Inflammation and future risk of symptomatic heart failure in patients with stable coronary artery disease Alon Eisen, MD, a,e Michal Benderly, PhD, b,c,e Solomon Behar, MD, d,† Uri Goldbourt, PhD, b,d and Moti Haim, MD a Petah-Tikva, Tel Aviv, and Tel-Hashomer, Israel
Background Heart failure (HF) carries poor prognosis in coronary artery disease (CAD) patients despite advances in therapy. Inflammation predicts recurrent cardiovascular events in CAD patients. It is unknown whether increased levels of inflammatory markers are associated with incident HF in these patients. Aim The aims of this study were to evaluate the association between inflammatory markers and future HF risk in patients with stable CAD and to explore possible mediation by myocardial infarction (MI). Methods The study comprised 2,945 patients with stable CAD without HF at baseline during a median follow-up of 7.9 years. Inflammatory baseline markers were the basis of this study. Results Heart failure was diagnosed in 508 patients (17.2%). Patients who developed HF were older and had more often previous MI, diabetes, hypertension, and peripheral vascular disease. Baseline levels of C-reactive protein (CRP), fibrinogen, and white blood cells (WBCs) were significantly higher in patients who developed HF compared with those who did not. Age-adjusted incident HF rates were related to elevated baseline inflammatory markers in a dose-response manner. Adjusting for multiple confounders, the HF hazard ratios were 1.38 (95% CI 1.11-1.72), 1.33 (95% CI 1.071.66), and 1.36 (95% CI 1.10-1.68) for the third tertiles of CRP, fibrinogen, and WBC levels, respectively. Hazard ratio for the fifth quintile of a combined “inflammation score” was 1.83 (95% CI 1.40-2.39). Mediation by MI preceding the HF onset during follow-up accounted for 10.4%, 10.8%, and 8.6% of the association of subsequent HF with CRP, fibrinogen, and WBC, respectively. Conclusions Increased levels of CRP, fibrinogen, and WBC are independently related to the incidence of HF in patients with stable CAD. (Am Heart J 2014;167:707-14.)
Heart failure (HF) is a major health problem with increasing incidence. 1 Traditional risk factors for HF include chronic coronary artery disease (CAD), myocardial infarction (MI), hypertension (HTN), diabetes mellitus, older age, and others. 1
From the aCardiology Department, Rabin Medical Center, Petah-Tikva, Israel, bDepartment of Epidemiology and Preventive Medicine, School of Public Health, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel, cGertner Institute for Epidemiology and Health Research Policy, Sheba Medical Center, Tel-Hashomer, Israel, and dThe Israel Society for the Prevention of Heart Attacks, Neufeld Cardiac Research Institute, Sheba Medical Center, Tel-Hashomer, Israel. e
A.E. and M.B. contributed equally to the study and are co-first authors of this manuscript. Deceased. Submitted July 6, 2013; accepted January 4, 2014. Reprint requests: Alon Eisen, MD, Cardiology Department, Rabin Medical Center, Jabotinsky St, 49100 Petah, Tikva, Israel. E-mail: [email protected] 0002-8703/$ - see front matter †
© 2014, Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.ahj.2014.01.008
Increased levels of various inflammatory markers predict increased risk of HF hospitalizations and poor outcome in healthy persons or in patients with preexisting HF. 2-5 Several studies reported an association between increased inflammatory marker levels and incidence of HF in patients with CAD, mainly in patients who underwent an acute event such as STelevation MI. 6-10 Data of such an association among patients with stable CAD are limited. In addition, most studies examined a single inflammatory marker, for example, white blood cell (WBC) count or C-reactive protein (CRP). It is also unclear to what extent inflammatory markers are directly associated with the development of HF or, rather, related to the extent of CAD and interim occurrence of MI, which subsequently leads to HF. The aims of the present study were to assess the association between levels of inflammatory markers and future incidence of HF in patients with stable CAD and no
American Heart Journal May 2014
708 Eisen et al
HF at baseline and to explore the role of MI as a possible mediator of such an association.
Hypertension was defined based on the study physician's report or blood pressure N140/90 mm Hg when measured at the time of screening. Diabetes was defined based on reports or treatment with a hypoglycemic drug.
Methods Patients
Laboratory and analytical methods
A total of 3,122 patients (2,854 men, 91%) participated in the Bezafibrate Infarction Prevention (BIP) study, a placebocontrolled randomized trial that assessed the efficacy of bezafibrate in reducing the risk of cardiac events and mortality. 11,12 Inclusion criteria for men and women comprised the following: age of 45 to 74 years, history of MI ≥6 months but b5 years before enrollment into the study and/or stable angina pectoris confirmed by coronary angiography, and/or radio-nuclear studies or standard exercise tests. In addition, a lipid profile of serum total cholesterol between 180 and 250 mg/dL, low-density lipoprotein cholesterol ≤180 mg/dL (≤160 mg/dL for patients b50 years), high-density lipoprotein cholesterol (HDL-C) ≤45 mg/dL, and triglycerides ≤300 mg/dL was required. Exclusion criteria included clinically significant HF (New York Heart Association classes III-IV) during the 3 months preceding the first screening visit, recent acute coronary syndrome (3 months for unstable angina pectoris and 6 months post-MI), idiopathic cardiomyopathy, and life-threatening arrhythmias within 3 months of screening. Institutional ethics committees in each of the participating centers and the central national ethics committee approved the study. Follow-up included an interview and a physical examination by a study cardiologist every 4 months for the study duration (5-7 years) and an additional 2 years of poststudy termination follow-up. Median follow-up length was 7.9 years (interquartile range 7.1-8.7). Patients with baseline HF (n = 171) or missing data on all inflammatory markers (n = 6) were excluded from the current analysis.
Blood samples for measurement of serum lipids, blood chemistry, complete blood count, and other laboratory tests were collected at randomization (baseline). All laboratory analyses were performed in a single central laboratory using standard automated procedures with commercial kits. Accuracy and precision for lipid measurements were under periodic surveillance by the Centers for Disease Control and Prevention/ National Heart, Lung, and Blood Institute's Lipids Standardization Program and by the Wellcome-Murex Diagnostic Clinical Chemistry Quality Assessment Program. Inflammatory markers at baseline (fibrinogen, CRP, and WBC) are the basis of this study. Fibrinogen was determined in fresh citrated plasma using a kinetic method, as described by Hemker et al. 13 High sensitive CRP concentrations were measured in samples of citrated plasma stored at −70°C obtained at baseline on an IMMULITE 2000 analyzer (Diagnostics Products Corporation, Los Angeles, CA) with the manufacturer's reagent solidphase, chemiluminescent immunometric assay. The validity of using plasma citrate compared with serum was tested by analysis of samples from 30 randomly selected individuals. C-reactive protein levels in plasma citrate were 68% of the serum levels (coefficient of variation 3.0, range 65%-74%). The correlation between plasma and serum levels was high (r = 0.998). Values of inflammation marker tertiles were determined for each gender separately: For men, CRP b2.3 (2.3-5.1), N5.1 mg/ dL; fibrinogen b310.3 (310.3-366.8), N366.8 mg/dL; WBC b5.9 (5.9-7.3), N7.3 × 10 3/μL. For women, CRP b2.9 (2.9-6.6), N6.6 mg/dL; fibrinogen b342.2 (342.2- 402.4), N402.4 mg/dL; WBC b5.9 (5.9-7.2), N7.2 × 10 3 /μL. C-reactive protein, fibrinogen, and WBC values were missing for 121, 96, and 151 patients, respectively.
Variable definition and classification Heart failure incidents were ascertained in 2 steps. First, HF diagnosis was based on diagnoses made on each study visit by the study cardiologist in each participating center. This was based on discharge diagnoses from hospitalization records as well as on the presence of HF symptoms, physical findings, and other supporting evidence such as congestion on chest x-ray. Second, all the cases were validated by reviewing the original study records by one of the authors (M.H.) and verifying that the diagnosis was based on documented hospital records, symptoms, or imaging. Functional capacity was classified according to the New York Heart Association classification. Metabolic syndrome (MS) was defined based on the Adult Treatment Panel III report classification as at least 3 of the following: HDL-C b40 mg/dL (1.04 mmol/L), triglycerides N150 mg/dL (1.69 mmol/L), blood pressure N135/85 mm Hg, glucose level N110 mg/dL (6.11 mmol/L), and replacing waist circumference criteria, which was not available, with body mass index (BMI) N28 kg/m 2. Glomerular filtration rate (GFR) was estimated using the formula derived by the Modification of Diet in Renal Disease study: 186 [serum creatinine (mg/dL) − 1.154] × [age (y) − 0.203]. Renal failure was defined as GFR b60 mL/min per 1.73 m 2. Baseline blood pressure is the mean of 2 preinclusion measurements.
