Abstract MP70: Ten Year Coronary Heart Disease Risk Prediction Using Coronary Artery Calcium and Traditional Risk Factors: Results from the Multi-Ethnic Study of Atherosclerosis (MESA)

• Published in: Circulation (New York, N.Y.) N.Y.), 2014-03, Vol.129 (suppl_1)
• Main Authors: McClelland, Robyn L, Jorgensen, Neal, Bild, Diane, Burke, Greg, Post, Wendy, Shea, Steve, Liu, Kiang, Watson, Karol, Folsom, Aaron, Budoff, Matthew, Carr, J. Jeffrey, Blaha, Michael, Kronmal, Richard
• Format: Article
• Language: English
• ISSN: 0009-7322; 1524-4539
• DOI: 10.1161/circ.129.suppl_1.mp70

Abstract only Objectives: In the modern era existing risk scores are not well calibrated, and may overestimate or underestimate risk in different populations. Several studies have demonstrated the added value of coronary artery calcium (CAC) over and above traditional risk factors for risk prediction. Our goal was to develop a risk score to estimate 10-year CHD risk using CAC and traditional risk factors. Methods: The MESA study is a cohort of 6814 participants aged 45-84 and free of clinical heart disease at baseline (2000-2002) in which CAC (Agatston score) was measured. Incident coronary heart disease (CHD) events included myocardial infarction, CHD death, and angina when accompanied by revascularization. Penalized Cox regression models (“shrinkage” models) were used to perform variable selection and estimate the risk score coefficients. Covariates which were considered for inclusion in the risk score model were: age, gender, race/ethnicity, HDL and total cholesterol, lipid lowering medication use, systolic blood pressure (SBP), anti-hypertensive medication use, body mass index (BMI), current smoking, family history of heart attack, diabetes, and CAC. Pre-specified interactions considered included: age, gender, race/ethnicity and CAC with all other predictors; anti-hypertensive medications-by-SBP; and lipid lowering medications-by- total cholesterol. Bootstrapping was used to establish the internal validity of the model to avoid over-optimism. Results: Participants were followed for a median of 10.2 years, and 393 CHD events were observed. With the exception of BMI and diastolic blood pressure, all the risk factors were included in the risk score. The risk score demonstrates good discrimination and calibration, with survival adapted area under the ROC curve (AUC) of 0.81 and discrimination slope of 0.91. The model exhibits improved performance compared to existing risk scores, even after recalibration of traditional risk scores to the MESA population, which have yield an AUC for Framingham risk score of 0.72, and an AUC for traditional risk factors fit to MESA of 0.76. The improvement reflects both the addition of CAC and a tuning of the parameter estimates for the traditional risk factors. The MESA 10-year CHD risk calculator will be available online to facilitate clinical use. Conclusions: An accurate estimate of 10-year CHD risk can be obtained using traditional cardiovascular risk factors and CAC.