Current Atherosclerosis Reports

, 16:379

Clinical and Biological Relevance of Statin-Mediated Changes in HDL Metabolism


    • Centre de Recherche de l’Institut Universitaire de Cardiologie et de Pneumologie de Quebec
    • Department of Medicine, Faculty of MedicineUniversité Laval
  • S. Matthijs Boekholdt
    • Department of CardiologyAcademic Medical Center
Statin Drugs (MB Clearfield, Section Editor)

DOI: 10.1007/s11883-013-0379-8

Cite this article as:
Arsenault, B.J. & Boekholdt, S.M. Curr Atheroscler Rep (2014) 16: 379. doi:10.1007/s11883-013-0379-8
Part of the following topical collections:
  1. Topical Collection on Statin Drugs


Although prospective studies consistently show that individuals with low levels of high-density lipoprotein (HDL) cholesterol are at increased cardiovascular risk, it is still not clear whether or not this relationship still holds in patients treated with statins who have achieved optimal low-density lipoprotein cholesterol levels. Additionally, the hypothesis that statin-mediated increases in HDL cholesterol levels have a impact on cardiovascular risk has not been clearly demonstrated. Statin therapy has little impact on the cholesterol content carried by HDL in the bloodstream (i.e., HDL cholesterol levels), but statins can induce a significant redistribution of HDL particle size, particle concentration, and physicochemical and functional properties. Our objective is to summarize the evidence arising from epidemiological, clinical, and experimental studies that suggests a potential role for statin therapy in the modulation of parameters of HDL metabolism and reverse cholesterol transport.


High-density lipoproteinsHigh-density lipoprotein functionalityCholesterol efflux capacitiesCholesteryl ester transfer proteinStatinsCardiovascular disease


Several studies have suggested that the balance of proatherogenic versus antiatherogenic lipoprotein subfractions is one of the most important predictors of cardiovascular diseases (CVD) [1, 2]. Proatherogenic lipoproteins include all apolipoprotein B (apoB)-containing lipoproteins such as low-density lipoproteins (LDL) and very low density lipoproteins, whereas the antiatherogenic subfraction is represented mostly by high-density lipoproteins (HDL) and their core protein component apolipoprotein A-I (apoA-I). Statin therapy is the cornerstone of cardiovascular risk management. Statins reduce cardiovascular risk by targeting atherogenic lipid subfractions, by inhibiting the rate-limiting step in the production of intracellular cholesterol, which raises the expression of the LDL receptor, thereby increasing the clearance of LDL particles from the circulation [3]. Thus, statin therapy has a profound impact on atherogenic lipoprotein subfractions and subsequent cardiovascular risk, but it is still debated whether or not antiatherogenic lipoprotein subfractions predict cardiovascular risk once LDL cholesterol or apoB levels have been reduced to very low levels. Additionally, several studies conducted over the past 20 years have shown that intracellular cholesterol levels, which are strongly impacted by statin therapy, regulate several transcription factors that target genes implicated in lipoprotein-lipid metabolism [4]. Therefore, statin therapy may influence other lipoprotein subfractions such as HDL. However, it is not known whether increases in HDL levels following statin therapy are to a certain extent responsible for the antiatherogenic effects of statin therapy. Our objectives are to review the role of HDL cholesterol as a predictor of cardiovascular risk in patients treated with statins and to shed light on the role of statin-induced changes in HDL levels as potential mediators of cardiovascular risk associated with statin therapy.

HDL and Cardiovascular Risk in Lipid-Lowering Randomized Clinical Trials

Statin therapy typically reduces cardiovascular risk by 20–40 % depending on the population studied and the statin type and dosage. This leaves an important residual risk that needs to be considered for the optimal management of patients at risk of CVD [5, 6]. Several randomized clinical trials that have tested the efficacy of statin therapy in reducing cardiovascular risk have documented the important predictive value of HDL cholesterol levels for recurrent CVD events. The Prospective Pravastatin Pooling Project (PPPP) included three outcome trials of pravastatin therapy including a total of 19,768 patients, among which 2,194 cardiovascular events were reported. In PPPP, HDL cholesterol levels were negatively associated with the risk of coronary artery disease death and myocardial infarction both in placebo-treated patients and in pravastatin-treated patients. However, pravastatin at a dosage of 40 mg/day is usually considered to be a low-dose statin, and thus many patients in PPPP did not achieve very low levels of LDL cholesterol, so the hypothesis stipulating that HDL cholesterol levels predict cardiovascular outcomes in patients achieving very low levels of LDL could not be tested. This hypothesis was tested, however, in trials comparing high-dose versus low-dose statin therapy. The Treating to New Targets (TNT) trial showed that 80 mg atorvastatin provided a 22 % cardiovascular risk reduction compared with 10 mg atorvastatin in patients with stable coronary heart disease [7]. A post hoc study of the TNT trial performed by Barter et al. [8] showed that even in patients achieving LDL cholesterol levels below 70 mg/dL, patients in the top HDL cholesterol quintile had a multivariate adjusted 39 % decrease in CVD risk compared with patients in the bottom HDL cholesterol quintile. The association between HDL cholesterol levels and CVD risk, however, appeared to be stronger in the 10-mg arm than in the 80-mg arm. An association between baseline and on-treatment HDL cholesterol levels with atherosclerosis burden was also reported by Nicholls et al. [9] in a pooled analysis of 1,455 statin-treated patients included in four randomized controlled trials that used serial coronary intravascular ultrasonography to quantify atherosclerosis progression. In that study, HDL cholesterol was associated with both total atheroma volume and percent atheroma volume in the coronary arteries.