Statistical analysis Data were analyzed using the SAS software (version 9.2; SAS, Cary, NC). Characteristics of patients are presented as frequencies or mean ± SD unless otherwise specified and compared by χ 2 tests for categorical variables or analysis of variance for normally distributed continuous variables. Triglycerides, fibrinogen, and CRP, which were not normally distributed, are presented as geometric mean (95% CI) and compared by the nonparametric Kruskal-Wallis test. Trends in proportions by tertiles of inflammatory markers were assessed by the Mantel-Haenszel χ 2 test. Direct adjustment using the entire group included in the analysis as the reference group was used for computation of age-adjusted mortality rates per 1,000 person-years (PY). To account for the competing risk of premature death that may preclude HF onset, HF incidence probabilities over followup time according to inflammation markers were estimated by the cumulative incidence function using the method suggested by Gooley et al. 14 Principal component analysis with varimax (orthogonal) rotation was used to create a single score for the 3 inflammatory markers. A proportional subdistribution hazards model 15 was fitted to the data using a SAS macro developed by Kohl and Heinze. 16 The macro first modifies the input data to
American Heart Journal Volume 167, Number 5
account for competing risks and then applies SAS standard Cox regression procedure, PROC PHREG, using weights and counting-process style of specifying survival times to the modified data set as suggested by Geskus. 17 A fixed set of variables including variables reported as possible confounders or mediators or identified in univariate analysis in the current study (age, history of MI, angina pectoris, peripheral vascular disease, diabetes, and MS) were included in all models. The predictive discrimination ability of each model was evaluated using a C-statistic 18 corresponding to the area under a receiver operating characteristics curve. The C-statistic for the multivariable models for each inflammatory marker was around 0.7 for all models. The validity of the proportional hazard assumption for each of the inflammatory markers was tested by running a Cox proportional hazards model including time-dependent explanatory variables for each inflammatory marker to test the assumption of no time-dependent effect. No significant deviation from the proportional hazard assumption was detected. The extent to which the association between inflammation and HF onset was mediated by MI was explored according to the approach suggested by Baron and Kenny. 19 Mediation models were developed using a SAS macro developed by Jasti et al. 20 The Sobel test was used to assess the significance of mediation. No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents.
Results Of 2,945 patients without HF at baseline, 508 patients (17%) developed HF over a median period of 7.9 years. Patients who developed HF during follow-up were older and had a higher prevalence of previous MI, angina pectoris, HTN (and correspondently, higher systolic and diastolic baseline blood pressure), diabetes, and MS compared with patients without HF (Table I). Patients who later developed HF were treated more frequently by nitrates, diuretics, digitalis, and angiotensin-convertingenzyme inhibitors at baseline (Table I). There was no difference in the rate of HF in patients allocated to the bezafibrate or placebo study arms. Baseline mean levels of inflammatory markers in patients with HF were higher than among those who remained free of HF during follow-up (CRP 4.2 mg/dL [95% CI 3.8-4.5] vs 3.4 mg/dL [95% CI 3.3-3.5], fibrinogen 353.0 mg/dL [95% CI 347.1-359.0] vs 339.8 mg/dL [95% CI 337.0-342.7], WBC 7.1 × 10 3/μL ± 1.8 vs 6.7 × 10 3/μL ± 1.9 [P b .001 for all]). The incidence of MI during follow-up was highest among patients in the third tertile of baseline CRP, fibrinogen, and WBC levels. Data regarding coronary revascularization and MI preceding HF are depicted in Table II. The age-adjusted HF incidence rate (per 1,000 PY) increased with increasing tertiles of the various
Eisen et al 709
Table I. Baseline characteristics of patients with and without HF during follow-up HF during follow-up Characteristics Age (mean ± SD) Men, n (%) History, n (%) MI Angina pectoris PVD Stroke HTN Diabetes MS BMI, mean ± SD Obesity (BMI N30) GFR ≤60 mL/min per 1.73 m2 Lipid profile Total cholesterol, mg/dL HDL-C, mg/dL LDL-C, mg/dL Past smoking Smoking Medical treatment β-Blockers Antidiabetic Nitrates Antiplatelets Digitalis Diuretics ACE-I Blood pressure, mm Hg (mean ± SD) Systolic Diastolic Pulse pressure
No (n = 2,437) Yes (n = 508)
P
59.8 ± 6.8 2238 (91.7)
61.5 ± 6.5 461 (90.2)
b.0001 NS
1844 (75.7) 1327 (54.5) 66 (2.7) 25 (1.0) 1007 (41.3) 215 (8.8) 1086 (44.9) 26.6 ± 3.2 298 (12.2) 432 (17.8)
424 (83.8) 335 (65.9) 29 (5.8) 10 (2.0) 271 (53.3) 71 (14.0) 287 (56.9) 27.3 ± 3.5 108 (21.3) 106 (20.9)
b.0001 b.0001 .0005 .07 b.0001 .0004 b.0001 b.0001 b.0001 NS
212.4 ± 17.5 34.7 ± 5.5 148.9 ± 16.5 1419 (58.2) 289 (11.9)
211.4 ± 17.9 34.0 ± 5.6 147.6 ± 16.5 297 (58.5) 62 (12.2)
NS .006 NS NS NS
971 (39.8) 102 (4.2) 1160 (47.6) 1741 (71.4) 46 (1.9) 219 (9.0) 241 (9.9)
191 (37.6) 41 (8.0) 299 (58.9) 508 (67.3) 27 (5.3) 126 (24.8) 91 (17.9)
NS .0002 b.0001 .06 b.0001 b.0001 b.0001
132.7 ± 15.6 80.9 ± 7.9 51.8 ± 11.6
137.2 ± 16.9 82.3 ± 8.2 54.9 ± 12.9
b.0001 .0003 b.0001
Abbreviations: NS, Not significant; PVD, peripheral vascular disease; LDL-C, lowdensity lipoprotein cholesterol; ACE-I, angiotensin-converting enzyme inhibitor.
inflammatory markers (Figure 1). The cumulative incidence of HF according to tertiles of inflammatory markers is depicted in Figure 2. Age-adjusted hazard ratios (HRs) of HF incidence are presented in Table III. For all inflammatory markers, elevated levels were associated with increased risk of developing HF during the follow-up period. Hazard ratios for developing HF in the third tertile as compared with the first tertile ranged between 1.41 (for fibrinogen) and 1.61 (for CRP) (Table III). Adjusting for additional possible confounders or mediators attenuated the HRs; still, the third tertile of all inflammatory markers was associated with HF (Figure 3). Similar results were obtained after adjusting for diuretic or antiplatelet treatment (data not shown). Patients were divided to quintiles according to their “inflammation score,” which was composed of CRP, WBC, and fibrinogen (Figure 4). Patients in the fifth quintile had a remarkably significantly higher incidence
American Heart Journal May 2014
Tertile values: for men, CRP b2.3 (2.3-5.1), N5.1 mg/dL; fibrinogen b310.3 (310.3-366.8), N366.8 mg/dL; WBC b5.9 (5.9-7.3), N7.3 × 103/μL; for women, CRP b2.9 (2.9-6.6), N6.6 mg/dL; fibrinogen b342.2 (342.2-402.4), N402.4 mg/dL; WBC b5.9 (5.9-7.2), N7.2 × 103/μL. Abbreviations: PTCA, Percutaneous transluminal coronary angioplasty; CABG, coronary artery bypass graft.
