Niacin therapy increases HDL particles and total cholesterol efflux capacity but not ABCA1-specific cholesterol efflux in statin-treated subjects

Abstract

Objective

We investigated relationships between statin and niacin/statin combination therapy and the concentration of high density lipoprotein particles (HDL-P) and cholesterol efflux capacity, two HDL metrics that might better assess cardiovascular disease (CVD) risk than HDL-cholesterol (HDL-C) levels.

Approach

In the Carotid Plaque Composition Study, 126 subjects with a history of CVD were randomized to atorvastatin or combination therapy (atorvastatin/niacin). At baseline and after 1 year of treatment, the concentration of HDL and its three subclasses (small, medium, and large) were quantified by calibrated ion mobility analysis (HDL-P_IMA). We also measured total cholesterol efflux from macrophages and ABCA1-specific cholesterol efflux capacity.

Results

Atorvastatin decreased low density lipoprotein cholesterol (LDL-C) by 39% and raised HDL-C by 11% (P=0.0001), but did not increase HDL-P_IMA or macrophage cholesterol efflux. Combination therapy raised HDL-C by 39% (P<0.0001), but increased HDL-P_IMA by only 14%. Triglyceride levels did not correlate with HDL-P_IMA (P=0.39), in contrast to their strongly negative correlation with HDL-C (P<0.0001). Combination therapy increased macrophage cholesterol efflux capacity (16%, P<0.0001) but not ABCA1-specific efflux. ABCA1-specific cholesterol efflux capacity decreased significantly (P=0.013) in statin-treated subjects, with or without niacin therapy.

Conclusions

Statin therapy increased HDL-C levels but failed to increase HDL-P_IMA. It also reduced ABCA1-specific cholesterol efflux capacity. Adding niacin to statin therapy increased HDL-C and macrophage efflux, but had much less effect on HDL-P_IMA. It also failed to improve ABCA1-specific efflux, a key cholesterol exporter in macrophages. Our observations raise the possibility that niacin might not target the relevant atheroprotective population of HDL.

Introduction

The risk of cardiovascular disease (CVD) associates inversely and robustly with levels of high density lipoprotein-cholesterol (HDL-C).¹ However, recent trials of drugs that elevate HDL-C levels, such as niacin and cholesteryl ester transfer protein (CETP) inhibitors, failed to reduce cardiac events in statin-treated subjects. ^2-4 These observations indicate that pharmacologically induced increases in HDL-C are not necessarily therapeutic. Therefore, metrics based on HDL’s role in cardioprotection are urgently needed.

Recent studies demonstrate that cholesterol efflux capacity, the ability of serum HDL (serum depleted of apolipoprotein B (apoB)-containing lipoproteins) to remove cholesterol from cultured macrophages,^{5, 6} might be a better metric for CVD risk than HDL-C. First, in two large studies, impaired cholesterol efflux capacity was a much stronger predictor of prevalent and incident coronary artery disease status than was HDL-C.^{7, 8} This relationship remained significant after adjusting for HDL-C levels.^{7, 8} It is important to note that this assay measures cholesterol efflux by multiple pathways, including ATP-binding cassette transporter A1 (ABCA1), ATP-binding cassette transporter G1 (ABCG1), scavenger receptor B1 (SR-B1), and aqueous diffusion. Second, cholesterol efflux capacity, quantified with an assay specific for the ABCA1 pathway,⁵ was strongly and inversely associated with future cardiovascular events in a large population free of CVD at baseline.⁹ Moreover, this relationship remained highly significant after correction for other classic lipid risk factors for CVD. Taken together, these observations provide strong evidence that cholesterol efflux capacity is a clinically relevant measure of functional HDL that is independent of HDL-C.

The concentrations of HDL particles (HDL-P) in plasma or serum might be another useful metric for assessing CVD risk, because HDL particles vary widely in size, protein composition,^{10, 11} and efflux capacity with different pathways.^{12, 13} Lipid composition and content also vary greatly, ranging from a few percent to up to 50% of HDL particle mass.¹⁴ Thus, measuring HDL-C creates a bias toward larger, cholesterol rich particles.¹⁵

We recently showed that ion mobility analysis (IMA)¹⁶ can accurately quantify HDL-P when it is calibrated with protein standards. This approach, termed calibrated-IMA, reproducibly detects three major HDL species: small, medium and large.¹⁷ The stoichiometry of apoA-I and the sizes and relative abundances of HDL subspecies determined by calibrated-IMA are in excellent agreement with those determined by orthogonal methods.^{10, 17} Moreover, calibrated-IMA accurately determines the concentration of gold nanoparticles and reconstituted HDL, validating the strength of this approach. We therefore propose a new term, HDL-P_IMA, to represent the quantification of an HDL species whose size has been determined by calibrated-IMA.