In the EPIC-Norfolk prospective population study, we have shown that physicochemical properties of HDL particles such as size and particle concentration measured by nuclear magnetic resonance (NMR) spectroscopy were equally as predictive of CVD risk as HDL cholesterol concentrations [10]. Interestingly, after adjustment for triglyceride and apoB concentrations, the relationship between HDL cholesterol concentration and size and CVD risk was no longer observed, but the association between total HDL particle concentrations and CVD risk was not affected. These results were recently replicated by Mackey et al. [11] with data from the Multi-Ethnic Study of Atherosclerosis. The impact of NMR-measured characteristics of HDL particles and the risk of CVD was recently documented by the investigators of the Heart Protection Study [12]. In both placebo-treated individuals and simvastatin-treated individuals, there was a negative association between HDL cholesterol levels, apoA-I levels, HDL particle number, and the risk of occlusive cardiovascular events. Although the association between HDL size and CVD risk was not studied in both arms separately, there was a negative association between HDL size and CVD risk in the combined study arms. This association was, however, largely explained by the close association between HDL size and HDL particle number.

Not all statin trials have reported significant associations between HDL cholesterol levels and CVD risk. For instance, in the Collaborative Atorvastatin Diabetes Study (CARDS), HDL cholesterol and apoA-I levels hardly changed during the trial, and on-treatment levels did not predict CVD risk in diabetic patients treated with atorvastatin [13]. There was, however, a treatment-related change in HDL size, which may have indicated that quite profound changes in HDL metabolism occurred [14]. In the Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin (JUPITER) trial, HDL cholesterol levels predicted cardiovascular risk in patients allocated to placebo treatment, but not to rosuvastatin treatment [15]. Similar results were also reported for apoA-I levels. It is not clear, however, if this lack of association between HDL cholesterol and apoA-I levels and CVD risk in rosuvastatin-treated individuals is because the overall study population was rather healthy or because of the lack of statistical power (251 and 142 events, respectively for placebo-treated and rosuvastatin-treated individuals [16]), or whether these findings have a true biological significance.

To determine whether or not HDL cholesterol and apoA-I levels were associated with increased cardiovascular risk in patients treated with statins, and especially in those who achieved very low LDL cholesterol levels, we performed a meta-analysis on individual patients that included 38,153 patients treated with statins in eight large-scale randomized clinical trials [17•]. During the course of the study, 5,387 study participants had a cardiovascular event. Our results show unequivocally that even in patients who achieved very low levels of LDL cholesterol (below 50 mg/dL) after 1 year of statin therapy, those who had low HDL cholesterol and/or apoA-I levels were at increased cardiovascular risk compared with those with higher HDL cholesterol and/or apoA-I levels.