.71 .62 .14 108 (11.4) 161 (16.8) 146 (15.2) 99 (10.7) 163 (17.6) 105 (11.4) 101 (10.5) 159 (16.5) 159 (16.5) 118 (12.6) 142 (15.2) 123 (13.2) 97 (10.3) 162 (17.2) 110 (11.7)
97 (10.2) 175 (18.5) 151 (15.9)
.97 .45 .007
97 (10.2) 154 (16.1) 119 (12.5)
122 (13.1) 166 (17.8) 111 (11.9)
.81 .81 .009
104 (11.4) 145 (15.9) 122 (13.4)
.0008 775 (80.8) 685 (75.3) 685 (74.3) 758 (79.0) 712 (76.3)
Previous MI, n (%) Revascularization during follow-up, n (%) PTCA CABG MI during follow-up, n (%)
722 (76.6)
747 (79.0)
.23
733 (76.8)
711 (76.3)
.27
2 1 P 3 2 1 1
2
3
P
Fibrinogen (mg/dL) tertiles CRP (mg/dL) tertiles
Table II. Coronary revascularization and MI preceding HF onset among study patients according to inflammatory markers tertiles
WBC (×10 3/μL) tertiles
3
P
710 Eisen et al
Figure 1
Age-adjusted HF rates (per 1,000 PY) by baseline tertiles of inflammatory markers.
of HF as compared with the first quintile (36.9/1,000 PY vs 14.3/1,000 PY). Cumulative incidence curve for HF during follow-up according to quintiles of inflammation score is depicted in Figure 5. After multivariable analysis, the HR for HF in the fifth quintile as compared with the first quintile according to inflammatory score was 1.83 (95% CI 1.40-2.39). To assess the possibility of interaction between MI and subsequent onset of HF during follow-up, we repeated the analyses separately for patients with and without MI preceding HF onset. In patients with MI (n = 322), 92 patients (28.6%) developed HF during follow-up. The multivariate adjusted HRs for HF according to baseline inflammatory marker tertiles were 1.22 (95% CI 0.702.12) and 1.37 (95% CI 0.81-2.31) for the second and third tertiles of CRP, respectively; 1.00 (95% CI 0.581.78) and 1.08 (95% CI 0.66-1.76) for the second and third tertiles of fibrinogen, respectively; and 1.06 (95% CI 0.63-1.76) for both the second and third tertiles of WBC levels. Among patients without MI during followup (n = 2,623), 416 patients (15.6%) were diagnosed with HF. In these patients, the multivariate adjusted HRs for HF were 1.14 (95% CI 0.89-1.47) and 1.39 (95% CI 1.09-1.76) for the second and third tertiles of CRP, respectively; 1.30 (95% CI 1.01-1.67) and 1.39 (95% CI 1.08-1.78) for the second and third tertiles of fibrinogen, respectively; and 1.02 (95% CI 0.80-1.32) and 1.40 (95% CI 1.11-1.77) for the second and third tertiles of WBC levels, respectively. To further examine the role of MI as a mediator between an inflammatory marker and HF, mediation analysis was performed. Mediation by MI before HF onset (before or during follow-up) accounted for 10.4% of the association between CRP and HF (P = .004). The degree
American Heart Journal Volume 167, Number 5
Tertile values: for men, CRP b2.3 (2.3-5.1), N5.1 mg/dL; fibrinogen b310.3 (310.3-366.8), N366.8 mg/dL; WBC b5.9 (5.9-7.3), N7.3 × 103/μL; for women, CRP b2.9 (2.9-6.6), N6.6 mg/dL; fibrinogen b342.2 (342.2-402.4), N402.4 mg/dL; WBC b5.9 (5.9-7.2), N7.2 × 103/μL.
910 146 (16.0) 1.07 (0.86-1.34) 924 132 (14.3) 1.0 961 193 (20.1) 1.41 (1.14-1.75) 933 169 (18.1) 1.27 (1.02-1.59) 955 130 (13.6) 1.0 947 201 (21.2) 1.61 (1.30-1.99)
3 2
934 161 (17.2) 1.24 (0.99-1.55) 943 130 (13.8) 1.0
The main finding of our study is that elevated levels of baseline inflammatory markers were associated with increased risk of developing HF among patients with stable CAD. This held true after adjusting for multiple confounders associated with HF and inflammation. By using mediation analysis, we found that the association between the inflammatory markers and HF was only partially explained by the occurrence of MI before onset of HF. Inflammation is associated with the development of CAD in different populations. 21-26 There is also growing evidence that supports the importance of inflammation in the development as well as progression of HF, described as the “cytokine hypothesis.” 27-35 In recent years,
1
Discussion
CRP (mg/dL) tertiles
of mediation observed for WBC and fibrinogen was of similar magnitude (8.6% and 10.8%, respectively).
Table III. Heart failure by tertiles of CRP, fibrinogen, and WBC count
Cumulative incidence of HF according to tertiles of CRP (A), fibrinogen (B), and WBC (C).
No. of patients HF incidence (%) Age-adjusted HR (95% CI)
2 2 1
Fibrinogen (mg/dL) tertiles
3
1
WBC (×10 3/μL) tertiles
3
Figure 2
960 201 (21.0) 1.48 (1.21-1.83)
Eisen et al 711
American Heart Journal May 2014
712 Eisen et al
Figure 3
Figure 5
Cumulative incidence of HF according to quintiles of inflammation score. Multivariate adjusted HF HR by inflammatory marker levels (adjusted for age, angina pectoris, diabetes, HTN, previous MI, and peripheral vascular disease).
Figure 4
Age-adjusted rate per 1,000 PY of HF according to inflammation score quintiles.
accumulating data support the association between elevated inflammatory biomarkers and incidence of HF. 2-10 In patients with HF, circulating inflammatory markers such as tumor necrosis factor α, interleukin 6, and CRP were shown to have prognostic value and, therefore, investigated as possible targets for HF therapy. 36,37 Elevated levels of CRP were reported to be associated with increased risk of developing HF in the general population, 4,9 patients with atherosclerotic risk factors, 9 obese patients 5, and the elderly. 38 Elevated WBC counts have also been linked to subsequent HF, in particular, in patients presenting with ST-elevation MI as well as in the general population. 7-10 Our study is the first to our knowledge that investigated the association between inflammation and future HF in a large cohort of stable CAD patients without HF at baseline, a population at risk for developing HF. We
have demonstrated that, regardless of traditional risk factors, elevated levels of CRP, fibrinogen, and WBC count were all associated in a dose-response manner with the risk of HF. Because MI is the leading cause of HF in CAD patients, we asked ourselves whether MI interacts with inflammation in the process of developing HF. Among patients without MI during follow-up, CRP, fibrinogen, and elevated WBC were all associated with future risk of HF. These findings suggest that inflammation may have a role in HF development at least partly independent of recurrent cardiovascular events such as MI. The mediation analysis provided support to our hypothesis, showing that only a modest portion of the association between inflammation and HF is mediated by MI. Several mechanisms could explain this association: an inflammatory state is associated with weight loss, anemia, endothelial dysfunction, and a prooxidative stress 35; CRP may promote HF by activating the complement system, stimulating cytokine production, and causing myocyte loss 32; and finally, elevated WBC count may promote abnormal leukocyte aggregation, vessel obstruction, endothelial injury, and decreased perfusion in heart muscle. 35 Therefore, although MI is the leading cause of HF in CAD patients, it seems that the development of HF involves multiple mechanisms. 39,40 Most studies that examined the association between inflammation and future risk of HF used a single inflammation marker, mainly CRP or WBC level. The correlation between inflammatory markers hampers any attempt to evaluate their joint effect in multivariable analysis. In this study, we were able to study such an effect by creating a score accounting for all 3 markers, using principal component analysis. Using this score, we found that the extent and “burden” of inflammation were related to the incidence of HF with the fifth score quintile carrying 1.8 times the risk of the first quintile. This finding, which was consistent over the follow-up years, suggests a quantitative as well as a qualitative
American Heart Journal Volume 167, Number 5
association between inflammation and HF in patients with CAD. In our cohort, patients who developed HF during follow-up had higher levels of inflammatory markers at baseline with the highest HF incidence in patients who had highest values of baseline inflammatory markers. The clinical implication of this finding may be of importance in identifying stable CAD patients who are at greater risk for developing future HF. These patients might need a more aggressive risk stratification and management. Nevertheless, it is unclear whether inflammation per se is contributing to the development of HF or whether the elevated inflammatory markers reflect another underlying mechanism. Further research is necessary to explore the possible efficacy of treatment of CAD patients by direct antiinflammatory interventions to delay HF development, if indeed inflammation has a direct role. Some limitations apply to our analysis. First, this study is retrospective, and its observational nature does not allow determining if the relationship between inflammation and subsequent HF onset is causal. Second, echocardiography studies were not available and, therefore, were not included in this study. Heart failure was not a prespecified primary end point in the BIP study, and therefore, HF diagnoses were largely based on the clinical judgment of the study cardiologist in each center. Nevertheless, all HF cases were validated by reviewing the patient files and verifying the diagnosis. In addition, although HF hospitalizations are considered to be the most well-delineated HF outcome measure, 41 we also included patients diagnosed with HF who were not hospitalized. We had no data on the WBC differential count and, therefore, could not explore whether granulocyte count, which was examined in prior studies, is a stronger predictor for HF as compared with other WBC components. Several baseline characteristics differ between patients who developed HF versus those who did not. Still, the association persisted (albeit attenuated) after adjustment for these differences. Data on inflammatory diseases and antiinflammatory drug use during study period were lacking. Inflammatory markers used in the study are imperfect and subject to fluctuation. Measurement of other more specific biomarkers may have improved the scientific validity of the study. Furthermore, baseline evaluation does not necessarily reflect long-term inflammatory status. The strengths of data including the large cohort and long follow-up allowed us to provide robust evidence of an association between inflammation and risk of HF in stable CAD patients.
