The role of high-density lipoprotein (HDL) function in cardiovascular disease represents an important emerging concept. The present study investigated whether HDL anti-inflammatory capacity is prospectively associated with first cardiovascular events in the general population.
HDL anti-inflammatory capacity was determined as its ability to suppress TNFα (tumor necrosis factor α)–induced VCAM-1 (vascular cell adhesion molecule-1) mRNA expression in endothelial cells in vitro (results expressed as achieved percent reduction by individual HDL related to the maximum TNFα effect with no HDL present). In a nested case-control design of the PREVEND (Prevention of Renal and Vascular End Stage Disease) study, 369 cases experiencing a first cardiovascular event (combined end point of death from cardiovascular causes, ischemic heart disease, nonfatal myocardial infarction, and coronary revascularization) during a median of 10.5 years of follow-up were identified and individually matched to 369 controls with respect to age, sex, smoking status, and HDL cholesterol. Baseline samples were available in 340 cases and 340 matched controls.
The median follow-up time was not different in cases (IQR, 10.5 [9.9–10.8] years) and in controls (IQR, 10.4 [9.9–10.8] years). Cases had a significantly higher prevalence of hypertension and diabetes, used more lipid-lowering drugs, and had higher hsCRP, total cholesterol, low-density lipoprotein cholesterol, triglyceride, and apoB levels. Cases had a significantly lower cholesterol efflux capacity. HDL cholesterol and apoA1 concentrations were similar between groups as a consequence of the matching procedure. It is notable that the HDL anti-inflammatory capacity was significantly lower in cases, despite virtually no difference in HDL cholesterol levels.
In a univariate conditional logistic regression analysis, the HDL anti-inflammatory capacity showed an inverse association with incident CVD events. When adjusting for BMI, alcohol intake, diabetes status, hypertension, and use of lipid lowering drugs, this association remained essentially unaltered. After further adjustment for total cholesterol, apoA1 and triglyceride levels, and hsCRP, UAE, and eGFR, the inverse association of the anti-inflammatory capacity with incident CVD also did not materially change.
In sensitivity analyses adjustment was performed for individual CVD risk factors, which did not significantly alter the association. Analyses were also repeated for individual components of the combined CVD end point. In crude analyses, the HDL anti-inflammatory capacity was suggested to be inversely associated with incident CVD death, ischemic heart disease, hospitalization for MI, percutaneous transluminal coronary angioplasty, and coronary artery bypass graft. Furthermore, in sensitivity analyses the association between cholesterol efflux capacity and incident CVD events was assessed as a comparison exposure. When combining the cholesterol efflux capacity and the anti-inflammatory capacity in 1 model, both HDL function metrics were significantly and independently associated with incident CVD events in a crude, as well as fully adjusted model.
The association of the HDL anti-inflammatory capacity with CVD was different for males and females and participants with high versus low BMI and high versus low triglyceride levels. These interactions were also present with BMI and triglycerides as continuous variables.
Restricted cubic spline analysis showed that the probability of a CVD event is approximately constant for values of HDL anti-inflammatory capacity ranging from −40% to 10%, while at higher values of the HDL anti-inflammatory capacity, an increase of antiinflammatory capacity is directly proportional to a decrease in risk. When adding anti-inflammatory capacity to the wellestablished Framingham risk score, the model likelihood ratio statistic increases from 10.50 to 20.40. A significant likelihood-ratio test (P=0.002) indicated a statistically significant greater predictive power. With the further addition of cholesterol efflux capacity, the model likelihood-ratio statistic again increased to 32.84, with a significant likelihood-ratio test (P=0.0005). When substituting HDL cholesterol with anti-inflammatory capacity in the Framingham risk score equation, the Akaike information criteria decreases from 936 to 542 and the Bayesian information criteria from 945 to 550, again suggesting an increase in model fit.
The HDL anti-inflammatory capacity was inversely associated with risk of CVD events in a CVD risk-adjusted model, which remained unchanged after further adjustment for HDL cholesterol. On the other hand, plasma HDL cholesterol was not significantly associated with a lower risk of CVD events in a model adjusted for CVD risk factorsFurther adjustment for antiinflammatory capacity changed the OR slightly.
Sherry
Phone: 
E-mail: 
Address: