Author + information
- Peter Bogaty, MD⁎ (, )
- Serge Simard, MSc,
- Gilles R. Dagenais, MD, FACC and
- James M. Brophy, MD, PhD, FACC
- ↵⁎Quebec Heart Institute/Laval Hospital, 2725 Chemin Ste-Foy, Quebec City, Quebec G1V 4G5, Canada
We appreciate the interest expressed by Dr. Biasucci and colleagues in our work (1). We prospectively found, in 1,210 patients representing a broad spectrum of acute coronary syndromes, that the inflammatory marker C-reactive protein (CRP), whether measured on admission, at hospital discharge, or 30 days later, did not have incremental clinical utility to predict the occurrence of death, myocardial infarction, or unstable angina (UA) with electrocardiogram changes at 1 year (primary end point) (1). Dr. Biasucci and colleagues argue that our inclusion of patients with significant acute myocardial necrosis (about two-thirds), in itself a stimulator of CRP release, and those with concomitant inflammatory conditions confounds the predictive ability of CRP. We believe this points precisely to the major limitation of using a nonspecific marker like CRP for coronary risk prediction. Similarly, why not also exclude patients with other possible confounders like obesity, depression, fatigue, diet, medications, and so on? Who would ultimately remain to be tested with CRP in clinical practice? Indeed, is it not questionable to exclude patients with inflammatory conditions when background (nonspecific) inflammation as reflected by raised CRP might increase coronary risk in these patients (2)?
Nonetheless, we report here the predictive ability of CRP in our study when the analysis is restricted to the 461 UA patients without significant myocardial damage. The odds ratio (OR) and 95% confidence intervals (CIs) for admission CRP to predict the primary end point at 1 year was 1.21 (95% CI: 0.97 to 1.51; p = 0.1) and adjusted it was 1.07 (95% CI: 0.85 to 1.36; p = 0.6). The OR of CRP measured at hospital discharge was 1.03 (95% CI: 0.83 to 1.28; p = 0.8) and adjusted OR was 0.99 (95% CI: 0.79 to 1.25; p = 1.0). The OR of CRP measured 30 days later was 1.16 (95% CI: 0.88 to 1.52; p = 0.3) and adjusted OR was 1.10 (95% CI: 0.83 to 1.47; p = 0.5). Findings were unchanged when the analysis was restricted to the 326 UA patients with negative troponin levels. Finally, results were unchanged when patients with concomitant inflammatory conditions (about 25%) were excluded.
Dr. Biasucci and colleagues suggest using a CRP cutoff of 10 mg/l instead of analyses with continuous values. Besides the potential loss of information with this approach, such a cutoff is quite arbitrary. The scientific statement of the Centers for Disease Control and Prevention/American Heart Association only speculated on use of this value without supporting references and did not actually recommend it (3). Nevertheless, we redid all of the previously mentioned analyses using the 10 mg/l cutoff and found it to have no clinical utility.
Finally, Dr. Biasucci and colleagues critique our study for not providing pathogenetic insight into the relation between CRP and risk of death. Our study was designed as a clinical, not mechanistic, study. The univariate link we found between CRP and death dissolved in multivariate analyses. Our study supports the notion that clinicians do not need CRP to know who are the sicker patients who are more likely to die as well as to have raised inflammatory markers (4). While we appreciate the pathogenetic importance of inflammation in acute coronary disease, this is insufficient to establish a seamless link to clinical use of an inflammatory marker for risk prediction. To CRP or not to CRP? For risk prediction in our study, CRP appears to be “dead, for a ducat, dead” (the Bard gratefully acknowledged).
- American College of Cardiology Foundation
- Bogaty P.,
- Boyer L.,
- Simard S.,
- et al.
- Pearson T.A.,
- Mensah G.A.,
- Alexander R.W.,
- et al.