Conclusion Increased levels of CRP, fibrinogen, and WBC were independently associated with the incidence of HF in patients with stable CAD. Further studies are required to
Eisen et al 713
determine whether inflammatory markers may be helpful in prediction and prevention of HF in this population.
Disclosures All authors are affiliated to Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel. All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. Relationship with industry and financial disclosure— none to disclose.
References 1. Ho KK, Anderson KM, Kannel WB, et al. Survival after the onset of congestive heart failure in Framingham Heart Study subjects. Circulation 1993;88:107-15. 2. Vasan RS, Sullivan LM, Roubenoff R, et al. Inflammatory markers and risk of heart failure in elderly subjects without prior myocardial infarction: the Framingham Heart Study. Circulation 2003;107:1486-91. 3. Cesari M, Penninx BW, Newman AB, et al. Inflammatory markers and onset of cardiovascular events: results from the Health ABC Study. Circulation 2003;108:2317-22. 4. Kardys I, Knetsch AM, Bleumink GS, et al. C-reactive protein and risk of heart failure. The Rotterdam Study. Am Heart J 2006;152: 514-20. 5. Bahrami H, Bluemke DA, Kronmal R, et al. Novel metabolic risk factors for incident heart failure and their relationship with obesity: the MESA (Multi-Ethnic Study of Atherosclerosis) Study. J Am Coll Cardiol 2008;51:1775-83. 6. Suleiman M, Khatib R, Agmon Y, et al. Early inflammation and risk of long-term development of heart failure and mortality in survivors of acute myocardial infarction: predictive role of C-reactive protein. J Am Coll Cardiol 2006;47:962-8. 7. Barron HV, Cannon CP, Murphy SA, et al. Association between white blood cell count, epicardial blood flow, myocardial perfusion, and clinical outcomes in the setting of acute myocardial infarction: a Thrombolysis In Myocardial Infarction 10 Substudy. Circulation 2000;102:2329-34. 8. Dragu R, Huri S, Zuckerman R, et al. Predictive value of white blood cell subtypes for long-term outcome following myocardial infarction. Atherosclerosis 2008;196:405-12. 9. Bekwelem W, Lutsey PL, Loehr LR, et al. White blood cell count, C-reactive protein, and incident heart failure in the Atherosclerosis Risk in Communities (ARIC) study. Ann Epidemiol 2011;21:739-48. 10. Arruda-Olson AM, Reeder GS, Bell MR, et al. Neutrophilia predicts death and heart failure after myocardial infarction: a community-based study. Circ Cardiovasc Qual Outcomes 2009;2:656-62. 11. Goldbourt U, Behar S, Reicher-Reiss H, et al. Rationale and design of a secondary prevention trial of increasing serum HDL cholesterol and reducing triglycerides in patients with clinically manifest atherosclerotic heart disease. Am J Cardiol 1993;71: 909-15. 12. The BIP study group. Secondary prevention by raising HDL cholesterol and reducing triglycerides in patients with coronary artery disease: the Bezafibrate Infarction Prevention (BIP) study. Circulation 2000;102:21-7.
American Heart Journal May 2014
714 Eisen et al
13. Hemker HC, Kop J, Willems GM. Kinetic aspects of the interaction of blood clotting enzymes, VII: the relation between clotting time and clotting velocity. Thromb Haemost 1979;41:309-13. 14. Gooley TA, Leisenring W, Crowley J, et al. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med 1999;18:695-706. 15. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94: 496-500. 16. Kohl M, Heinze G. PSHREG: a SAS macro for proportional and nonproportional subdistribution hazards regression with competing risk data. Tech Rep 2012;1-33. 17. Geskus RB. Cause-specific cumulative incidence estimation and the fine and gray model under both left truncation and right censoring. Biometrics 2011;67:39-49. 18. Harrell FE, Lee KL, Mark DB. Tutorial in biostatistics: multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361-87. 19. Baron RM, Kenney DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations. J Pers Soc Psychol 1986;51(6):1173-82. 20. Jasti S, Dudley WN, Goldwater E. SAS macro for testing statistical mediation in data with binary mediators or outcomes. Nurs Res 2008;57:118-22. 21. Kaptoge S, Di Angelantonio E, Lowe G, et al, for the Emerging Risk Factors Collaboration. C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet 2010;375: 132-40. 22. Danesh J, Whincup P, Walker M, et al. Low grade inflammation and coronary heart disease: prospective study and updated meta-analysis. BMJ 2000;321:199-204. 23. Madjid M, Awan I, Willerson JT, et al. Leukocyte count and coronary heart disease: implications for risk assessment. J Am Coll Cardiol 2004;44:1945-56. 24. Ballantyne CM, Hoogeveen RC, Bang H, et al. Lipoprotein-associated phospholipase A2, high-sensitivity C-reactive protein, and risk for incident coronary heart disease in middle-aged men and women in the Atherosclerosis Risk in Communities (ARIC) Study. Circulation 2004;109:837-42. 25. Wheeler JG, Mussolino ME, Gillum RF, et al. Associations between differential leukocyte count and incident coronary heart disease: 1764 incident cases from seven prospective studies of 30,374 individuals. Eur Heart J 2004;25:1287-92. 26. Folsom AR, Aleksic N, Catellier D, et al. C-reactive protein and incident coronary heart disease in the Atherosclerosis
27.
28.
29.
30. 31.
32.
33. 34.
35. 36.
37.
38.
39.
40.
41.
Risk in Communities (ARIC) study. Am Heart J 2002;144: 233-8. Devaux B, Scholz D, Hirche A, et al. Upregulation of cell adhesion molecules and the presence of low grade inflammation in human chronic heart failure. Eur Heart J 1997;18:470-9. Fildes JE, Shaw SM, Yonan N, et al. The immune system and chronic heart failure: is the heart in control? J Am Coll Cardiol 2009;53:1013-20. Rauchhaus M, Doehner W, Francis DP, et al. Plasma cytokine parameters and mortality in patients with heart failure. Circulation 2000;102:3060-7. Kaneko K, Kanda T, Yamauchi Y, et al. C-reactive protein in dilated cardiomyopathy. Cardiology 1999;91:215-7. Deswal A, Petersen NJ, Feldman AM, et al. Cytokines and cytokine receptors in advanced heart failure: an analysis of the cytokine database from the Vesnarinone Trial (VEST). Circulation 2001;103:2055-9. Mann DL, Young JB. Basic mechanisms in congestive heart failure: recognizing the role of proinflammatory cytokines. Chest 1994;105:897-904. Yndestad A, Damås JK, Øie E, et al. Systemic inflammation in heart failure: the whys and wherefores. Heart Fail Rev 2006;11:83-92. Tracchi I, Ghigliotti G, Mura M, et al. Increased neutrophil lifespan in patients with congestive heart failure. Eur J Heart Fail 2009;11:378-85. Peterson JW, Felker GM. Inflammatory biomarkers in heart failure. CHF 2006;12:324-8. Leone AM, Giannico MB, Bruno I, et al. Safety and efficacy of G-CSF in patients with ischemic heart failure: the CORNER (Cell Option for Recovery in the Non-Eligible patients for Revascularization) study. Int J Cardiol 2011;150:75-8. Verma S, Li SH, Badiwala MV, et al. Endothelin antagonism and interleukin-6 inhibition attenuate the proatherogenic effects of C-reactive protein. Circulation 2002;105:1890-6. Gottdiener JS, Arnold AM, Aurigemma GP, et al. Predictors of congestive heart failure in the elderly: the Cardiovascular Health Study. J Am Coll Cardiol 2000;35:1628-37. Loehr LR, Rosamond WD, Chang PP, et al. Heart failure incidence and survival (from the Atherosclerosis Risk in Communities Study). Am J Cardiol 2008;101:1016-22. Wilhelmsen L, Eriksson H, Svärdsudd K, et al. Improving the detection and diagnosis of congestive heart failure. Eur Heart J 1989;10(Suppl C):13-8. Morrow DA, de Lemos JA, Blazing MA, et al. Prognostic value of serial B-type natriuretic peptide testing during follow-up of patients with unstable coronary artery disease. JAMA 2005;14 (294):2866-71.
