C-Reactive Protein and Heart Failure after Myocardial Infarction in the Community

Study Population 

Olmsted County, Minnesota, is relatively isolated from other urban centers, and nearly all medical care in virtually every specialty is delivered to residents by a few providers,20 which include the Mayo Clinic and its affiliated hospitals; the Olmsted Medical Center and its affiliated community hospital; local nursing homes; and a few private practitioners. Each provider in the community uses 1 medical record, whereby all medical information for each individual is in a single file. Through the Rochester Epidemiology Project, medical diagnoses, surgical interventions, and other key information from the dossier are abstracted and coded, thereby allowing the linkage of medical records from all sources of care. This provides a unique infrastructure to analyze disease outcomes. The county population was 106,479 in 199020 and increased to 124,277 in 2000 while becoming ethnically more diverse.

Patient Enrollment 

All Olmsted County residents hospitalized between November 2002 and December 2004 and presenting with a troponin T value greater than or equal to 0.03 ng/mL (upper limit of normal defined using the value at which the coefficient of variation is <10%) were prospectively identified within 12 hours of the blood draw through the electronic files of the Department of Laboratory Medicine. Consent was sought from patients (or the next of kin) to measure CRP in unused serum initially stored for additional clinical need. Three patients did not have serum available in which to measure CRP; thus, they were not included in the analyses.

Of the approached patients, 82% consented for the study. Cases were classified using published recommendations,21 and definite or probable myocardial infarction was included as defined by the combination of cardiac pain, biomarkers, and Minnesota coding of electrocardiograms.22 The reliability of this methodology is excellent.1, 23

Time to presentation was defined as the interval from the self-reported onset of symptoms to the first electrocardiogram in hours. Killip class was assessed within 24 hours of admission. Comorbidity was measured by the Charlson index.24 Clinical diagnoses were used to ascertain hypertension, diabetes, hyperlipidemia, family history of coronary disease (defined as coronary disease in first-line male descendants aged < 55 years and in first-line female descendants aged < 65 years), and smoking. The extent of angiographic coronary disease was measured as the number of vessels with stenosis greater than 50%; multivessel disease was defined as at least 2 vessels with greater than 50% stenosis. Recurrent ischemic events included recurrent myocardial infarction or unstable angina and were defined by physicians’ diagnoses.

Biomarkers 

Patients who undergo a clinically indicated blood draw have blood stored to allow for tests without additional phlebotomy. These samples are stored at −70°C and held for 6 days. All patients potentially experiencing a myocardial infarction or next of kin were contacted for permission to use these samples for measurement of CRP.

CRP was measured on serum from the first draw after symptom onset using a latex-enhanced immunoturbidimetric assay on a Hitachi 912 automated analyzer (Hitachi Ltd, Fukushima, Japan) and reagents from Diasorin (Stillwater, Minn). The reference interval for the assay was 0.20 to 0.8 mg/L. The interassay and intra-assay coefficients of variation of the high-sensitivity CRP method were less than 10% for the lower limit and less than 5% for the upper limit; interassay precision was 8.5% at a mean CRP of 1.0 mg/L, 4.6% at a mean CRP of 2.3 mg/L, and 3.4% at a mean CRP of 52.0 mg/L.

Venous samples for troponin T were obtained at the time of admission and 6 to 9 hours after the symptom onset. Serum stored from clinically indicated draws was used to measure creatine kinase-MB (CKMB) with the Elecsys 2010 automated immunochemistry analyzer (Roche Diagnostic, Indianapolis, Ind). Peak troponin T and CKMB values were used in the analyses. Biomarkers were measured in the Immunochemical Core Laboratory of Mayo Medical Laboratories, where all quality control and quality assurance procedures are in place.

Follow-up 

The complete (inpatient and outpatient) medical record for each participant was reviewed by abstractors who were unaware of the CRP value. This process yields information that is complete, because more than 90% of the population receives care at Mayo Clinic or Olmsted Medical Center, and residents are seen on average every 3 years at Mayo Clinic.20 Heart failure, including both inpatient and outpatient events, was validated using the Framingham criteria,25 the reliability of which has been published for Olmsted County studies.2 Outpatient events included all episodes of heart failure that were not diagnosed in the hospital (ie, outpatient diagnoses made by primary care physicians or cardiologists during clinic or nursing home visits). The ascertainment of death incorporated death certificates filed in Olmsted County, autopsy reports, obituary notices, and death certificates obtained from the State of Minnesota Department of Vital and Health Statistics.1, 23 The 10th version of the International Classification of Diseases was used to classify the underlying cause of death. Coronary deaths were defined with codes I20 to I25, including in-hospital and out-of-hospital deaths.26

Statistical Analysis 

Data are presented as frequencies for categoric variables, mean ± standard deviation for continuous variables, or median (25th-75th percentile) for skewed variables.