The effect of HDL-targeted therapies on total HDL and its subpopulations is important because it may be necessary to increase the total concentration of HDL and/or specific subspecies to provide cardioprotection and improve cholesterol efflux capacity. Three studies reported that niacin therapy had little impact on HDL particle number ^18-20 but the relationship of HDL-P determined by nuclear magnetic resonance spectroscopy (HDL-P_NMR) to the size and concentration of HDL is unclear.¹⁷ In a small clinical study, adding niacin to statin therapy did not improve cholesterol efflux capacity, even though HDL-C levels rose considerably.²¹

In the current study, we examine the impact of statin therapy alone or in combination with niacin on HDL-C, LDL-C, cholesterol efflux capacity, and HDL-P_IMA. Samples from baseline (off therapy) and after 1 year of treatment with atorvastatin or atorvastatin plus niacin were obtained from the Carotid Plaque Composition (CPC) study.²²

Materials and Methods

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

Results

The baseline clinical characteristics of the subjects in the monotherapy (n=46) and combination therapy groups (n=80) are shown in Table 1. None of the subjects had received lipid therapy for at least 1 year before entering the study. The two groups were well randomized, with similar levels of LDL-C and HDL-C. All other baseline characteristics were also similar in the two groups, with no significant differences in the fractions of subjects with a family history of CVD, prior MI, smoking status, or diabetes. Supplemental Table I reports baseline clinical characteristics of HDL-related measurements according to coronary disease phenotype (stable vascular disease or acute coronary syndrome and/or revascularization).

Table 1.

Baseline characteristics of study subjects.

Characteristic	Atorvastatin	Atorvastatin + niacin	P value
n	46	80
Age	54.9 ± 8.5	55.1 ± 8.07	0.89
Gender (% female)	37.0	31.0	0.51
Family history of CVD (%)	50.0	43.8	0.50
History of MI	28.3	37.5	0.29
CVD (%)	87.0	87.5	0.93
Current smoking status (%)	23.9	18.8	0.49
DM (%)	6.5	10.0	0.51
Total cholesterol (mg/dL)	235.5 ± 37.5	243.8 ± 42.86	0.28
Triglycerides (mg/dL)†	168 (102-221)*	151 (114-227.5)*	0.83
LDL-C (mg/dL)	161.6 ± 32.7	167.5 ± 42.44	0.41
HDL-C (mg/dL)	43.3 ± 14.3	41.4 ± 10.78	0.42
ApoB (mg/dL)	130.4 ± 24.0	134.2 ± 24.57	0.41
ApoA-I (mg/dL)	134.1 ± 26.3	130.6 ± 21.96	0.42
C-reactive protein (mg/dL)†	0.21 (0.09-0.37)*	0.17 (0.08-0.44)*	0.89
S-HDL-PIMA (μM)	6.2 ± 2.4	6.4 ± 2.4	0.75
M-HDL-PIMA (μM)	5.9 ± 2.5	6.0 ± 2.7	0.83
L-HDL-PIMA (μM)†	1.46 (0.79-2.97)*	1.35 (0.72-2.37)*	0.40
HDL-PIMA (μM)	14.2 ± 3.6	14.0 ± 3.8	0.79
Macrophage cholesterol efflux (%)	6.7 ± 1.6	6.3 ± 1.4	0.21
ABCA1 cholesterol efflux (%)	12.6 ± 2.1	12.5 ± 2.4	0.77

^*median and interquartile range.

^†Two-sample Wilcoxon rank-sum (Mann-Whitney) test. ±, mean and standard deviation.