Background Heart failure (HF) carries poor prognosis in coronary artery disease (CAD) patients despite advances in therapy. Inflammation predicts recurrent cardiovascular events in CAD patients. It is unknown whether increased levels of inflammatory markers are associated with incident HF in these patients. Aim The aims of this study were to evaluate the association between inflammatory markers and future HF risk in patients with stable CAD and to explore possible mediation by myocardial infarction (MI). Methods The study comprised 2,945 patients with stable CAD without HF at baseline during a median follow-up of 7.9 years. Inflammatory baseline markers were the basis of this study. Results Heart failure was diagnosed in 508 patients (17.2%). Patients who developed HF were older and had more often previous MI, diabetes, hypertension, and peripheral vascular disease. Baseline levels of C-reactive protein (CRP), fibrinogen, and white blood cells (WBCs) were significantly higher in patients who developed HF compared with those who did not. Age-adjusted incident HF rates were related to elevated baseline inflammatory markers in a dose-response manner. Adjusting for multiple confounders, the HF hazard ratios were 1.38 (95% CI 1.11-1.72), 1.33 (95% CI 1.071.66), and 1.36 (95% CI 1.10-1.68) for the third tertiles of CRP, fibrinogen, and WBC levels, respectively. Hazard ratio for the fifth quintile of a combined “inflammation score” was 1.83 (95% CI 1.40-2.39). Mediation by MI preceding the HF onset during follow-up accounted for 10.4%, 10.8%, and 8.6% of the association of subsequent HF with CRP, fibrinogen, and WBC, respectively. Conclusions Increased levels of CRP, fibrinogen, and WBC are independently related to the incidence of HF in patients with stable CAD. (Am Heart J 2014;167:707-14.)
Heart failure (HF) is a major health problem with increasing incidence. 1 Traditional risk factors for HF include chronic coronary artery disease (CAD), myocardial infarction (MI), hypertension (HTN), diabetes mellitus, older age, and others. 1
From the aCardiology Department, Rabin Medical Center, Petah-Tikva, Israel, bDepartment of Epidemiology and Preventive Medicine, School of Public Health, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel, cGertner Institute for Epidemiology and Health Research Policy, Sheba Medical Center, Tel-Hashomer, Israel, and dThe Israel Society for the Prevention of Heart Attacks, Neufeld Cardiac Research Institute, Sheba Medical Center, Tel-Hashomer, Israel. e
A.E. and M.B. contributed equally to the study and are co-first authors of this manuscript. Deceased. Submitted July 6, 2013; accepted January 4, 2014. Reprint requests: Alon Eisen, MD, Cardiology Department, Rabin Medical Center, Jabotinsky St, 49100 Petah, Tikva, Israel. E-mail: [email protected] 0002-8703/$ - see front matter †
© 2014, Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.ahj.2014.01.008
Increased levels of various inflammatory markers predict increased risk of HF hospitalizations and poor outcome in healthy persons or in patients with preexisting HF. 2-5 Several studies reported an association between increased inflammatory marker levels and incidence of HF in patients with CAD, mainly in patients who underwent an acute event such as STelevation MI. 6-10 Data of such an association among patients with stable CAD are limited. In addition, most studies examined a single inflammatory marker, for example, white blood cell (WBC) count or C-reactive protein (CRP). It is also unclear to what extent inflammatory markers are directly associated with the development of HF or, rather, related to the extent of CAD and interim occurrence of MI, which subsequently leads to HF. The aims of the present study were to assess the association between levels of inflammatory markers and future incidence of HF in patients with stable CAD and no
American Heart Journal May 2014
708 Eisen et al
HF at baseline and to explore the role of MI as a possible mediator of such an association.
Hypertension was defined based on the study physician's report or blood pressure N140/90 mm Hg when measured at the time of screening. Diabetes was defined based on reports or treatment with a hypoglycemic drug.
Methods Patients
Laboratory and analytical methods
A total of 3,122 patients (2,854 men, 91%) participated in the Bezafibrate Infarction Prevention (BIP) study, a placebocontrolled randomized trial that assessed the efficacy of bezafibrate in reducing the risk of cardiac events and mortality. 11,12 Inclusion criteria for men and women comprised the following: age of 45 to 74 years, history of MI ≥6 months but b5 years before enrollment into the study and/or stable angina pectoris confirmed by coronary angiography, and/or radio-nuclear studies or standard exercise tests. In addition, a lipid profile of serum total cholesterol between 180 and 250 mg/dL, low-density lipoprotein cholesterol ≤180 mg/dL (≤160 mg/dL for patients b50 years), high-density lipoprotein cholesterol (HDL-C) ≤45 mg/dL, and triglycerides ≤300 mg/dL was required. Exclusion criteria included clinically significant HF (New York Heart Association classes III-IV) during the 3 months preceding the first screening visit, recent acute coronary syndrome (3 months for unstable angina pectoris and 6 months post-MI), idiopathic cardiomyopathy, and life-threatening arrhythmias within 3 months of screening. Institutional ethics committees in each of the participating centers and the central national ethics committee approved the study. Follow-up included an interview and a physical examination by a study cardiologist every 4 months for the study duration (5-7 years) and an additional 2 years of poststudy termination follow-up. Median follow-up length was 7.9 years (interquartile range 7.1-8.7). Patients with baseline HF (n = 171) or missing data on all inflammatory markers (n = 6) were excluded from the current analysis.
Blood samples for measurement of serum lipids, blood chemistry, complete blood count, and other laboratory tests were collected at randomization (baseline). All laboratory analyses were performed in a single central laboratory using standard automated procedures with commercial kits. Accuracy and precision for lipid measurements were under periodic surveillance by the Centers for Disease Control and Prevention/ National Heart, Lung, and Blood Institute's Lipids Standardization Program and by the Wellcome-Murex Diagnostic Clinical Chemistry Quality Assessment Program. Inflammatory markers at baseline (fibrinogen, CRP, and WBC) are the basis of this study. Fibrinogen was determined in fresh citrated plasma using a kinetic method, as described by Hemker et al. 13 High sensitive CRP concentrations were measured in samples of citrated plasma stored at −70°C obtained at baseline on an IMMULITE 2000 analyzer (Diagnostics Products Corporation, Los Angeles, CA) with the manufacturer's reagent solidphase, chemiluminescent immunometric assay. The validity of using plasma citrate compared with serum was tested by analysis of samples from 30 randomly selected individuals. C-reactive protein levels in plasma citrate were 68% of the serum levels (coefficient of variation 3.0, range 65%-74%). The correlation between plasma and serum levels was high (r = 0.998). Values of inflammation marker tertiles were determined for each gender separately: For men, CRP b2.3 (2.3-5.1), N5.1 mg/ dL; fibrinogen b310.3 (310.3-366.8), N366.8 mg/dL; WBC b5.9 (5.9-7.3), N7.3 × 10 3/μL. For women, CRP b2.9 (2.9-6.6), N6.6 mg/dL; fibrinogen b342.2 (342.2- 402.4), N402.4 mg/dL; WBC b5.9 (5.9-7.2), N7.2 × 10 3 /μL. C-reactive protein, fibrinogen, and WBC values were missing for 121, 96, and 151 patients, respectively.
Variable definition and classification Heart failure incidents were ascertained in 2 steps. First, HF diagnosis was based on diagnoses made on each study visit by the study cardiologist in each participating center. This was based on discharge diagnoses from hospitalization records as well as on the presence of HF symptoms, physical findings, and other supporting evidence such as congestion on chest x-ray. Second, all the cases were validated by reviewing the original study records by one of the authors (M.H.) and verifying that the diagnosis was based on documented hospital records, symptoms, or imaging. Functional capacity was classified according to the New York Heart Association classification. Metabolic syndrome (MS) was defined based on the Adult Treatment Panel III report classification as at least 3 of the following: HDL-C b40 mg/dL (1.04 mmol/L), triglycerides N150 mg/dL (1.69 mmol/L), blood pressure N135/85 mm Hg, glucose level N110 mg/dL (6.11 mmol/L), and replacing waist circumference criteria, which was not available, with body mass index (BMI) N28 kg/m 2. Glomerular filtration rate (GFR) was estimated using the formula derived by the Modification of Diet in Renal Disease study: 186 [serum creatinine (mg/dL) − 1.154] × [age (y) − 0.203]. Renal failure was defined as GFR b60 mL/min per 1.73 m 2. Baseline blood pressure is the mean of 2 preinclusion measurements.
