Abstract 3571: 20-Year Prospective Associations between Selected Lipid Parameters in Young Adulthood and Subsequent Measures of Subclinical Atherosclerosis
Background: The role of apolipoproteins in prediction of clinical CVD remains controversial. Data are sparse regarding their longitudinal associations and predictive utility for subclinical atherosclerosisis.
Methods: We included participants in the Coronary Artery Risk Development in Young Adults (CARDIA) Study, a multicenter cohort of adults aged 18 to 30, and balanced in distribution of sex, race and education at enrollment in 1985– 6. Serum lipids and apolipoproteins (LDL-c, HDL-c, Apo B and ApoA1) were measured at baseline (Year 0). Coronary artery calcification (CAC) and carotid intimal-medial thickness (IMT) were measured at Year 20. We used logistic regression (for CAC) and linear regression (for IMT) to estimate odds ratios, beta coefficients, likelihood ratio X2, F values, Akaike information criteria and c-statistics or adjusted R2 for each lipid measure in association with CAC and IMT.
Results: There were 2977 men and women (mean age 25 at baseline). Correlations between cholesterol measures and their corresponding apolipoproteins were high (r 0.55 to 0.80), as expected. Table 1⇓ shows univariable results for the association and predictive utility of each lipid measure with CAC and IMT. In univariable analyses of both CAC and IMT, ApoB levels had the strongest predictive utility of any single measure; the ApoB/A1 and LDL-c/HDL-c ratios had still stronger and similar predictive utility. In multivariable-adjusted analyses for CAC, LDL-c had the strongest predictive utility. In multivariable analyses of IMT, ApoB and the ApoB/A1 ratio were strongest. When the ApoB/A1 ratio and LDL-c/HDL-c ratio were included in the model simultaneously, the ApoB/ApoA1 ratio was a modestly superior predictor for both CAC and IMT.
Conclusions: Apolipoprotein levels measured in young adulthood have significant associations with subclinical atherosclerosis in middle age, but their predictive utility is only marginally better than standard lipid measures.