Correlations of HDL-C and HDL-P_IMA with other lipid metrics

Table 2 shows the correlation coefficients of HDL-C and HDL-P_IMA with baseline characteristics. HDL-C and HDL-P_IMA correlated positively (r=0.59, P<0.0001), but only 35% of the variation in HDL-P_IMA was explained by HDL-C. ApoA-I levels correlated more strongly with HDL-C (r=0.91, P<0.0001) than with HDL-P_IMA (r=0.62, P<0.0001). As previously reported,¹⁷ HDL-C correlated negatively with S-HDL-P_IMA (r=−0.33, P<0.0001). L-HDL-P_IMA correlated more strongly with HDL-C (r=0.72, P<0.0001) than with HDL-P_IMA (r=0.51, P<0.0001), explaining 50% of the variance in HDL-C but only 26% of the variance in HDL-P_IMA.

Table 2.

Correlations of baseline HDL-C and HDL-P_IMA with lipid variables.

	HDL-C	HDL-PIMA
Triglycerides	−0.44**	−0.08
LDL-C	0.26	0.02
HDL-C	-------	0.59**
ApoB	−0.09	−0.07
ApoA-I	0.91**	0.62**
S-HDL-PIMA	−0.33*	0.36**
M-HDL-PIMA	0.70**	0.77**
L-HDL-PIMA	0.72**	0.51**
HDL-PIMA	0.59**	-------
Macrophage cholesterol efflux	0.59**	0.37**
ABCA1 cholesterol efflux	0.27*	0.32*

^*P<0.05,

^**P<0.0001.

Interestingly, triglycerides correlated negatively with HDL-C (r=−0.44, P<0.0001) but not with HDL-P_IMA (r=−0.08, P=0.39). In contrast, the concentration of individual HDL-P_IMA subclasses did correlate with triglyceride levels (Supplemental Table II). Triglycerides correlated positively with S-HDL-P_IMA (r=0.31, P=0.0003), but negatively with M- and L- HDL-P_IMA (r=−0.25 for both; P=0.006 and 0.005, respectively). At baseline, HDL-C and HDL-P_IMA did not correlate with carotid plaque parameters (plaque lumen or wall volume) or the presence/absence of necrotic core as assessed by MRI (data not shown).

Statin and niacin therapies have markedly different impacts on HDL-C and HDL-P_IMA

The effects of therapy allocation on lipid measurements are shown in Fig. 1. Atorvastatin significantly reduced LDL-C levels by 39% (−65 mg/dL, P<0.0001) and apoB by 35% (−47 mg/dL, P<0.0001). Adding niacin to atorvastatin reduced both LDL-C and apoB levels by an additional 11% (−83 mg/dL and −63 mg/dL, respectively, P<0.0001 for each). Combination therapy was more effective than monotherapy at lowering LDL-C and apoB (P=0.005 and<0.0001, respectively).

Figure 1. Effects of lipid-altering therapies on lipid parameters.

For each therapy, the change from baseline on lipid parameters is provided. The differences in change from baseline between therapies were compared using a linear regression model controlling for baseline values, and the P value for the difference is indicated. A, atorvastatin therapy; A+N, atorvastatin plus niacin therapy.

After 1 year of treatment with atorvastatin, HDL-C levels increased by 11% (P=0.0001). Combination therapy raised HDL-C by a further 18% (P<0.0001). The increase observed for combination therapy was significantly higher than the one seen for monotherapy (P<0.0001). For each treatment, the change in HDL-C depended on coronary disease status (presence of either stable coronary disease or prior acute coronary syndrome event and/or coronary revascularization, P for interaction, 0.017). Despite the increase in HDL-C levels with atorvastatin therapy, there was no change in HDL-P_IMA concentration (P=0.49). In contrast, combination therapy increased HDL-P_IMA levels by 14% (P<0.0001); it was more effective than monotherapy in raising HDL-P_IMA (P<0.0001). However, when controlling for the increase in HDL-C, the change in HDL-P_IMA for combination therapy was not different from the change seen with monotherapy (P=0.09). In contrast, the difference between HDL-C levels on combination therapy and monotherapy remained significant after controlling for the change in HDL-P_IMA concentration (P<0.0001). Coronary disease status (categorized as stable coronary disease or prior acute coronary syndrome event and/or coronary revascularization) did not affect the relationship of therapies to changes in HDL-P_IMA.

ApoA-I levels also increased after atorvastatin (6%, P=0.006) and combination therapy (13%, P<0.0001). As seen for HDL-C, the change in apoA-I levels after 1 year of treatment was higher for the combination therapy than for monotherapy (P=0.004).