Statistical analysis Data were analyzed using the SAS software (version 9.2; SAS, Cary, NC). Characteristics of patients are presented as frequencies or mean ± SD unless otherwise specified and compared by χ 2 tests for categorical variables or analysis of variance for normally distributed continuous variables. Triglycerides, fibrinogen, and CRP, which were not normally distributed, are presented as geometric mean (95% CI) and compared by the nonparametric Kruskal-Wallis test. Trends in proportions by tertiles of inflammatory markers were assessed by the Mantel-Haenszel χ 2 test. Direct adjustment using the entire group included in the analysis as the reference group was used for computation of age-adjusted mortality rates per 1,000 person-years (PY). To account for the competing risk of premature death that may preclude HF onset, HF incidence probabilities over followup time according to inflammation markers were estimated by the cumulative incidence function using the method suggested by Gooley et al. 14 Principal component analysis with varimax (orthogonal) rotation was used to create a single score for the 3 inflammatory markers. A proportional subdistribution hazards model 15 was fitted to the data using a SAS macro developed by Kohl and Heinze. 16 The macro first modifies the input data to
American Heart Journal Volume 167, Number 5
account for competing risks and then applies SAS standard Cox regression procedure, PROC PHREG, using weights and counting-process style of specifying survival times to the modified data set as suggested by Geskus. 17 A fixed set of variables including variables reported as possible confounders or mediators or identified in univariate analysis in the current study (age, history of MI, angina pectoris, peripheral vascular disease, diabetes, and MS) were included in all models. The predictive discrimination ability of each model was evaluated using a C-statistic 18 corresponding to the area under a receiver operating characteristics curve. The C-statistic for the multivariable models for each inflammatory marker was around 0.7 for all models. The validity of the proportional hazard assumption for each of the inflammatory markers was tested by running a Cox proportional hazards model including time-dependent explanatory variables for each inflammatory marker to test the assumption of no time-dependent effect. No significant deviation from the proportional hazard assumption was detected. The extent to which the association between inflammation and HF onset was mediated by MI was explored according to the approach suggested by Baron and Kenny. 19 Mediation models were developed using a SAS macro developed by Jasti et al. 20 The Sobel test was used to assess the significance of mediation. No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents.
Results Of 2,945 patients without HF at baseline, 508 patients (17%) developed HF over a median period of 7.9 years. Patients who developed HF during follow-up were older and had a higher prevalence of previous MI, angina pectoris, HTN (and correspondently, higher systolic and diastolic baseline blood pressure), diabetes, and MS compared with patients without HF (Table I). Patients who later developed HF were treated more frequently by nitrates, diuretics, digitalis, and angiotensin-convertingenzyme inhibitors at baseline (Table I). There was no difference in the rate of HF in patients allocated to the bezafibrate or placebo study arms. Baseline mean levels of inflammatory markers in patients with HF were higher than among those who remained free of HF during follow-up (CRP 4.2 mg/dL [95% CI 3.8-4.5] vs 3.4 mg/dL [95% CI 3.3-3.5], fibrinogen 353.0 mg/dL [95% CI 347.1-359.0] vs 339.8 mg/dL [95% CI 337.0-342.7], WBC 7.1 × 10 3/μL ± 1.8 vs 6.7 × 10 3/μL ± 1.9 [P b .001 for all]). The incidence of MI during follow-up was highest among patients in the third tertile of baseline CRP, fibrinogen, and WBC levels. Data regarding coronary revascularization and MI preceding HF are depicted in Table II. The age-adjusted HF incidence rate (per 1,000 PY) increased with increasing tertiles of the various
Eisen et al 709
Table I. Baseline characteristics of patients with and without HF during follow-up HF during follow-up Characteristics Age (mean ± SD) Men, n (%) History, n (%) MI Angina pectoris PVD Stroke HTN Diabetes MS BMI, mean ± SD Obesity (BMI N30) GFR ≤60 mL/min per 1.73 m2 Lipid profile Total cholesterol, mg/dL HDL-C, mg/dL LDL-C, mg/dL Past smoking Smoking Medical treatment β-Blockers Antidiabetic Nitrates Antiplatelets Digitalis Diuretics ACE-I Blood pressure, mm Hg (mean ± SD) Systolic Diastolic Pulse pressure
No (n = 2,437) Yes (n = 508)
P
59.8 ± 6.8 2238 (91.7)
61.5 ± 6.5 461 (90.2)
b.0001 NS
1844 (75.7) 1327 (54.5) 66 (2.7) 25 (1.0) 1007 (41.3) 215 (8.8) 1086 (44.9) 26.6 ± 3.2 298 (12.2) 432 (17.8)
424 (83.8) 335 (65.9) 29 (5.8) 10 (2.0) 271 (53.3) 71 (14.0) 287 (56.9) 27.3 ± 3.5 108 (21.3) 106 (20.9)
b.0001 b.0001 .0005 .07 b.0001 .0004 b.0001 b.0001 b.0001 NS
212.4 ± 17.5 34.7 ± 5.5 148.9 ± 16.5 1419 (58.2) 289 (11.9)
211.4 ± 17.9 34.0 ± 5.6 147.6 ± 16.5 297 (58.5) 62 (12.2)
NS .006 NS NS NS
971 (39.8) 102 (4.2) 1160 (47.6) 1741 (71.4) 46 (1.9) 219 (9.0) 241 (9.9)
191 (37.6) 41 (8.0) 299 (58.9) 508 (67.3) 27 (5.3) 126 (24.8) 91 (17.9)
NS .0002 b.0001 .06 b.0001 b.0001 b.0001
132.7 ± 15.6 80.9 ± 7.9 51.8 ± 11.6
137.2 ± 16.9 82.3 ± 8.2 54.9 ± 12.9
b.0001 .0003 b.0001
Abbreviations: NS, Not significant; PVD, peripheral vascular disease; LDL-C, lowdensity lipoprotein cholesterol; ACE-I, angiotensin-converting enzyme inhibitor.
inflammatory markers (Figure 1). The cumulative incidence of HF according to tertiles of inflammatory markers is depicted in Figure 2. Age-adjusted hazard ratios (HRs) of HF incidence are presented in Table III. For all inflammatory markers, elevated levels were associated with increased risk of developing HF during the follow-up period. Hazard ratios for developing HF in the third tertile as compared with the first tertile ranged between 1.41 (for fibrinogen) and 1.61 (for CRP) (Table III). Adjusting for additional possible confounders or mediators attenuated the HRs; still, the third tertile of all inflammatory markers was associated with HF (Figure 3). Similar results were obtained after adjusting for diuretic or antiplatelet treatment (data not shown). Patients were divided to quintiles according to their “inflammation score,” which was composed of CRP, WBC, and fibrinogen (Figure 4). Patients in the fifth quintile had a remarkably significantly higher incidence
American Heart Journal May 2014
Tertile values: for men, CRP b2.3 (2.3-5.1), N5.1 mg/dL; fibrinogen b310.3 (310.3-366.8), N366.8 mg/dL; WBC b5.9 (5.9-7.3), N7.3 × 103/μL; for women, CRP b2.9 (2.9-6.6), N6.6 mg/dL; fibrinogen b342.2 (342.2-402.4), N402.4 mg/dL; WBC b5.9 (5.9-7.2), N7.2 × 103/μL. Abbreviations: PTCA, Percutaneous transluminal coronary angioplasty; CABG, coronary artery bypass graft.
