HDL phospholipid content and cholesterol efflux capacity are reduced in patients with very high HDL-C and coronary disease

Abstract

Objective

Plasma levels of high-density lipoprotein cholesterol (HDL-C) are strongly inversely associated with coronary artery disease (CAD), and high HDL-C is generally associated with reduced risk of CAD. Extremely high HDL-C with CAD is an unusual phenotype, and we hypothesized that the HDL in such individuals may have an altered composition and reduced function when compared to controls with similarly high HDL-C and no CAD.

Approach

55 subjects with very high HDL-C (mean 86 mg/dL) and onset of CAD around age 60 with no known risk factors for CAD (‘cases’) were identified through systematic recruitment. 120 control subjects without CAD, matched for race, gender, and HDL-C level (‘controls’), were identified. In all subjects, HDL composition was analyzed and HDL cholesterol efflux capacity was assessed.

Results

HDL phospholipid composition was significantly lower in cases (92 ± 37 mg/dL) than in controls (109 ± 43 mg/dL, p= 0.0095). HDL cholesterol efflux capacity was significantly lower in cases (1.96 ± 0.39) compared with controls (2.11 ± 0.43, p= 0.04).

Conclusions

In persons with very high HDL-C, reduced HDL phospholipid content and cholesterol efflux capacity is associated with the paradoxical development of CAD.

Introduction

Plasma high-density lipoprotein cholesterol (HDL-C) levels are strongly inversely correlated with the incidence of coronary artery disease (CAD)¹. It has been estimated that for each mg/dL increase in HDL-C, the risk of cardiovascular events is decreased by 2–3%². Consequently, levels of HDL-C are factored into many cardiovascular risk assessments, and HDL has been intensively pursued as a secondary goal for risk reduction after low-density lipoprotein cholesterol (LDL-C) lowering. The belief that levels of HDL-C have a causal relationship to the prevention of CAD has been referred to as ‘the HDL cholesterol hypothesis’³.

There have been recent challenges to the HDL-C hypothesis. Common variations associated with small changes in HDL levels are not associated with protection from coronary disease, in contrast to variants that affect LDL-C and triglycerides^{4, 5}. Recently, several clinical trials using agents that raise HDL-C have failed to show any clinical benefit. In the dal- OUTCOMES trial of the cholesteryl ester transfer protein (CETP) inhibitor dalcetrapib, patients received dalcetrapib in addition to other agents that lower LDL-C. Though a significant elevation in HDL-C levels was noted in patients treated with dalcetrapib, the trial was terminated due to futility of the study⁶. The HPS2- THRIVE trial was designed to assess cardiovascular outcomes in patients treated with extended release (ER)- niacin and laropiprant, an antiflushing agent, in addition to a statin. However, HPS2- THRIVE missed its primary endpoint of reducing the risk of MI, stroke, or coronary revascularizations compared to statin therapy alone⁷. These studies have fueled the debate about a causal role of HDL-C in heart disease, and whether raising HDL-C levels is a viable therapeutic strategy.

HDL has several properties that may offer protection against CAD, including its role in promoting cholesterol efflux and reverse cholesterol transport⁸. Genetic and pharmacological manipulations of HDL that increase reverse cholesterol transport in animal models are generally protective against atherosclerosis⁹. However, HDL-C concentration does not always reflect its functionality. For example, even after controlling for HDL-C the cholesterol efflux capacity of HDL was inversely associated with prevalent carotid and coronary atherosclerosis¹⁰ and with incident cardiovascular events¹¹.

Extremely high HDL-C levels are generally associated with reduced risk of CAD. However, an unusual phenotype is that of very high HDL-C with development of CAD in the absence of traditional risk factors. We hypothesized that these individuals have altered composition and/or reduced function of their HDL that may predispose them to increased risk of CAD. We systematically recruited individuals with very high HDL both with and without CAD and compared the composition and function of HDL. We found that the HDL from high HDL-C subjects with CAD had reduced phospholipid content and reduced cholesterol efflux capacity when compared with the HDL from high HDL-C subjects without CAD.

Material and Methods

Materials and Methods are available in the online- only Data Supplement

Results

Clinical characteristics and plasma lipids and apolipoproteins

The clinical characteristics of the 55 ‘cases’ with high HDL-C and CAD and the 120 matched ‘controls’ with high HDL-C and no CAD are shown in Table 1. Mean age was 64 ± 11 for the cases and 69 ± 12 for the controls with approximately 40% of the subjects being female. The mean age of onset of CAD was approximately 60 in the cases for both men and women, although this was not reliably ascertained in all subjects.

Table 1.