Both atorvastatin and niacin alter HDL particle distribution

Although atorvastatin monotherapy failed to affect HDL-P_IMA, it reduced S-HDL-P_IMA by 10% (Fig. 2, P=0.0008). A similar decline was observed in subjects allocated to atorvastatin plus niacin (16%, P<0.0001). The between-group comparison for the change in means was not significant (P=0.31, Fig. 2). There was a non-significant trend toward an increase in M-HDL-P_IMA concentration (P=0.068) in the monotherapy group, while combination therapy increased M-HDL-P_IMA by 32% (P=0.0001). However, no significant differences were found when comparing the changes in means observed for the mono and combination therapies (Fig. 2, P=0.11). Large-HDL-P_IMA concentration was not affected by atorvastatin treatment (P=0.99, skewed distribution, Wilcoxon signed rank test). In contrast, combination therapy increased L-HDL-P_IMA by 103% (P<0.0001, Wilcoxon signed rank test). Thus, compared with atorvastatin, combination therapy associated with a significant increase in L-HDL-P_IMA (Fig. 2, P=0.009, median regression).

Figure 2. HDL particle distribution according to therapy.

S-HDL-P_IMA, M-HDL-P_IMA, and L-HDL-P_IMA were measured by calibrated-IMA. HDL-P_IMA was obtained by summing the different HDL-P_IMA subspecies. For each HDL-P_IMA subspecies, P values comparing on treatment and baseline measurements for the same therapy are provided. P values comparing differences in changes between two therapies are also indicated. The box plots show the distribution of the data (median, interquartile ranges), while the dots represent outliers. A, atorvastatin therapy; A+N, atorvastatin plus niacin therapy.

Lipid-altering therapies affect macrophage and ABCA1 cholesterol efflux capacity differently

Atorvastatin therapy did not affect the cholesterol efflux capacity of serum HDL (serum depleted of apoB-containing lipoproteins) with macrophages (P=0.86). However, the efflux capacity of serum HDL rose by 16% with combination therapy (P<0.0001); the between-group comparison for the change in means was significant (Fig. 3A, P<0.0001). The difference lost significance when we controlled for the change in HDL-C (P=0.13), but it remained significant when we controlled for the change in HDL-P_IMA (P=0.003). Plots showing the change in macrophage cholesterol efflux for each individual, according to treatment group, are shown in Supplemental Figure IA. The change in macrophage cholesterol efflux seen for combination therapy correlated more strongly with the change in HDL-C levels (r=0.60, P<0.0001) than with the change in HDL-P_IMA (r=0.35, P=0.002). Monotherapy with atorvastatin did not affect macrophage efflux with serum HDL, and we observed no correlation of efflux capacity with the increase in HDL-C (r=−0.12, P=0.42) or the variation in HDL-P_IMA (r=0.07, P=0.63) in this group of subjects.

Figure 3. Effect of lipid-altering therapies on cholesterol efflux capacity.

(A) Macrophage cholesterol efflux. (B) ABCA1-specific efflux. (C) ABCA1-specific efflux combining all subjects (n=126). P values comparing on treatment and baseline measurements for the same therapy are provided. P values comparing differences in changes between two therapies are also indicated. The box plots show the distribution of the data (median, interquartile ranges), while the dots represent outliers. A, atorvastatin therapy; A+N, atorvastatin plus niacin therapy.

To further investigate what factors might affect changes in macrophage cholesterol efflux, we used regression models that included clinically relevant covariates. Changes in macrophage efflux were not predicted by baseline characteristics (Supplemental Table III). In a multiple regression model that included changes in lipid metrics, therapy and coronary disease status as predictors, the major determinants of change in macrophage efflux capacity were therapy (if the patient was receiving either atorvastatin or combination therapy, P=0.009) and changes in apoA-I levels (P=0.031).

Although trending to a decrease, ABCA1-specific cholesterol efflux was not significantly affected by either monotherapy (P=0.065) or combination therapy (P=0.086). Plots showing the change in ABCA-specific cholesterol efflux for each individual, according to treatment are showed in Supplemental Figure IB. Because the between-group comparison for the change in means was not significant (Fig. 3B, P=0.78), we also determined whether statin therapy, with or without niacin, had any effect on the efflux capacity of serum HDL with the ABCA1 pathway. This analysis demonstrated a significant negative effect of statin therapy on ABCA1 efflux by serum HDL (Fig. 3C, P=0.013). Because both groups of subjects were on statin therapy, this observation suggests that atorvastatin therapy associates with impaired efflux capacity by the ABCA1 pathway.