Despite the significantly increased risk associated with low HDL cholesterol levels, whether or not increasing HDL cholesterol levels will translate into direct cardiovascular benefits is unknown. The recent outcome trials with drugs that raise HDL cholesterol levels have yielded rather negative results (reviewed in [18]). However, whether or not statin-induced changes in HDL cholesterol levels are associated with a reduced CVD risk is still a matter of debate. The first study that addressed this question was a meta-analysis of 101 studies of lipid-modifying agents and cardiovascular outcomes performed by Briel et al. [19]. The results of this meta-analysis suggested that changes in LDL cholesterol levels, but not HDL cholesterol levels, were associated with cardiovascular benefits, which is to be expected given that the vast majority of studies were LDL-lowering trials in which HDL cholesterol levels were barely affected. In our view, the results of this study, which significantly challenged the HDL hypothesis, were to a large extent explained by the fact that this study was a study-level meta-analysis, which may have led to cancellation of opposite effects within trials, by the fact that the association between changes in apoA-I levels and cardiovascular risk was not reported, and by the fact that most of the analyses accounted for the changes in LDL cholesterol levels. Two other smaller studies have investigated the impact of lipid-lowering therapy-induced changes in HDL cholesterol levels and cardiovascular risk. The first study used data from the Framingham Heart Study and the second combined data from the EPIC-Norfolk and Rotterdam studies [20, 21]. Although the report from the Framingham study suggests that changes in HDL cholesterol levels were significantly associated with a decreased risk of CVD even after adjustment for changes in LDL cholesterol levels, the report from the EPIC-Norfolk and Rotterdam studies showed that changes in HDL cholesterol levels did predict the risk of CVD, but not after adjustment for traditional CVD risk factors. However, these studies included a very low number of cardiovascular events. In our meta-analysis of individual patient data discussed above, we were able to document the association between changes in HDL cholesterol and apoA-I levels between the baseline and after 1-year of statin therapy and the risk of subsequent cardiovascular events. In fact, both baseline and on-trial HDL cholesterol and apoA-I levels were available for 37,747 study participants, of whom 3,902 developed a major cardiovascular event after the 1-year HDL cholesterol and apoA-I measurements. The multivariate adjusted hazard ratio for cardiovascular events per standard deviation HDL cholesterol increment was 0.98 [95 % confidence interval (CI), 0.94–1.01], which is in line with some of the studies discussed above. However, the multivariate adjusted hazard ratio for cardiovascular events per standard deviation apoA-I increment was 0.93 (95 % CI, 0.90–0.97), which suggests that changes in apoA-I levels but not HDL cholesterol levels provide significant cardiovascular benefits. On the basis of these results, we believe that targeting apoA-I may be as important as, if not more important than, targeting HDL cholesterol levels to reduce cardiovascular risk.

Impact of Statin Therapy on HDL Functionality

Cardiovascular medicine and research focusing on HDL cholesterol has developed considerably over the past decade, and is now looking beyond the simple measurement of the static pool of cholesterol carried by HDL particles (HDL cholesterol) to determine whether or not the assessment of HDL functional properties could provide more information on cardiovascular risk. Indeed, recent technological advances have made the assessment of functional properties of HDL particles possible and ready to be used in large-scale prospective studies. To date, the most important proxy of HDL functionality is the measurement of HDL cholesterol efflux capacities. By incubating macrophages containing labeled free cholesterol in vitro with isolated HDL particles from study patients, one can now estimate the capacity of HDL particles to promote cholesterol efflux from macrophages for each study patient. Although a definite association between HDL cholesterol efflux capacities and the onset of CVD has not been established, some have suggested that HDL cholesterol efflux capacities could be a better biomarker of the presence of atherosclerosis than HDL cholesterol levels [22•, 23•]. Khera et al. [22•] measured HDL cholesterol efflux capacities in approximately 100 patients treated with 10 mg atorvastatin, 80 mg atorvastatin, or 40 mg pravastatin for 16 weeks and documented no effects of statin therapy on HDL cholesterol efflux capacities. However, they did report that treatment with 30 mg pioglitazone for 6 weeks increased HDL cholesterol efflux capacities by approximately 11 % in a small sample of 16 patients. The pathways that mediate cholesterol efflux from lipid-laden macrophages involve several membrane transporters such as ATP-binding cassette A1 (ABCA1) and ATP-binding cassette G1 (ABCG1) as well as scavenger receptor class B type 1 (SR-B1) which interact mostly with apoA-I, which is the core protein component of HDL particles [24]. Using two-dimensional gradient gel electrophoresis, Asztalos et al. [25] have provided evidence that larger, antiatherogenic HDL particles promote cholesterol efflux via the SR-B1 transporter, whereas pre-β-1 HDL promote efflux via the ABCA1 transporter. We have also recently shown that larger HDL particles as assessed by NMR measurements promoted cholesterol efflux via the SR-B1 pathway, and we documented a similar association between pre-β-1 HDL levels and ABCA1-mediated efflux [26]. In the study of Khera et al. [22•], HDL cholesterol efflux capacities from all transporters were measured simultaneously. As opposed to these results, Guerin et al. [27] have shown that atorvastatin therapy could increase SR-B1-dependent cholesterol efflux capacities in a dose-dependent manner in patients with hyperlipidemia. Using samples from the STELLAR study, Asztalos et al. [28] documented the impact of high-dose statin therapy (40 mg rosuvastatin and 80 mg atorvastatin) on HDL particle profiles obtained by two-dimensional gradient gel electrophoresis. Statin therapy profoundly decreased pre-β-1 HDL levels (by -41 and -40 % for 40 mg rosuvastatin and 80 mg atorvastatin, respectively) and increased the large α-1 HDL particles (by 24 and 12 % for 40 mg rosuvastatin and 80 mg atorvastatin, respectively). In another statin trial, however, the same group found little or no impact of statin therapy on pre-β-1 HDL levels, but did confirm the impact of statins on larger α-1 HDL particles [29]. In a type 2 diabetic population, atorvastatin therapy was found to lower the amount of pre-β-1 HDL particles, which appeared to be closely associated with the decrease in phospholipid transfer protein activity [30]. Therefore, we believe that the impact of statin therapy on transporter-specific cholesterol efflux needs to be documented in studies using more than one cellular model at the same time (Fig. 1).
Fig. 1