Basic Demographics

	Cases	Controls	P value
N	55	120	-
Age (mean ± SD)	64 ± 11	69 ± 12	0.07
Female (%)	36.4	40	0.65
African American (%)	3.6	6.1	0.95
BMI	24.6 ± 3.5	24.4 ± 3.3	0.81
Mean age of onset of CAD (%reported)	60 (44)	NA	-
% statin users*	80	32

^*64 % of cases and 79 % of controls provided information on statin usage

Plasma lipid and apolipoprotein values for the cases and controls are depicted in Table 2. There was no difference in the mean HDL-C between the cases and controls as they were matched for HDL-C level by study design. Triglyceride levels were also not different between the two groups. LDL-C was significantly lower in the cases as compared to the controls (97 ± 38 mg/dL vs.125 ± 33 mg/dL, p<0.0001), as was apoB (77 ± 21 mg/dL vs. 89 ± 19 mg/dL, p<0.001), and LDL particle number. Total plasma phospholipids were lower in cases as compared with controls (253 ± 55 mg/dL vs. 274 ± 52 mg/dL, p= 0.017). ApoE levels were modestly lower in cases (5.0 ± 2 mg/dL vs. 6.0 ± 2 mg/dL, p= 0.046). No differences were observed in apoA-I, apoA-II, or apoC-III levels between cases and controls. No difference was observed between cases and controls in measurements of the inflammatory marker, GlycA.

Table 2.

Plasma Lipids and Apolipoproteins

	Cases	Controls	P value
Triglycerides	80 ± 34	85 ± 38	0.34
Total Cholesterol	201 ± 47	228 ± 37	0.0003
Total phospholipid	253 ± 55	274 ± 52	0.017
HDL-C	86 ± 21	86 ± 20	0.97
LDL-C	97 ± 38	125 ± 33	0.000016
ApoB	77 ± 21	89 ± 19	0.0007
LDL Particle Number	894 ± 318	998 ± 297	0.048
Lp(a) (min, max)	23* (0, 221)	14* (2, 165)	0.1
ApoA-I	195 ± 42	194 ± 40	0.91
ApoA-II	43 ± 11	40 ± 15	0.40
ApoC-III	15 ± 5	13 ± 5	0.89
ApoE	5 ± 2	6 ± 2	0.046
GlycA	332 ± 65	319 ± 61	0.26

Abbreviations: lipoprotein (a) (Lp(a))

Lipid and apolipoprotein parameters are mg/dl.

^*Data reported as median

HDL lipids and subclasses

A comparison of the HDL lipids and particle subclasses is shown in Table 3. HDL phospholipid (HDL- PL) concentrations were significantly lower in cases as compared to controls (92 ± 37 mg/dL vs. 109 ± 43 mg/dL, p= 0.0095), fully accounting for the difference in total plasma phospholipids. HDL triglycerides were modestly elevated in cases as compared to controls. No differences were observed in total HDL particle number or in the number of large, medium, or small HDL particles.

Table 3.

HDL lipids and subclasses

	Cases	Controls	P value	P value adjusted age, gender	P value adjusted age, gender, BMI
HDL Phospholipid	92 ± 37	109 ± 43	0.0095	-	-
HDL Triglyceride	13 ± 9	11 ± 4	0.049	-	-
Total HDL Particle number	40± 8	39 ± 8	0.46	0.46	0.41
Large HDL Particle number	12 ± 5	13 ± 4	0.87	0.56	0.63
Medium HDL Particle number	11 ± 6	11 ± 7	0.68	0.75	1.00
Small HDL Particle number	17 ± 6	16 ± 7	0.55	0.79	0.41

Lipid and apolipoprotein parameters are mg/dL. HDL particle number is reported in µmol/L

HDL cholesterol efflux capacity, cholesterol esterification rate, and PLTP activity

The capacity of HDL to promote cholesterol efflux from J774 macrophages in the presence and absence of cAMP is shown in Table 4. After adjusting for age, sex, and BMI, total HDL cholesterol efflux capacity was significantly lower in cases compared with controls (p= 0.03). The ratio of cholesterol efflux/HDL-C was also significantly lower in cases (p=0.006). Furthermore, cAMP-inducible cholesterol efflux capacity was significantly lower in cases (p= 0.025). HDL- PL was a significant predictor of total cholesterol efflux capacity (p = 0.009, slope 0.0025, R² 0.06). No differences were observed between cases and controls in cholesterol esterification rate or PLTP activity as shown in Table 4.

Table 4.

Cholesterol efflux capacity, CER, and PLTP

	Cases	Controls	P value	P value adjusted age, gender	P value adjusted age, gender, BMI
Total efflux	1.96 ± 0.39	2.11 ± 0.43	0.040	0.047	0.03
Total efflux/HDL-C	0.023 ± 0.005	0.025 ± 0.006	0.029	0.009	0.006
cAMP- inducible efflux	0.60 ± 0.24	0.71 ± 0.32	0.033	0.40	0.025
Cholesterol esterification rate	153 ± 55	169 ± 56	0.30	0.16	0.17
PLTP activity	0.49 ± 0.18	0.51 ± 0.20	0.46	0.51	0.53

Efflux activity is a unitless measure due to normalization, cholesterol esterification rate is given in nmol esterified/ hr/ mL. PLTP activity is given in nmol/mL/min

Discussion

In this study, we investigated a paradoxical phenotype, extremely high HDL-C associated with CAD. We hypothesized that individuals with this phenotype have altered HDL composition and function that put them at greater risk for CAD in the setting of high HDL-C. We found that individuals with very high HDL-C and CAD have reduced levels of HDL phospholipids and reduced HDL cholesterol efflux capacity. These findings add to the growing body of data linking HDL composition and function to clinical cardiovascular disease as distinct from HDL-C concentrations.

