predictor of CHD: CRP may not be causally related to CHD; it remains more expensive than asking patients about their health, lifestyle, and socioeconomic background; and it adds only modest additional predictive ability over conventional risk factors, even in Cook and colleagues' study."