Abstract 16612: Left Ventricular Mass and Geometry for Enhancing Risk Prediction of Cardiovascular Disease in Older Adults: The Cardiovascular Health Study (chs)

Abstract

BACKGROUND: Increased left ventricular (LV) mass is associated with higher risk for clinical cardiovascular disease (CVD) events. We sought to determine whether addition of LV mass and geometry to standard risk prediction models improves CVD risk reclassification in older adults. METHODS: We included CHS participants (ppts) who underwent echocardiography (echo) at the baseline examination. We calculated 10-year risk prediction models for coronary heart disease (CHD) and CVD (CHD, heart failure, stroke) using CHS-specific coefficients for Framingham risk factors (RFs). LV mass index (LVMI) was calculated using the Devereux equation and adjusted for age, height, weight, and sex to obtain partition values, based on echo data from healthy CHS ppts. We used these values to define LV geometry as normal, concentric remodeling, and eccentric or concentric LV hypertrophy. We then added echo-LVMI and LV geometry category to baseline risk prediction models and determined the C-statistic and net reclassification improvement (NRI) using < 10%, 10-20%, and ≥ 20% for CHD risk and < 20%, 20-30%, and ≥ 30% for CVD risk. RESULTS: Over 10 years of follow up in 2577 CHS ppts (64% women, 15% black, baseline age 72 years), the multivariable-adjusted hazards ratios for CHD and CVD for a 1-unit increase in echo-LVMI were 1.92 (95% CI: 1.47-2.50) and 1.85 (95% CI: 1.48-2.30), respectively. For CVD, the C-statistic in models with and without echo LVMI were 0.666 and 0.659, respectively (P = 0.03 for difference). The NRI for CHD risk prediction with the addition of echo-LVMI and geometry to a model containing RFs was 0.034 (P = 0.06). As shown in the Table, the NRI for CVD risk prediction with the addition of echo-LVMI and geometry was 0.055 (P = 0.001).

CONCLUSION: In an elderly cohort, increased echo-LVMI was significantly associated with increased risk for CHD and CVD. Addition of echo-LVMI and LV geometry to a prediction model for CVD was found to modestly improve risk classification beyond Framingham RFs.