Cholesterol efflux is the first step of the reverse cholesterol pathway that can be assessed ex vivo by a method first developed by Rothblat and colleagues¹². Khera et al. demonstrated that HDL cholesterol efflux capacity was inversely associated with prevalent carotid and coronary atherosclerosis even after adjusting for HDL-C,¹⁰ a finding confirmed by Li et al¹³. Recently, Rohatgi et al showed that HDL efflux capacity was inversely associated with incident cardiovascular events after adjusting for HDL-C¹¹. Similarly, Hafiane, A. et al¹⁴ and Shao B et al¹⁵ have demonstrated impaired cholesterol efflux capacity in acute coronary syndrome. In contrast, Li et al¹³ reported a positive association of cholesterol efflux capacity with incident cardiovascular events in an angiographic cohort. Thus there remain questions about the relationship of efflux capacity to cardiovascular disease.

The HDL from subjects with high HDL-C and CAD was reduced in phospholipid content. An inverse relationship has been noted between HDL- PL and the CAD¹⁶. Furthermore, HDL- PL composition has been positively associated with cholesterol efflux capacity^17–21. In our study, HDL- PL was a significant predictor of total cholesterol efflux capacity. Taken together, the reduction in HDL- PL levels may play a causative role in the reduced cholesterol efflux capacity of HDL. The reduced HDL- PL was not associated with a difference in HDL subfraction particle numbers or size. A more thorough lipidomic analysis of the HDL particles is of interest but outside the scope of this study.

LCAT hydrolyzes HDL phospholipids and could influence HDL-PL content as well as potentially cholesterol efflux capacity. We measured the cholesterol esterification rate as an assay of ‘endogenous’ LCAT that is also influenced by the endogenous lipoproteins, but found no evidence of a difference between cases and controls. Prior studies have demonstrated an inverse association between PLTP activity and cholesterol efflux capacity²². We measured PLTP activity and found no significant differences between the two groups. One limitation inherent in the PLTP activity assay is that it measures transfer between synthetic donor and acceptor particles, rather than between native lipoproteins where lipid and protein composition may influence activity. It is possible that increased phospholipase activity (ie. hepatic and endothelial lipase) may underlie the lower phospholipid content in the HDL of these patients, but measurement of these lipases requires post-heparin plasma which was not available.

A limitation of this study is incomplete data on medical history (such as date of onset of CAD) and lack of prospective data. To minimize this problem, we selected controls that were either the same age or older than the cases. Furthermore, prior studies have demonstrated that CAD can reduce cholesterol efflux capacity^{14, 15}. Another limitation in our study is the assessment of only one of the major functions of HDL. HDL is a heterogeneous particle that has additional antioxidant and anti-inflammatory functions, as well as unknown differences in lipidomic and proteomic compositions between the cases and controls, which when all considered together, may give a more comprehensive picture of HDL functionality.

The significant differences in LDL-C and apoB levels between the cases and controls are attributed to higher use of statins in the cases than in the controls. Of the subjects that provided information on statin usage, 80% of the cases and 32% of controls reported being on statins. Recent studies have shown conflicting data regarding the effect of statins on cholesterol efflux capacity making it difficult to assess the nature of their effect on our study population^{10, 23–26}. However, it is unlikely that statins from the PEG precipitated supernatants would have a substantial effect on the cells over a two hour efflux assay. Recently, Miyamoto-Sasaki et al. determined cholesterol efflux capacity in dyslipidemic patients before and after treatment with pitavastatin²⁷. The statin increased serum HDL-C levels, HDL- PL levels, and enhanced cholesterol efflux capacity, suggesting it is unlikely that statins decreased the HDL- PL content and efflux capacity of the cases. In fact, it is possible that the observed differences may actually have been greater if measurements were taken prior to statin therapy.

In conclusion, individuals with the paradoxical phenotype of very high HDL-C and CAD were found to have reduced HDL phospholipid and HDL cholesterol efflux capacity as compared to controls with very high HDL-C and no CAD.

Significance.

The present study describes characterization of the HDL in individuals with a paradoxical phenotype of very high HDL-C and CAD. We demonstrate that the HDL from these individuals has reduced HDL- PL content and reduced cholesterol efflux capacity compared with individuals with comparably high HDL-C but no CAD. These data add to the growing body of data suggesting that HDL quality, not quantity, is important in influencing risk of CAD.

Abbreviations

LCAT

Lecithin-cholesterol acyltransferase

PLTP

Phospholipid transfer protein

CETP

Cholesterol Ester Transfer Protein

ABCA1

ATP-binding cassette transporter-1

HDL-PL

HDL Phospholipid