.71 .62 .14 108 (11.4) 161 (16.8) 146 (15.2) 99 (10.7) 163 (17.6) 105 (11.4) 101 (10.5) 159 (16.5) 159 (16.5) 118 (12.6) 142 (15.2) 123 (13.2) 97 (10.3) 162 (17.2) 110 (11.7)
97 (10.2) 175 (18.5) 151 (15.9)
.97 .45 .007
97 (10.2) 154 (16.1) 119 (12.5)
122 (13.1) 166 (17.8) 111 (11.9)
.81 .81 .009
104 (11.4) 145 (15.9) 122 (13.4)
.0008 775 (80.8) 685 (75.3) 685 (74.3) 758 (79.0) 712 (76.3)
Previous MI, n (%) Revascularization during follow-up, n (%) PTCA CABG MI during follow-up, n (%)
722 (76.6)
747 (79.0)
.23
733 (76.8)
711 (76.3)
.27
2 1 P 3 2 1 1
2
3
P
Fibrinogen (mg/dL) tertiles CRP (mg/dL) tertiles
Table II. Coronary revascularization and MI preceding HF onset among study patients according to inflammatory markers tertiles
WBC (×10 3/μL) tertiles
3
P
710 Eisen et al
Figure 1
Age-adjusted HF rates (per 1,000 PY) by baseline tertiles of inflammatory markers.
of HF as compared with the first quintile (36.9/1,000 PY vs 14.3/1,000 PY). Cumulative incidence curve for HF during follow-up according to quintiles of inflammation score is depicted in Figure 5. After multivariable analysis, the HR for HF in the fifth quintile as compared with the first quintile according to inflammatory score was 1.83 (95% CI 1.40-2.39). To assess the possibility of interaction between MI and subsequent onset of HF during follow-up, we repeated the analyses separately for patients with and without MI preceding HF onset. In patients with MI (n = 322), 92 patients (28.6%) developed HF during follow-up. The multivariate adjusted HRs for HF according to baseline inflammatory marker tertiles were 1.22 (95% CI 0.702.12) and 1.37 (95% CI 0.81-2.31) for the second and third tertiles of CRP, respectively; 1.00 (95% CI 0.581.78) and 1.08 (95% CI 0.66-1.76) for the second and third tertiles of fibrinogen, respectively; and 1.06 (95% CI 0.63-1.76) for both the second and third tertiles of WBC levels. Among patients without MI during followup (n = 2,623), 416 patients (15.6%) were diagnosed with HF. In these patients, the multivariate adjusted HRs for HF were 1.14 (95% CI 0.89-1.47) and 1.39 (95% CI 1.09-1.76) for the second and third tertiles of CRP, respectively; 1.30 (95% CI 1.01-1.67) and 1.39 (95% CI 1.08-1.78) for the second and third tertiles of fibrinogen, respectively; and 1.02 (95% CI 0.80-1.32) and 1.40 (95% CI 1.11-1.77) for the second and third tertiles of WBC levels, respectively. To further examine the role of MI as a mediator between an inflammatory marker and HF, mediation analysis was performed. Mediation by MI before HF onset (before or during follow-up) accounted for 10.4% of the association between CRP and HF (P = .004). The degree
American Heart Journal Volume 167, Number 5
Tertile values: for men, CRP b2.3 (2.3-5.1), N5.1 mg/dL; fibrinogen b310.3 (310.3-366.8), N366.8 mg/dL; WBC b5.9 (5.9-7.3), N7.3 × 103/μL; for women, CRP b2.9 (2.9-6.6), N6.6 mg/dL; fibrinogen b342.2 (342.2-402.4), N402.4 mg/dL; WBC b5.9 (5.9-7.2), N7.2 × 103/μL.
910 146 (16.0) 1.07 (0.86-1.34) 924 132 (14.3) 1.0 961 193 (20.1) 1.41 (1.14-1.75) 933 169 (18.1) 1.27 (1.02-1.59) 955 130 (13.6) 1.0 947 201 (21.2) 1.61 (1.30-1.99)
3 2
934 161 (17.2) 1.24 (0.99-1.55) 943 130 (13.8) 1.0
The main finding of our study is that elevated levels of baseline inflammatory markers were associated with increased risk of developing HF among patients with stable CAD. This held true after adjusting for multiple confounders associated with HF and inflammation. By using mediation analysis, we found that the association between the inflammatory markers and HF was only partially explained by the occurrence of MI before onset of HF. Inflammation is associated with the development of CAD in different populations. 21-26 There is also growing evidence that supports the importance of inflammation in the development as well as progression of HF, described as the “cytokine hypothesis.” 27-35 In recent years,
1
Discussion
CRP (mg/dL) tertiles
of mediation observed for WBC and fibrinogen was of similar magnitude (8.6% and 10.8%, respectively).
Table III. Heart failure by tertiles of CRP, fibrinogen, and WBC count
Cumulative incidence of HF according to tertiles of CRP (A), fibrinogen (B), and WBC (C).
No. of patients HF incidence (%) Age-adjusted HR (95% CI)
2 2 1
Fibrinogen (mg/dL) tertiles
3
1
WBC (×10 3/μL) tertiles
3
Figure 2
960 201 (21.0) 1.48 (1.21-1.83)
Eisen et al 711
American Heart Journal May 2014
712 Eisen et al
Figure 3
Figure 5
Cumulative incidence of HF according to quintiles of inflammation score. Multivariate adjusted HF HR by inflammatory marker levels (adjusted for age, angina pectoris, diabetes, HTN, previous MI, and peripheral vascular disease).
Figure 4
Age-adjusted rate per 1,000 PY of HF according to inflammation score quintiles.
accumulating data support the association between elevated inflammatory biomarkers and incidence of HF. 2-10 In patients with HF, circulating inflammatory markers such as tumor necrosis factor α, interleukin 6, and CRP were shown to have prognostic value and, therefore, investigated as possible targets for HF therapy. 36,37 Elevated levels of CRP were reported to be associated with increased risk of developing HF in the general population, 4,9 patients with atherosclerotic risk factors, 9 obese patients 5, and the elderly. 38 Elevated WBC counts have also been linked to subsequent HF, in particular, in patients presenting with ST-elevation MI as well as in the general population. 7-10 Our study is the first to our knowledge that investigated the association between inflammation and future HF in a large cohort of stable CAD patients without HF at baseline, a population at risk for developing HF. We
have demonstrated that, regardless of traditional risk factors, elevated levels of CRP, fibrinogen, and WBC count were all associated in a dose-response manner with the risk of HF. Because MI is the leading cause of HF in CAD patients, we asked ourselves whether MI interacts with inflammation in the process of developing HF. Among patients without MI during follow-up, CRP, fibrinogen, and elevated WBC were all associated with future risk of HF. These findings suggest that inflammation may have a role in HF development at least partly independent of recurrent cardiovascular events such as MI. The mediation analysis provided support to our hypothesis, showing that only a modest portion of the association between inflammation and HF is mediated by MI. Several mechanisms could explain this association: an inflammatory state is associated with weight loss, anemia, endothelial dysfunction, and a prooxidative stress 35; CRP may promote HF by activating the complement system, stimulating cytokine production, and causing myocyte loss 32; and finally, elevated WBC count may promote abnormal leukocyte aggregation, vessel obstruction, endothelial injury, and decreased perfusion in heart muscle. 35 Therefore, although MI is the leading cause of HF in CAD patients, it seems that the development of HF involves multiple mechanisms. 39,40 Most studies that examined the association between inflammation and future risk of HF used a single inflammation marker, mainly CRP or WBC level. The correlation between inflammatory markers hampers any attempt to evaluate their joint effect in multivariable analysis. In this study, we were able to study such an effect by creating a score accounting for all 3 markers, using principal component analysis. Using this score, we found that the extent and “burden” of inflammation were related to the incidence of HF with the fifth score quintile carrying 1.8 times the risk of the first quintile. This finding, which was consistent over the follow-up years, suggests a quantitative as well as a qualitative
American Heart Journal Volume 167, Number 5
association between inflammation and HF in patients with CAD. In our cohort, patients who developed HF during follow-up had higher levels of inflammatory markers at baseline with the highest HF incidence in patients who had highest values of baseline inflammatory markers. The clinical implication of this finding may be of importance in identifying stable CAD patients who are at greater risk for developing future HF. These patients might need a more aggressive risk stratification and management. Nevertheless, it is unclear whether inflammation per se is contributing to the development of HF or whether the elevated inflammatory markers reflect another underlying mechanism. Further research is necessary to explore the possible efficacy of treatment of CAD patients by direct antiinflammatory interventions to delay HF development, if indeed inflammation has a direct role. Some limitations apply to our analysis. First, this study is retrospective, and its observational nature does not allow determining if the relationship between inflammation and subsequent HF onset is causal. Second, echocardiography studies were not available and, therefore, were not included in this study. Heart failure was not a prespecified primary end point in the BIP study, and therefore, HF diagnoses were largely based on the clinical judgment of the study cardiologist in each center. Nevertheless, all HF cases were validated by reviewing the patient files and verifying the diagnosis. In addition, although HF hospitalizations are considered to be the most well-delineated HF outcome measure, 41 we also included patients diagnosed with HF who were not hospitalized. We had no data on the WBC differential count and, therefore, could not explore whether granulocyte count, which was examined in prior studies, is a stronger predictor for HF as compared with other WBC components. Several baseline characteristics differ between patients who developed HF versus those who did not. Still, the association persisted (albeit attenuated) after adjustment for these differences. Data on inflammatory diseases and antiinflammatory drug use during study period were lacking. Inflammatory markers used in the study are imperfect and subject to fluctuation. Measurement of other more specific biomarkers may have improved the scientific validity of the study. Furthermore, baseline evaluation does not necessarily reflect long-term inflammatory status. The strengths of data including the large cohort and long follow-up allowed us to provide robust evidence of an association between inflammation and risk of HF in stable CAD patients.
