October 3rd, 2012

CRP and Cholesterol Emerge as Equally Strong Predictors of Cardiovascular Risk

In this week’s issue of the NEJM, the Emerging Risk Factors Collaboration, led by Stephen Kaptoge, present data confirming that the inflammatory biomarker C-reactive protein (CRP) provides incremental risk information comparable to that of total cholesterol (TC) and HDL cholesterol (HDL-c). (I am a member of this study group.) The analysis makes four important points.

Click to enlarge

Point 1: As shown in the upper left panel of Figure 1 (right), for a prediction model that already includes age, smoking, systolic blood pressure, and diabetes status, the magnitude of change in the C-statistic associated with adding TC to the model was 0.0043; subsequently adding HDL-c to the latter model changed the C-statistic by 0.0050. These are critical benchmarks for comparison, as all cardiovascular screening programs worldwide include TC and HDL-c.

Point 2: Once TC and HDL-c are included in the prediction model, the incremental change in the C-statistic associated with adding CRP was 0.0039, an effect magnitude fully comparable with that of the two prior benchmarks. As each new biomarker is added to a prediction model, the bar it must clear becomes progressively higher. Therefore, these data confirm that CRP’s value in predicting cardiovascular risk is at least similar to that of standard lipid measures. This is borne out by the observation, shown in Table 1, that the multivariable-adjusted hazard ratio associated with a one standard-deviation increase in CRP level was 1.20 (versus 1.17 for a comparable 1-SD increase in TC level). In other words, if on the basis of discrimination and magnitude of effect lipids are considered important for risk prediction, so too must be CRP.

Point 3: Compared head-to-head, the incremental predictive value of CRP was modestly greater than that of fibrinogen, and the value of combining the two was broadly similar to that of either alone.

Point 4: Most important, by affirming that the role of inflammation in atherothrombosis is comparable to that of cholesterol, these analyses provide support for ongoing trials that target inflammation as a novel treatment strategy. We have recently initiated two large-scale, multinational randomized trials directly testing the inflammatory hypothesis of atherothombosis: one evaluating the interleukin-1-beta inhibitor canakinumab (CANTOS) and the other evaluating low-dose methotrexate (CIRT). If either or both of these trials are positive, we will have direct evidence that reducing inflammation reduces vascular risk.

Whether clinicians ought to use any biomarkers for risk prediction should be based on hard trial evidence and on the concepts of “what works?” and “in whom?”. In primary prevention, randomized trials have shown statins to be highly effective at reducing event rates among patients with elevated LDL-c levels (WOSCOPS), low HDL-c levels (AFCAPS/TexCAPS), and elevated CRP levels (JUPITER). A simple set of evidence-based prevention guidelines, derived from these hard trial outcomes, is easy to follow.

Disclosure: Dr. Ridker, a member of the Emerging Risk Factors Collaboration, is listed as a co-inventor on patents, held by Brigham and Women’s Hospital (BWH), that relate to the use of inflammatory biomarkers licensed to AstraZeneca and Siemens. He is also the principal investigator of the NHLBI-supported CIRT trial and the Novartis-supported CANTOS trial. However, neither Dr. Ridker nor the BWH receives any royalties related to these patents for use of the CRP test in either clinical trial.

Share your thoughts on the CRP findings from the Emerging Risk Factors Collaboration.

8 Responses to “CRP and Cholesterol Emerge as Equally Strong Predictors of Cardiovascular Risk”

  1. Judith Andersen, AB, MD says:

    I think we get it: inflammation, as indicated by increased levels of hs-cRP, is not just a biomarker but a participant in creating cardiovascular risk. Unfortunately, the mechanisms needed to decrease these levels are really problematical: We need to convince our patients to change many things: their eating habits. their exercise habits; their medication habits. We can modulate with medication, but the important modifiers of risk are those that have to do with increased exercise , decreased dietary fat intake, and other modulators of inflammation( . Unfortunately, these are the things we are least good at — and least able to show success at enforcing.

