Biomarkers for the Prediction of Incident Coronary Heart Disease in the MESA Cohort
Embargo until
Date
2015-05-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Johns Hopkins University
Abstract
Introduction
Heart disease is the leading cause of death in the United States. Biomarkers have been integral to the advancement of preventive cardiology. They have been used to understand the mechanisms of heart disease, and have gone from merely being used in the diagnosis of disease to being important in predicting risk.
This study examined the association between biomarkers and cardiovascular events, and sought to identify which biomarkers would be best improve predictions of such events. Heterogeneity by sex and race on these associations were also examined.
Methods
In a prospective multi-ethnic study of men and women ages 45 to 84 years old without baseline clinical cardiovascular disease, the associations between eight biomarkers and incident coronary heart disease (CHD) were determined by Cox Regression. Tests for heterogeneity by race and by sex were performed. Models containing traditional risk factors were compared to those containing biomarkers to assess the discriminatory and predictive powers of adding biomarkers to the model.
Results
Among 5914 participants, the incident rate of CHD was 7.0 cases per 1000 person-years. After adjustment for the traditional risk factors of heart disease, three biomarkers were found to have significant associations with incident CHD –mean Agatston calcium score (adjusted hazard ratio = 1.0009 per one Agatston unit increase, 95% C.I.: 1.0006 - 1.001), urinary albumin (adjusted hazard ratio = 1.0025 per 1 mg/dL urinary albumin increase, 95% C.I: 1.0003 - 1.0048), and fibrinogen (adjusted hazard ratio = 1.003 per 1mg/dL fibrinogen increase, 95% C.I.: 1.001 – 1.004). Generally, the addition of biomarkers to traditional risk factor models improved model fit and the predictive power of the models, with the inclusion of mean Agatston calcium score and fibrinogen antigen showing the greatest improvement in model prediction.
Conclusions
In a cohort characterized by ethnic diversity, increasing coronary calcium, fibrinogen, and urinary albumin were found to be significantly associated with incident CHD. Prognostic models for CHD containing biomarkers had better predictive power than models containing only traditional risk factors. Therefore, when developing models for predicting heart disease, researchers may want to consider including biomarkers as they improve the discrimination afforded by current heart disease risk factors.
Description
Keywords
biomarkers, epidemiology, prediction, heart disease