Association between inflammatory markers, hemostatic, and lipid factors in postinfarction patients

Abstract

A therosclerosis is increasingly recognized as a complex phenomenon involving the interaction of several mechanisms. Dyslipidemia, thrombosis, and other metabolic dysfunctional syndromes have been implicated in atherogenesis and its secondary complications.1 More recently, inflammation has been proposed to play an important role in the pathophysiology of atherosclerosis and to be an important determinant of plaque vulnerability.2 Although thrombosis, dyslipidemia, and proinflammatory conditions contribute to atherosclerosis and increased risk of cardiac events, the associations among these 3 pathologic states are not well understood. Therefore, the aim of this study was to evaluate the relations among inflammation, thrombosis, and dyslipidemia, by comparing the levels of their respective serum markers measured in a population of patients in a stable phase after myocardial infarction. • • • For the purpose of this analysis, we used data and stored blood samples collected during the Thrombogenic Factors and Recurrent Coronary Events (THROMBO) study. THROMBO was a prospective, multicenter study that evaluated the prognostic significance of lipid and hemostatic factors measured in 1,045 stable patients (791 men and 254 women) 2 months after an index myocardial infarction. Patient eligibility criteria and conclusions of the study have been previously published in detail.3 The hemostatic factor analyses included measurement of D-dimer, factor VII, fibrinogen, von Willebrand factor (vWF), and plasminogen activator inhibitor-1 (PAI-1). The lipid parameters included measurement of cholesterol, apolipoprotein A, apolipoprotein B, lipoprotein(a), triglycerides, and highdensity lipoprotein cholesterol. All markers were assayed using previously published methods as detailed in the original THROMBO study.3 Low-density lipoprotein cholesterol was not measured directly in this study. Analyses were performed according to manufacturers’ specifications, and quality control was within the recommended precision for each test. For this analysis, the stored blood samples were retrieved and sent to Dr. Ridker’s laboratory (Boston, Massachusetts) where they were assayed for the inflammatory markers C-reactive protein (CRP) and serum amyloid A (SAA).4 CRP and SAA were measured in 957 and 907 patients, respectively, of the 1,045 patients enrolled in THROMBO. High-sensitivity assays (Dade Behring, Newark, Delaware) were used for measurement of both markers. There were 88 patients who did not have CRP levels measured and 138 patients who did not have SAA levels measured due to inadequate blood samples or technical difficulties relating to the assay. The subsets of patients who did not have CRP or SAA measured for logistic reasons did not differ significantly in their clinical profile from patients included in the analysis. Statistical analysis for a particular inflammatory marker was based on the subpopulation of patients who had that marker measured of the 1,045 patients enrolled in the THROMBO study. As an initial step, we evaluated the trends in the distribution of the levels of the lipid and hemostatic factors by CRP quartile using the general linear model for continuous variables. Then, using the fourth CRP or SAA quartile (coded as CRP Q4, SAA Q4) as the response variable, we applied univariate and multivariate analyses to determine which of the various lipid and thrombogenic markers measured in THROMBO were associated with elevated plasma CRP or SAA levels. For the multivariate portion of the analysis, we first constructed a basic clinical model using logistic regression (SAS, PHREG computer program, SAS Institute, Cary, North Carolina) to identify factors associated with elevated CRP or SAA levels. This was done by a forward selection process to identify the clinical factors associated with CRP Q4 or SAA Q4 from 13 preselected covariates (age, gender, race, body mass index, prior myocardial infarction, prior stroke, history of diabetes mellitus, history of claudication, index infarct type by electrocardiogram [Q-wave vs non–Qwave], thrombolytic therapy, pulmonary congestion by chest x-ray, ejection fraction during the index coronary event, and smoking status at enrollment). A significance level of p 0.10 was used for entering a variable into the basic clinical model. Each of the lipid From the Cardiology Unit, Department of Medicine and Department of Biostatistics, University of Rochester Medical Center, Rochester, New York; Vascular Medicine Program, Orthopaedic Hospital, UCLA School of Medicine, Los Angeles, California; Departments of Laboratory Medicine and Pathology, Children’s Hospital and Harvard Medical School, Boston, Massachusetts; and Center for Cardiovascular Disease Prevention and Division of Cardiology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts. This study was supported by grants 5R01 HL-48259 and HL-30616 from the National Institutes of Health, Bethesda, Maryland. Dr. Zareba’s address is: Heart Research Follow-Up Program, University of Rochester Medical Center, Box 653, Rochester, New York 14642. E-mail: heartwz@heart.rochester.edu. Manuscript received August 26, 2002; revised manuscript received and accepted January 7, 2003.