Conclusion Increased levels of CRP, fibrinogen, and WBC were independently associated with the incidence of HF in patients with stable CAD. Further studies are required to
Eisen et al 713
determine whether inflammatory markers may be helpful in prediction and prevention of HF in this population.
Disclosures All authors are affiliated to Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel. All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. Relationship with industry and financial disclosure— none to disclose.
References 1. Ho KK, Anderson KM, Kannel WB, et al. Survival after the onset of congestive heart failure in Framingham Heart Study subjects. Circulation 1993;88:107-15. 2. Vasan RS, Sullivan LM, Roubenoff R, et al. Inflammatory markers and risk of heart failure in elderly subjects without prior myocardial infarction: the Framingham Heart Study. Circulation 2003;107:1486-91. 3. Cesari M, Penninx BW, Newman AB, et al. Inflammatory markers and onset of cardiovascular events: results from the Health ABC Study. Circulation 2003;108:2317-22. 4. Kardys I, Knetsch AM, Bleumink GS, et al. C-reactive protein and risk of heart failure. The Rotterdam Study. Am Heart J 2006;152: 514-20. 5. Bahrami H, Bluemke DA, Kronmal R, et al. Novel metabolic risk factors for incident heart failure and their relationship with obesity: the MESA (Multi-Ethnic Study of Atherosclerosis) Study. J Am Coll Cardiol 2008;51:1775-83. 6. Suleiman M, Khatib R, Agmon Y, et al. Early inflammation and risk of long-term development of heart failure and mortality in survivors of acute myocardial infarction: predictive role of C-reactive protein. J Am Coll Cardiol 2006;47:962-8. 7. Barron HV, Cannon CP, Murphy SA, et al. Association between white blood cell count, epicardial blood flow, myocardial perfusion, and clinical outcomes in the setting of acute myocardial infarction: a Thrombolysis In Myocardial Infarction 10 Substudy. Circulation 2000;102:2329-34. 8. Dragu R, Huri S, Zuckerman R, et al. Predictive value of white blood cell subtypes for long-term outcome following myocardial infarction. Atherosclerosis 2008;196:405-12. 9. Bekwelem W, Lutsey PL, Loehr LR, et al. White blood cell count, C-reactive protein, and incident heart failure in the Atherosclerosis Risk in Communities (ARIC) study. Ann Epidemiol 2011;21:739-48. 10. Arruda-Olson AM, Reeder GS, Bell MR, et al. Neutrophilia predicts death and heart failure after myocardial infarction: a community-based study. Circ Cardiovasc Qual Outcomes 2009;2:656-62. 11. Goldbourt U, Behar S, Reicher-Reiss H, et al. Rationale and design of a secondary prevention trial of increasing serum HDL cholesterol and reducing triglycerides in patients with clinically manifest atherosclerotic heart disease. Am J Cardiol 1993;71: 909-15. 12. The BIP study group. Secondary prevention by raising HDL cholesterol and reducing triglycerides in patients with coronary artery disease: the Bezafibrate Infarction Prevention (BIP) study. Circulation 2000;102:21-7.
American Heart Journal May 2014
714 Eisen et al
13. Hemker HC, Kop J, Willems GM. Kinetic aspects of the interaction of blood clotting enzymes, VII: the relation between clotting time and clotting velocity. Thromb Haemost 1979;41:309-13. 14. Gooley TA, Leisenring W, Crowley J, et al. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med 1999;18:695-706. 15. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94: 496-500. 16. Kohl M, Heinze G. PSHREG: a SAS macro for proportional and nonproportional subdistribution hazards regression with competing risk data. Tech Rep 2012;1-33. 17. Geskus RB. Cause-specific cumulative incidence estimation and the fine and gray model under both left truncation and right censoring. Biometrics 2011;67:39-49. 18. Harrell FE, Lee KL, Mark DB. Tutorial in biostatistics: multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361-87. 19. Baron RM, Kenney DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations. J Pers Soc Psychol 1986;51(6):1173-82. 20. Jasti S, Dudley WN, Goldwater E. SAS macro for testing statistical mediation in data with binary mediators or outcomes. Nurs Res 2008;57:118-22. 21. Kaptoge S, Di Angelantonio E, Lowe G, et al, for the Emerging Risk Factors Collaboration. C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet 2010;375: 132-40. 22. Danesh J, Whincup P, Walker M, et al. Low grade inflammation and coronary heart disease: prospective study and updated meta-analysis. BMJ 2000;321:199-204. 23. Madjid M, Awan I, Willerson JT, et al. Leukocyte count and coronary heart disease: implications for risk assessment. J Am Coll Cardiol 2004;44:1945-56. 24. Ballantyne CM, Hoogeveen RC, Bang H, et al. Lipoprotein-associated phospholipase A2, high-sensitivity C-reactive protein, and risk for incident coronary heart disease in middle-aged men and women in the Atherosclerosis Risk in Communities (ARIC) Study. Circulation 2004;109:837-42. 25. Wheeler JG, Mussolino ME, Gillum RF, et al. Associations between differential leukocyte count and incident coronary heart disease: 1764 incident cases from seven prospective studies of 30,374 individuals. Eur Heart J 2004;25:1287-92. 26. Folsom AR, Aleksic N, Catellier D, et al. C-reactive protein and incident coronary heart disease in the Atherosclerosis
27.
28.
29.
30. 31.
32.
33. 34.
35. 36.
37.
38.
39.
40.
41.
Risk in Communities (ARIC) study. Am Heart J 2002;144: 233-8. Devaux B, Scholz D, Hirche A, et al. Upregulation of cell adhesion molecules and the presence of low grade inflammation in human chronic heart failure. Eur Heart J 1997;18:470-9. Fildes JE, Shaw SM, Yonan N, et al. The immune system and chronic heart failure: is the heart in control? J Am Coll Cardiol 2009;53:1013-20. Rauchhaus M, Doehner W, Francis DP, et al. Plasma cytokine parameters and mortality in patients with heart failure. Circulation 2000;102:3060-7. Kaneko K, Kanda T, Yamauchi Y, et al. C-reactive protein in dilated cardiomyopathy. Cardiology 1999;91:215-7. Deswal A, Petersen NJ, Feldman AM, et al. Cytokines and cytokine receptors in advanced heart failure: an analysis of the cytokine database from the Vesnarinone Trial (VEST). Circulation 2001;103:2055-9. Mann DL, Young JB. Basic mechanisms in congestive heart failure: recognizing the role of proinflammatory cytokines. Chest 1994;105:897-904. Yndestad A, Damås JK, Øie E, et al. Systemic inflammation in heart failure: the whys and wherefores. Heart Fail Rev 2006;11:83-92. Tracchi I, Ghigliotti G, Mura M, et al. Increased neutrophil lifespan in patients with congestive heart failure. Eur J Heart Fail 2009;11:378-85. Peterson JW, Felker GM. Inflammatory biomarkers in heart failure. CHF 2006;12:324-8. Leone AM, Giannico MB, Bruno I, et al. Safety and efficacy of G-CSF in patients with ischemic heart failure: the CORNER (Cell Option for Recovery in the Non-Eligible patients for Revascularization) study. Int J Cardiol 2011;150:75-8. Verma S, Li SH, Badiwala MV, et al. Endothelin antagonism and interleukin-6 inhibition attenuate the proatherogenic effects of C-reactive protein. Circulation 2002;105:1890-6. Gottdiener JS, Arnold AM, Aurigemma GP, et al. Predictors of congestive heart failure in the elderly: the Cardiovascular Health Study. J Am Coll Cardiol 2000;35:1628-37. Loehr LR, Rosamond WD, Chang PP, et al. Heart failure incidence and survival (from the Atherosclerosis Risk in Communities Study). Am J Cardiol 2008;101:1016-22. Wilhelmsen L, Eriksson H, Svärdsudd K, et al. Improving the detection and diagnosis of congestive heart failure. Eur Heart J 1989;10(Suppl C):13-8. Morrow DA, de Lemos JA, Blazing MA, et al. Prognostic value of serial B-type natriuretic peptide testing during follow-up of patients with unstable coronary artery disease. JAMA 2005;14 (294):2866-71.