Trends in variables across tertiles of CRP were tested using the Mantel-Haenszel chi-square test for categoric variables and analysis of variance for continuous variables treating the tertiles of CRP as a 3-level variable. Trends in skewed variables were analyzed after logarithmic transformation.

Correlations between CRP and troponin T and CKMB were assessed with the Pearson correlation coefficient (r) after logarithmic transformation.

Kaplan-Meier curves estimated survival according to CRP tertiles and were compared using the log-rank test. For survival free from heart failure, the analysis was repeated treating death as a competing risk.27 Cox proportional hazards regression estimated the hazard ratio (HR) and 95% confidence interval (CI) for death and heart failure. Recurrent ischemic events were analyzed as a time-dependent covariate. Analyses were performed using SAS statistical software, version 8 (SAS Institute Inc, Cary, NC). The institutional review board approved the study.

C-Reactive Protein, Heart Failure, and Death After Myocardial Infarction 

The mean follow-up was 1.0 ± 0.6 years. At 1 year, 63 patients had died and 182 were still alive, whereas 84 had less than 1 year of follow-up. On the basis of Kaplan-Meier estimates, at 1 year 28% of patients (95% CI, 23%-33%) had experienced heart failure and 20% of patients (95% CI, 15%-24%) had died; 103 recurrent ischemic events occurred.

There was a strong positive graded association between CRP and the long-term development of heart failure. One-year survival free from heart failure was 88% (95% CI, 81%-94%) in the first tertile of CRP, 72% (95% CI, 64%-81%) in the second tertile, and 52% (95% CI, 43%-64%) in the third tertile (P < .01) (Figure 1). These estimates were unchanged after death was analyzed as a competing risk. Compared with patients in the first tertile, patients in the second and third tertiles had a markedly increased risk of heart failure independently of age, sex, and comorbidity. Further adjustment for peak troponin, Q wave, Killip class on admission, previous myocardial infarction, and recurrent ischemic events as a time-dependent covariate did not modify this association (Table 2), nor did further adjustments for cardiovascular risk factors and history of heart failure (data not shown).

View Large Image Download to PowerPoint

Figure 1. Survival free of heart failure (top) and overall survival (bottom) according to CRP tertiles. Hs-CRP = high sensitivity C-reactive protein.

Table 2. Hazard Ratios for Heart Failure and Death According to the Tertiles of C-Reactive Protein

		HR (95% CI)
		Unadjusted	Adjusted⁎	Adjusted†
	Heart failure
	CRP tertile 1 (referent)	1	1	1
	CRP tertile 2	2.45(1.30-4.62)	2.16(1.14-4.10)	1.92(1.01-3.67)
	CRP tertile 3	4.59(2.53-8.36)	2.83(1.50-5.36)	2.47(1.27-4.92)
	P value	<.01	<.01	.019
	Death
	CRP tertile 1 (referent)	1	1	1
	CRP tertile 2	2.39(1.04-5.50)	2.10(0.91-4.84)	1.73(0.72-4.15)
	CRP tertile 3	7.92(3.75-16.73)	4.28(1.95-9.38)	3.96(1.78-8.83)
	P value	<.01	<.01	<.01

HR = hazard ratio; CI = confidence interval; CRP = C-reactive protein.

⁎ Adjusted for age, sex, comorbidity. † Adjusted for age, sex, comorbidity, peak troponin T, Q wave, Killip class, previous myocardial infarction, and recurrent ischemic events as time-dependent covariate.

During follow-up, 75 deaths occurred. There was a strong positive association between CRP and 1-year survivals. One-year survivals were 93% (95% CI, 88%-98%) among patients in the first CRP tertile, 84% (95% CI, 77%-91%) in the second tertile, and 62% (95% CI, 54%-72%) in the third tertile (P < .01) (Figure 1). Patients in the third tertile had an 8-fold increase in the unadjusted risk of death compared with patients in the first tertile (HR 7.92, 95% CI, 3.75-16.73; P < .01). After further adjustment for age, sex, and comorbidity, patients in the third tertile had a 4-fold increase in the risk of death (adjusted HR 4.28, 95% CI, 1.95-9.38; P < .01) (Table 2). Further adjustment for peak troponin T, Q wave, Killip class, previous myocardial infarction, and recurrent ischemic events as a time-dependent covariate did not modify this association (Table 2). Adjustment for cardiovascular risk factors and history of heart failure did not modify this association (data not shown).