Impact of statin therapy on HDL subclasses and cholesterol efflux capacities. Studies have been consistent in showing that statins increase the large α-1 HDL subpopulations. Given that these subfractions promote cholesterol efflux via the scavenger receptor class B type 1 (SR-B1) and ATP-binding cassette G1 (ABCG1) transporters, it could be speculated that statin therapy increases cholesterol efflux induced by these transporters. The impact of statin therapy on pre-β HDL particles is more controversial. Given that these subfractions promote cholesterol efflux via the ATP-binding cassette A1 (ABCA1) transporter, the impact of statin therapy on HDL ABCA1-mediated cholesterol efflux capacities is uncertain. Studies have shown that statin therapy has no impact on HDL cholesterol efflux capacities measured in systems that use all transporters simultaneously

The functional properties of HDL particles are not limited to HDL cholesterol efflux capacities. The group of Fogelman has pioneered a technique that makes possible the assessment of the anti-inflammatory function of HDL particles on the basis of their ability to inhibit macrophage chemotactic activity in a coculture system. In a clinical investigation, they showed that treatment with 40 mg simvastatin favorably affected the anti-inflammatory properties of HDL [31]. Using a similar technique, Patel et al. [32] showed that statin therapy improved the so-called HDL inflammatory (HII) index. In another placebo-controlled study, Khera et al. [33] measured HDL cholesterol levels, HDL cholesterol efflux capacities, and HII in patients before and after extended-release niacin therapy in 19 patients. They showed that niacin therapy significantly raised HDL cholesterol levels by 29 %, but failed to improve significantly cholesterol efflux capacities (+2 %) and HII (+9 %). The failure of two recent large-scale phase 3 randomized clinical trials to provide any cardiovascular benefits associated with niacin therapy [34, 35] suggests that merely raising HDL cholesterol levels does not putatively lead to a reduction in cardiovascular risk. In fact, it could be speculated that the fact that niacin does not improve HDL functionality may account to a certain extent for these negative results.

How Do Statins Influence HDL Metabolism?

Depending on statin potency, statin dosage, the study design, and population characteristics, statins typically increase HDL cholesterol levels by 10–15 % [36]. At the cellular level, statins inhibit the factor Rho [37]. One of the consequences of this inhibition is the activation of the nuclear receptor peroxisome-proliferator-activated receptor α, which increases the genetic expression of several key targets involved in HDL metabolism such as apoA-I, hepatic lipase, and lipoprotein lipase. By simultaneously increasing the clearance of circulating LDL particles and reducing the cellular biosynthesis of cholesterol, statins decrease by approximately 30 % the number of circulating atherogenic lipoproteins as estimated by plasma apoB concentration [38]. As a direct consequence, the number of lipoproteins available that send triglyceride molecules in exchange for cholesterol from HDL, a process mediated by cholesteryl ester transfer protein (CETP), is decreased. Additionally, statins directly decrease the genetic expression of CETP at the cellular level [39] and reduce CETP activity in the bloodstream [40]. A dose-escalation study of rosuvastatin and atorvastatin further confirmed the important impact of statins on enzymes that modulate the physicochemical properties and functionality of several lipoproteins [41•]. In that trial, increasing doses of rosuvastatin and atorvastatin were associated with a reduction in CETP mass and activity, phospholipid transfer protein mass and activity, and lipoprotein-associated phospholipase A2 mass and activity, without affecting the mass of lecithin cholesterol acyltransferase , an enzyme implicated in the maturation of HDL particles. Therefore, in part because of these reasons, the esterified cholesterol pool is maintained, whereas the triglyceride content of HDL particles is decreased, which makes HDL particles less susceptible to hepatic lipase-dependent hydrolysis and removal. In parallel to these findings, investigators of the REGRESS study found that CETP genotype predicted CVD risk in coronary artery disease patients enrolled in that study and concluded that statin efficacy could depend on CETP genotype and associated circulating mass [42]. However, a meta-analysis of statin trials could not confirm an interaction between the efficacy of statin therapy and CETP genotype [43].

