Abstract P197: Non-HDL Cholesterol is the Most Related Lipid Measure to Lifestyle Factors in Japanese Mild Hyperlipidemic Patients
Introduction: MEGA Study is a primary prevention trial for coronary events in Japan, in which patients with mild hyperlipidemia were randomized either to diet-alone or diet-plus-pravastatin group. The study group previously reported that moderate non-HDL-C reduction was significantly associated with reduced coronary risk in the diet-alone group. We thus assessed the hypothesis that, among lipid measures, non-HDL-C most explained bodyweight change and dietary habit (“lifestyle factors”), which in turn seemed to be related to coronary risk reduction.
Methods: We analyzed 5-year follow-up data of the subjects in diet-alone group (n = 3966). To identify the lipid measure that most related to bodyweight change and dietary habit simultaneously, we performed a canonical correlation analysis (CCA) using data averaged over the on-treatment period. CCA procedure finds the linear combination of each variable set (canonical variable) so that a correlation between two canonical variables (canonical correlation) is maximized. We then find the variable most attributing to the relation by assessing weights in a linear combination (canonical coefficients) and correlations between each variable and the canonical variable. We also conducted generalized estimating equation (GEE) analyses for exploring relationship between lipid measures and lifestyle factors using original (not-averaged) data.
Results: Table (top) shows that non-HDL-C reduction was highly attributed to the lifestyle factors (ie, high canonical coefficients of non-HDL-C) and was the most influential measure of lifestyle factors (ie, large correlations between non-HDL-C and canonical variables) throughout the different combinations of lipid measures. GEEs (Table, bottom) affirmed that non-HDL-C reduction was associated with bodyweight loss and well adherence to dietary treatment throughout the follow-up.
Conclusion: We identified non-HDL-C reduction as the most representative lipid measure for lifestyle factors in Japanese mild hyperlipidemia.
|Model 1||Model 2||Model 3||Model 4||Model 5|
|Canonical correlation||0.23 (P <.01)||0.24 (P <.01)||0.23 (P <.01)||0.21 (P <.01)||0.21 (P <.01)|
|Lifestyle factors set|
|Adherence to dietary treatment||−0.24||−0.33||−0.24||−0.32||−0.24||−0.33||−0.32||−0.40||−0.31||−0.39|
|Weight change (%)||0.95||0.97||0.95||0.97||0.95||0.97||0.92||0.95||0.92||0.95|
|Lipid measures set|
|TC change (%)||−0.75||0.53||0.57||0.54||−1.22||0.60|
|LDL-C change (%)||−0.04||0.15||−0.05||0.14||−0.01||0.15||−0.23||0.18||−0.25||0.17|
|HDL-C change (%)||−0.28||−0.55||−0.48||−0.56||−0.46||−0.60|
|Non-HDL-C change (%)||0.59||0.76||1.32||0.75||0.90||0.85||2.09||0.84|
|TG change (%)||0.51||0.80||0.51||0.80||0.53||0.81|
Dietary habits were recorded as “adherence to dietary treatment” data recorded as: 0, not/seldom adhere; 1, almost/completely adhere. Data on lipid and bodyweight during the follow-up were used as change rates from baseline rather than recorded values themselves.
↵* Standardized canonical coefficient
↵† Correlation coefficient to each canonical variable
|Weight change*||Adherence to dietary treatment|
|Reduction (g)||95% CI||Odds ratio||95% CI|
|TC reduction (10%)||296.2||234.9, 357.6||1.34||1.27, 1.41|
|LDL-C reduction (10%)||67.2||−14.5, 149.0||1.08||1.00, 1.18|
|HDL-C increase (10%)||239.9||192.2, 287.7||0.97||0.94, 1.01|
|Non-HDL-C reduction (10%)||324.7||272.9, 376.6||1.25||1.20, 1.30|
|TG reduction (10%)||104.0||90.7, 117.2||1.03||1.02, 1.04|
All models adjusted for baseline characteristics: gender, blood pressure levels, and other lipid measures.
↵* Means (SDs) of weight change (g) among patients with decreasing non-HDL-C (n = 2586, median change rate was -6.4%) and with increasing non-HDL-C (n = 1371, median change rate was 4.5%) were -468.5 (924.8) and 24.1 (1747.9), respectively
- © 2012 by American Heart Association, Inc.