We also investigated the determinants of ABCA1-specific efflux. In a multiple regression model using changes in ABCA1-specific efflux as outcome and baseline measurements of lipid metrics, as well as coronary disease status, treatment, age, gender, smoking status and diabetes as predictors, no baseline variable predicted changes in ABCA1-specific cholesterol efflux (Supplemental Table IV). The major predictors of change in ABCA1-specific cholesterol efflux were changes in triglycerides (P=0.005) and changes in apoA-I levels (P=0.015) (Supplemental Table IV). Plots showing the correlation of changes in ABCA1-specific efflux and HDL metrics are showed in Supplemental Figure II.

Total and cAMP-dependent macrophage cholesterol efflux capacity correlate with ABCA1-specific cholesterol efflux capacity of BHK cells

Total macrophage efflux capacity⁸ and ABCA1-specific cholesterol efflux capacity assessed with J774 macrophages⁹ were strong, negative predictors of incident CVD events. We therefore examined the relationship between total efflux capacity with cAMP-stimulated J744 macrophages and ABCA1-specific efflux with BHK cells in all 126 CPC subjects. Cholesterol efflux capacity of the two assays correlated strongly (r=0.63). These results are in good agreement with previous observations that the ABCA1 pathway accounts for 30-40% of the total cholesterol efflux from J774 macrophages stimulated with cAMP.^{5, 7}

To further probe the relationship between ABCA1-specific cholesterol efflux capacity in J774 macrophages (with and without cAMP) and BHK cells expressing human ABCA1 (with and without mifepristone), we quantified the relationship between the two assays in 20 subjects with stable coronary disease. ABCA1-specific efflux in the two cell lines correlated strongly (r=0.56; Supplemental Figure III).

Discussion

Atorvastatin therapy reduced plasma LDL-C by 39% and apoB by 35%. Because LDL particles account for >85% of total apoB in the circulation, ²³ and because each LDL particle has one molecule of apoB, plasma apoB concentration can be considered a direct measure of LDL particle concentration.²⁴ Therefore, atorvastatin therapy decreased LDL concentration by about 35%.

We determined that combination therapy with statin and niacin increased HDL-C levels by 39% but increased HDL-P_IMA —the concentration of HDL particles determined by calibrated-IMA—by only 14%. Thus, the impact of combination therapy on HDL-P_IMA was much smaller than the statin’s effect on LDL concentration, as assessed by plasma levels of apoB. Our results show that lipid-altering therapies have can have markedly different effects on levels of HDL-C and HDL-P_IMA.

HDL subpopulations of different sizes were also affected differentially. Although L-HDL-P_IMA and M-HDL-P_IMA correlated positively with HDL-C, S-HDL-P_IMA correlated negatively. Moreover, atorvastatin monotherapy decreased the concentration of S-HDL-P_IMA by 10% without affecting total HDL-P_IMA, despite the significant 11% increase in HDL-C. Thus, changes in HDL-C levels can be dissociated from changes in HDL populations of specific sizes. Total HDL-P_IMA did not correlate with triglyceride levels (P=0.39, r=−0.08), in striking contrast to its strong negative correlation with HDL-C (P<0.0001, r=−0.44). Moreover, triglycerides correlated positively with S-HDL-P_IMA but negatively with M-HDL-P_IMA and L-HDL-P_IMA. Taken together, our observations suggest that HDL-P_IMA can offer clinically relevant information that cannot be obtained by measuring only levels of HDL-C.