Metabolic studies performed in humans have provided evidence that statin therapy can influence the kinetics of apoA-I [44]. In one such study, the impact of 20 mg rosuvastatin was compared with that of placebo in eight patients with type 2 diabetes [45]. Both apoA-I production rates and apoA-I fractional catabolic rates were significantly decreased, which ultimately led to a 27 % increase in apoA-I plasma residence time. The apoA-I fractional catabolic rate correlated positively with plasma CETP mass, which further confirms the pivotal role of CETP in statin-induced changes in HDL metabolism. However, whether the increase in plasma residence time further contributes to plasma reverse cholesterol transport and other antiatherogenic function of HDL is unknown and needs to be studied further.

Can Other LDL-Lowering Therapies Influence HDL Metabolism in Familial Hypercholesterolemia?

Several lines of evidence suggest that the reverse cholesterol pathway could be impaired in patients with familial hypercholesterolemia (FH). For instance, the fractional catabolic rate of apoA-I could be increased and its synthesis could be decreased in patients with FH, which could ultimately lead to lower HDL cholesterol levels [9]. Bellanger et al. [46•] recently measured several parameters of the reverse cholesterol pathway in 12 patients with FH and 12 participants without FH. They showed that larger HDL2 particles have lower cholesterol efflux capacities via the SR-B1 and ABCG1 pathways in FH patients. In another assay, they also showed that HDL particles from patients with FH had a considerably lower ability to deliver cholesterol esters to hepatic cells, which is one of the key steps in reverse cholesterol transport. Furthermore, they injected labeled HDL particles isolated from patients with FH and normolipidemic patients into mice with and without genetic ablation of the LDL receptor, and found that in both models labeled cholesterol levels were lower in the liver and in the feces of mice into which labeled HDL isolated from FH patients had bene injected. Plasma radioactivity counts, however, were not different.

The currently available treatment for FH does not correct this altered HDL particle function. However, emerging therapies for FH may soon provide a shift in paradigm and potentially address this issue. For instance, LDL apheresis has recently been shown to substantially reduce plasma levels of pre-β-1 HDL levels (53 % decrease), with a concomitant decrease in plasma CETP mass and activity [47•]. However, HDL cholesterol efflux capacities did not change consistently in the various cell models that were studied. Furthermore, although the impact of proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibition on HDL metabolism has not been fully documented, a recent placebo-controlled, multiple-dose study conducted with a humanized IgG2ΔA monoclonal antibody RN316 in hypercholesterolemic patients showed that even at the lowest RN316 dose studied (0.25 mg/kg, intravenously), patients treated with RN316 had an increase in NMR-measured HDL particle number of 11 % [48]. At the highest doses (1.0 and 1.5 mg/kg, intravenously), the mean HDL size also increased significantly following RN316 treatment. Therefore, on top of dramatically lowering LDL cholesterol levels, PCSK9 inhibition may have a positive role in modulating HDL physicochemical properties. Whether or not this will translate into improved HDL function is unknown.


A large body of evidence suggests that HDL cholesterol levels are a strong risk factor for CVD, both in low-risk individuals and in individuals treated with highly potent statins who meet their LDL cholesterol goals. However, whether or not changes in HDL cholesterol levels (induced or not induced by statins) over time reduce cardiovascular risk is still unknown. The results of our meta-analysis suggest that increments in apoA-I rather than HDL cholesterol levels are associated with a reduction in CVD risk. Still, the mechanisms responsible for these changes are poorly understood, and novel and innovative research tools such as those offered by genetic epidemiology may be helpful to better understand the impact of statin therapy on HDL metabolism. In this regard, we believe that the recent formation of the Genomic Investigation of Statin Therapy consortium should help identify genetic variants associated with statin-induced changes in LDL and HDL cholesterol levels by performing meta-analyses of genome-wide scans for these traits in randomized clinical trials and observational studies [49]. Further research should also be conducted toward the identification of the biological mechanisms that lead to improvements in the functionality of HDL particles (if any) and to identify the genomic regions associated with such improvements in order to fully appreciate the impact of statin therapy on HDL biology.

Conflict of Interest

Benoit J. Arsenault is a consultant to Pfizer.

S. Matthijs Boekholdt is a consultant to Pfizer.

Human and Animal Rights and Informed Consent

This article does not contain any studies with human or animal subjects performed by any of the authors.

Copyright information

© Springer Science+Business Media New York 2013