C-Reactive Protein, Heart Failure, and Death After Myocardial Infarction 

Heart failure and death remain frequent during the first year after an acute myocardial infarction, reaffirming data from earlier studies.2, 3, 4

After myocardial infarction, studies on CRP focused on all-cause and cardiac deaths, which were evaluated mainly in case series13, 17 and subgroup analysis of data from clinical trials.8, 9, 10, 11, 12 Yet, studies of patients referred to tertiary centers do not represent the entire spectrum of patients with myocardial infarction, and secondary analyses of clinical trials8, 9, 11, 19 pertain to selected patients who often have fewer comorbidities.29 Few studies have investigated the association of postmyocardial infarction heart failure with CRP.12, 14, 30, 31 Although suggesting a positive association, the studies focused chiefly on short-term risk,14, 30 used low-sensitivity assays,12 or considered only patients referred to intensive coronary care units and heart failure episodes that required hospitalization.31

The present study pertains to all patients with myocardial infarction within a geographically defined community, which enhances its generalizability. It indicates that elevated CRP is associated with a large increase in the risk of heart failure and death during the first year after myocardial infarction independently of known risk indicators, including other biomarkers, comorbidity, and recurrent ischemic events. To this end, although CRP was higher among subjects with higher Killip class, its predictive value for heart failure during follow-up is independent of Killip class, thus providing incremental information. The graded positive association between CRP and heart failure and death is consistent with a dose-response pattern.

Because few studies have investigated the association of CRP and heart failure, the mechanism for this association is unknown. It is unlikely related to recurrent ischemia because the associations between CRP and heart failure and death were not altered by adjusting for recurrent ischemic events. Alternative explanations include an exaggerated immune response to myocardial injury,32, 33, 34 as demonstrated by the association of inflammatory cytokines with ventricular remodeling after myocardial infarction, ejection fraction, and heart failure progression.33, 35, 36 Finally, experimental animal studies showed that CRP may have direct harmful effects on the ischemic myocardium. Griselli et al37 demonstrated that after coronary ligation, injection of CRP increased infarct size. Barrett et al38 demonstrated that elevated endogenous CRP is associated with an increase in ischemia/reperfusion injury. All these explanations remain hypothetical and addressing them directly is beyond the scope of this study, which, however, underscores the need of doing so.

A recent study reported that CRP measured at discharge was not associated with the combined end point of death, myocardial infarction, unstable angina, urgent revascularization, and stroke.39 Heart failure was not examined as an outcome such that these findings further support our hypothesis that the association between CRP and death is not mediated by ischemia, but rather that heart failure plays an important role.

Previous studies reported conflicting results on whether CRP was associated with myocardial infarction size as assessed by cardiac biomarkers.18, 19, 30 In acute coronary syndromes, a significant although weak association of CRP with CKMB40 and troponin was reported.8, 9, 11 Among community cases that met rigorous criteria for acute myocardial infarction,21, 41 CRP was not associated with higher troponin, CKMB, or Q waves. Thus, when measured early after myocardial infarction, CRP does not seem to be related to myocardial injury.7, 14, 42 This is consistent with the fact that CRP provides incremental information over troponin and CKMB to predict death and heart failure, as documented in this article.

These findings provide indirect evidence against the hypothesis that ischemia is the mechanism whereby CRP relates to heart failure and death, an observation further supported by the fact that greater CRP is associated with these outcomes independently of recurrent ischemic events. Indeed, our results resonate with data14, 43 suggesting that CRP elevation does not reflect plaque inflammation but rather the response of the “downstream myocardium” to the necrosis, and with the report that, compared with other cardiovascular risk predictors, CRP is only a modest predictor of recurrent ischemic events in patients with coronary disease.44

Several predictors of heart failure after myocardial infarction have been examined, such as age, female sex, diabetes, hypertension,45, 46 brain natriuretic peptide, ejection fraction,47 and wall motion score index.47, 48 Although no study has directly compared CRP with other predictors of heart failure after myocardial infarction, our finding of an almost 4-fold increase in the risk of heart failure was independent of several other known risk markers.

Some potential limitations should be acknowledged to interpret the data. Imaging to quantify myocardial infarction size was not performed, and we relied on electrocardiography and biomarkers, both imperfect surrogate measures.49 This may play a role in the lack of association of CRP with angiographic coronary disease or biomarkers. CRP changes dynamically after myocardial infarction, and in most studies, such as the present one, it was measured at only 1 time point;34, 42, 50 this may account for discrepant results observed in other studies.

The strengths of our study include its prospective community-based design, whereby all consecutive patients in a geographically defined population were included and among whom CRP was measured, as recommended,50 within 24 hours of admission using a high-sensitivity assay. Additional important strengths include rigorous ascertainment approaches that rely on standardized criteria to define myocardial infarction and heart failure, and the comprehensive nature of the follow-up, which includes all inpatient and outpatient events. These important methodologic strengths optimize the robustness of our findings.