We also investigated the impact of monotherapy and combination therapy on the cholesterol efflux capacity of serum HDL. Atorvastatin therapy did not change macrophage efflux capacity, as Khera and colleagues also concluded.⁷ Combination therapy, on the other hand, significantly increased macrophage efflux, contrasting with previous work.²¹ However, there are several notable differences between the two studies. First, the CPC subjects in our study were off lipid treatment at baseline, whereas 67% of the subjects in the other study were already taking a statin at entry. Second, the types and dosages of statin and niacin differed. Consistent with our observations, another small study showed that niacin therapy alone increased the cholesterol efflux capacity of HDL in both mouse macrophages and a human macrophage cell line.²⁵

To focus on ABCA1-specific cholesterol efflux capacity, we used BHK cells with and without inducible expression of this transporter.²⁶ In contrast to the results we obtained with J774 macrophages, combination therapy failed to improve the ABCA1-specific efflux capacity of serum HDL. Moreover, we observed a significant reduction (P=0.013) in ABCA1-specific cholesterol efflux capacity in the subjects on statin therapy, with or without niacin, suggesting that atorvastatin impairs cholesterol efflux to serum HDL by this pathway. These results indicate that adding niacin to statin therapy increases both HDL-C and efflux capacity with J774 macrophages but that statin therapy alone reduces cholesterol efflux capacity with the ABCA1 pathway. Impaired ABCA1-specific efflux capacity strongly and negatively associated with incident CVD events in the Dallas Heart Study.⁹ Moreover, patients with Tangier’s disease, who lack ABCA1 activity, accumulate cholesterol-laden macrophages in many different tissues.²⁷ Thus, the ABCA1 pathway appears to be a major route for cholesterol removal from macrophages in humans, and, in our study, statin therapy inhibited that pathway.

Our finding that atorvastatin therapy failed to alter HDL-P_IMA while significantly reducing both S-HDL-P_IMA and efflux via the ABCA1 pathway, suggests that atorvastatin might target the species of HDL that removes cholesterol from the artery wall. Indeed, a recent study provided strong evidence that cholesterol efflux through ABCA1 is promoted by small, dense HDL,¹² challenging the concept that only lipid-free/poorly lipidated apoA-I can promote efflux by this pathway.²⁸ In contrast, medium and large HDL particles are the preferred substrates for ABCG1, aqueous diffusion, and SR-B1,^{2, 13} and we found that combination therapy increased large HDL particles and cholesterol efflux capacity with macrophages. These observations may help explain why niacin increased HDL-C levels in clinical trials but failed to reduce cardiac risk.

It is currently not known if different statins have different effects on macrophage and ABCA1-specific cholesterol efflux capacity or how those effects might modulate CVD risk. In mouse studies rosuvastatin treatment enhanced ABCA1-specific cholesterol efflux capacity, but atorvastatin had no effect.²⁹ A small study showed that pitavastatin increased cholesterol efflux capacity from human macrophages by 8.6%.³⁰ In the JUPITER trial (Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin), rosuvastatin treatment increased HDL-C by 6%. In the same trial, rosuvastatin raised HDL-P_NMR by a lesser extent (3.8%).³¹ However, the concentration of HDL-P measured by NMR is not equivalent to that obtained by C-IMA.³² HDL-P_NMR, gives values for the stoichiometry of apoA-I and size distribution of HDL particles that are not consistent with those of orthogonal methods.^{10, 17} Nonetheless, HDL-P_NMR has been a better predictor of CVD than HDL-C in multiple studies,^{31, 33-36}, though this association has not been confirmed in other populations.³⁷

Our observations on the effects of atorvastatin and combination therapy on particle size and cholesterol efflux may have implications for the development of therapies that target HDL.³⁸ Because reduction in CVD risk strongly relates to the degree of LDL lowering,³⁹ it may be significant that the impact of niacin on HDL-P_IMA in statin-treated subjects was much smaller than that of statin monotherapy on LDL concentration. If cardioprotection relates to the total concentration of HDL particles, it may be necessary to identify agents that can elevate HDL-P_IMA more effectively than niacin. Moreover, niacin might fail to improve HDL’s atheroprotective properties because it selectively raises levels of L-HDL-P_IMA instead of boosting levels of pre-beta HDL (lipid-poor apoA-I) and S-HDL-P_IMA, which promote cholesterol efflux in vitro by the ABCA1 pathway.^{12, 28, 38} However, a recent study showed that evacetrapib, a CETP inhibitor that elevated pre-beta HDL levels and HDL-C levels, increased total and ABCA1-specific cholesterol efflux capacity in dyslipidemic patients.⁴⁰ Another CETP inhibitor failed to reduce cardiovascular risk in statin treated subjects,³ but the effect of CETP inhibitors on HDL subpopulations as quantified by HDL-P_ima is unknown. The relative contributions of specific HDL subpopulations to total cholesterol efflux capacity and ABCA1-specific cholesterol efflux capacity and their associations with CVD risk in humans deserves further investigation.