    • Yes correct. The anonymous member is also correct. This is why I have better than average outcomes with my patients. Its critical to integrate diet and exercise counseling along with and understanding of nutritional supplements and drug interactions. Most of my patients use prescriptions as well as nutritional supplements.
      Clinical pearl: try fish oil (4-6 g QD) and Curcumin 1 BID (Tumeric extract) for CRP when weight is not an issue.

      An average follow up visit is 30-45 min. Too bad insurance will not pay for this.

  2. Dan Hackam, MD PhD says:

    Weight reduction – through carbohydrate restriction – is highly effective at lowering hsCRP.

  3. Robin Motz, M.D., Ph.D. says:

    But the precise data stated that in patients at intermediate risk for ASCVD, 500 patients would have to be followed for 10 years for 1 patient to benefit from having the CRP measured. Once again, relative (CRP vs.cholesterol) significance is considerably greater than absoluyte significance.

  4. The title of this blog should read “CRP and Cholesterol Emerge as Equally WEAK Predictors of Cardiovascular Risk”. Nothing “strong” about the predictive ability of CRP or Cholesterol. It is amazing that this kind of statistical nonsense still passes the straight face test of any editorial board.

    Let’s look at real science such as ASCOT-LLA where baseline CRP had no correlation with subsequent events and change in CRP had no effect on outcomes.

    Let’s look at MESA where relative risk of CRP was trivial. In fact the baseline HS-CRP of those who did not have events during the study was higher than the baseline HS-CRP of those who had events.

    Continuing to promote HS-CRP as though it was a legitimate screening test is not consistent with any rigorous evaluation of the literature. It can lead to both inappropriate use of stain therapy in those who have such low risk they cannot benefit from it as well as giving false reassurance to those at risk who should be treated.

  5. Neville Wilson, M.B., Ch.B., D.Fam. Med., M.Sc., D.Obs (COG) SA., B.A. says:

    I agree with William. CRP and Cholesterol are equally weak predictors of cardiovascular risk,and as mere markers for inflammation should not be targets for chemotherapy.
    In CORONA both LDL-C and CRP were significantly lowered by rosuvastatin, to a similar degree as in JUPITER, and yet performed no better than placebo on the primary composite outcome of death from cardiovascular causes, nonfatal MI, or nonfatal stroke. (NEJM 2007;357(12):2248-61).

    Zacho et al have proposed that the relationship between CRP and ischaemic vascular disease is not causal (NEJM 2008 Oct 30, 359:1897)
    Furthermore, “genetic variants associated with lifelong elevations of CRP levels are not associated with an increase in cardiovascular risk” (J Watch Cardiol Oct 29, 2008)

    Improvements in lifestyle is likely to be reflected by decreasing levels of CRP.
    It is well established that exercise, vitamin C and Omega 3 can lower the risk of morbidity and mortality, and give rise to reductions in CRP levels.
    There should be serious concerns that CRP elevations could lead to widespread and inappropriate use of statin therapy.

  6. Ian Neeland, MD says:

    I echo the statements of Drs. Motz and Blanchet. The NRI for CRP in this study was a paltry 1.5%. Compare that to an NRI of 25% for CAC and incident CHD events (Polonsky et al. JAMA. 2010 April 28; 303(16): 1610–1616). Although the link between atherosclerosis and inflammation is clear, the link between clinical assessment of inflammation, treatment guided by that assessment, and clinical outcomes is clearly not. Additionally, the utility of CRP may be limited to those who are not obese (Gupta et al. J Am Coll Cardiol. 2012 Sep 25;60(13):1148-55, and personal communication), which if true diminishes its importance in the clinical arena significantly.

  7. Edgar Abovich, MD says:

    I didn’t find CRP to be helpful in my practice as far as making decisions in individual patients. It is too nonspecific. There are clear differences between large groups of patients in the studies because numerous factors affecting CRP levels cancel each other out.