It is noteworthy that ABCA1-specific cholesterol efflux capacity also failed to improve in subjects on combination therapy, despite the marked increase in macrophage cholesterol efflux capacity. This selective increase in macrophage efflux capacity likely reflects the 103% rise in L-HDL-P_IMA, because large HDL particles are the preferred substrates for cholesterol export from cells by ABCG1, SR-B1, and aqueous diffusion.² If all of these pathways are important contributors to cholesterol efflux from macrophages in the artery wall, it may be necessary to increase the concentrations of both small and large HDLs for cardioprotection. Indeed, small and large HDL particles have been proposed to work together with ABCA1and ABCG1 to promote maximal cholesterol efflux from macrophages.^{12, 41} Also, hypercholesterolemic mice deficient in both ABCA1 and ABCG1 develop much greater atherosclerosis than mice deficient in only one of the transporters.⁴¹

Strengths of this study include its randomized and prospective matched-pairs design and the use of robust validated assays to assess HDL function, concentration, and size. The study has also some potential limitations. Because the CPC subjects had preexisting CVD, our results might be most relevant to such individuals rather than to healthy people and other populations, such as people with diabetes and/ or hypertriglyceridemia. Also, we assessed ABCA1-specific cholesterol efflux capacity with BHK cells transfected with human ABCA1; this assay has not been shown to predict CVD in clinical studies. However, impaired ABCA1-specific efflux capacity with J744 macrophages was a strong predictor of incident CVD subjects in the Dallas Heart Study ⁹, and we found a strong correlation between the ABCA1-specific efflux capacity of BHK cells and J774 macrophages. Finally, because calibrated ion mobility analysis requires isolation of HDL by ultracentrifugation, we were not able to quantify pre-beta HDL.

In summary, niacin markedly increased HDL-C, large, cholesterol-enriched HDL particles, and the cholesterol efflux capacity of macrophages, but it had much less effect on HDL-P_IMA. It also failed to improve ABCA1-specific efflux. Furthermore, we showed that statin therapy did not increase HDL-P_IMA or cholesterol efflux capacity with macrophages, even though it raised HDL-C. Our observations raise the possibility that these lipid-altering therapies might not target the relevant atheroprotective population of HDL. They further suggest that HDL metrics distinct from cholesterol content might be useful for determining which therapies reduce cardiovascular risk.

Significance.

Clinical trials with niacin elevate HDL-cholesterol (HDL-C) but fail to reduce cardiovascular events in statin-treated subjects. In the current studies, we showed that combination therapy with statin and niacin increased HDL-C but was much less effective at raising the concentration of HDL particles. Combination therapy also failed to improve ABCA1-specific efflux, a key cholesterol exporter in human macrophages. Recent studies strongly suggest that both the cholesterol efflux capacity of HDL and HDL particle concentration are much better predictors of incident and prevalent cardiovascular disease. Our observations raise the possibility that statin/niacin therapy may not target the relevant atheroprotective population of HDL. They further suggest that HDL metrics distinct from cholesterol content might be useful for determining which therapies can reduce cardiovascular risk.

Abbreviations

ABCA1

ATP-binding cassette transporter A1

ABCG1

ATP-binding cassette transporter G1

apoA-I

apolipoprotein A-I

apoB

apolipoprotein B

BHK

baby hamster kidney

CETP

cholesteryl ester transfer protein

CPC study

Carotid Plaque Composition (CPC) study

CVD

cardiovascular disease

HDL-C

high-density lipoprotein-cholesterol

HDL-P

high density lipoprotein particle concentration

HDL-P_IMA

HDL particle concentration determined by calibrated-IMA

HDL-P_NMR

HDL particle concentration determined by NMR

IMA

ion mobility analysis

LDL-C

low-density lipoprotein cholesterol

L- HDL-P_IMA

large HDL particle concentration

M- HDL-P_IMA

medium HDL particle concentration

NMR

nuclear magnetic resonance spectroscopy

S- HDL-P_IMA

small HDL particle concentration

SRB1

scavenger receptor B1.