Cardiovascular disease risk reduction by raising HDL cholesterol – current therapies and future opportunities

Abstract

Since the first discovery of an inverse correlation between high-density lipoprotein-cholesterol (HDL-C) levels and coronary heart disease in the 1950s the life cycle of HDL, its role in atherosclerosis and the therapeutic modification of HDL-C levels have been major research topics. The Framingham study and others that followed could show that HDL-C is an independent cardiovascular risk factor and that the increase of HDL-C of only 10 mg·L⁻¹ leads to a risk reduction of 2–3%. While statin therapy and therefore low-density lipoprotein-cholesterol (LDL-C) reduction could lower coronary heart disease considerably; cardiovascular morbidity and mortality still occur in a significant portion of subjects already receiving therapy. Therefore, new strategies and therapies are needed to further reduce the risk. Raising HDL-C was thought to achieve this goal. However, established drug therapies resulting in substantial HDL-C increase are scarce and their effect is controversial. Furthermore, it is becoming increasingly evident that HDL particle functionality is at least as important as HDL-C levels since HDL particles not only promote reverse cholesterol transport from the periphery (mainly macrophages) to the liver but also exert pleiotropic effects on inflammation, haemostasis and apoptosis. This review deals with the biology of HDL particles, the established and future therapeutic options to increase HDL-C and discusses the results and conclusions of the most important studies published in the last years. Finally, an outlook on future diagnostic tools and therapeutic opportunities regarding coronary artery disease is given.

Introduction

Coronary artery disease (CAD) is one of the major causes of death, worldwide (Yusuf et al., 2001). Dyslipidaemia is characterized by elevated low-density lipoprotein cholesterol (LDL-C) and decreased high-density lipoprotein cholesterol (HDL-C) and is a known risk factor for development and progression of atherosclerosis in CAD (Arca et al., 2007). In the past, therapeutic strategies were focused on lowering LDL-C, primarily by the use of statins. However, as CAD events seem not to be satisfyingly prevented by current treatment schemes (LaRosa et al., 1999), therapeutic options to increase HDL-C stepped into spotlight, recently.

The role of HDL in atherosclerosis

Since the middle of the 20th century, research has been engaged with lipoproteins in health and in disease. In 1977, the Framingham study was the first large-scale study giving evidence that low levels of HDL-C is a major risk factor for CAD. The association between the incidence of CAD and HDL-C levels was stronger than for LDL levels (Gordon et al., 1977). In a later re-evaluation of this same study, which included follow-up data of 12 years, low HDL-C levels were even associated with increased mortality (Wilson et al., 1988). Jenkins et al. confirmed the correlation between HDL-C levels and CAD observed in epidemiological studies by performing coronary angiographies and found a significant association between HDL-C levels and the severity of atherosclerosis (Jenkins et al., 1978). Some years later, Gordon et al. noted a 2–3% decrease in CAD risk with each increase by 10 mg·L⁻¹ in HDL-C (Gordon et al., 1989). A recent meta-analysis, including 302.430 subjects from 68 long-term prospective studies, supported the importance of HDL-C measurement in the risk assessment for CAD (Di Angelantonio et al., 2009).

HDL metabolism and its atheroprotective properties

High-density lipoproteins are a heterogeneous group of particles that differ in size, shape, density, cholesterol and phospholipid content, as well as in apolipoprotein composition. The life cycle of HDL begins with apolipoprotein A-I (ApoA-I) being secreted by the liver. As ApoA-I binds circulating phospholipids and cholesterol, nascent discoid lipid-poor HDL particles are formed. These immature HDL particles trigger cholesterol efflux in subendothelial macrophages and fibroblasts and, via interactions with ATP-binding cassette transporter A1 (ABCA1), store the cholesterol in their core, after esterification by lecithin-cholesterol acyltransferase (LCAT). Such HDL particles obtain a spheric shape, resulting in the two main mature particles, HDL2 and HDL3. Subsequently, HDL deliver their cholesterol load either directly to the liver via scavenger-receptor B-I (SR-BI) or indirectly by shifting cholesterol to very low-density lipoprotein (VLDL) or LDL particles, which in turn are taken up by the liver via the LDL-receptor. This shift is carried out by cholesterol ester transfer protein (CETP), a protein associated with HDL. Either way, cholesterol finally gets excreted into the feces as neutral steroids or bile acids.

This process of cholesterol clearance was named ‘reverse cholesterol transfer’ (RCT) and has been the explanation for HDL's association with atheroprotection for a long time.

However, other anti-atherosclerotic properties of HDL have been discovered in recent years.

Anti-inflammatory effects of HDL

Atherosclerosis is an inflammatory disease. Adhesion and migration of immune cells into the vessel wall, as well as inflammatory cytokines and chemokines orchestrating these processes are essential (Ross, 1999b). In vitro studies have shown that HDL inhibits the expression of endothelial adhesion molecules, such as vascular cell adhesion molecule-1 (VCAM-1), intercellular adhesion molecule-1 (ICAM-1), and E-selectin (Cockerill et al., 1995; 1999; Baker et al., 1999). However, these observations could not be repeated in human aortic and coronary artery endothelial cells (Stannard et al., 2001; Zhang et al., 2002). Another anti-inflammatory effect of HDL is the inhibition of monocyte chemoattractant protein-1 (MCP-1) production, as demonstrated in human endothelial cells as well as murine vascular smooth muscle cells from the aorta (Mackness et al., 2004; Tolle et al., 2008). In vivo studies could confirm some of these in vitro observations. (Cockerill et al., 2001) Nicholls et al. inserted a non-occlusive collar around carotid arteries in rabbits to imitate carotid stenosis and infused reconstituted HDL (rHDL). The treatment with rHDL resulted in the inhibition of VCAM-1, ICAM-1 and MCP-1 production, as well as in inhibition of leukocyte infiltration and the abolishment of reactive oxygen species production in the artery wall (Nicholls et al., 2005).

Antioxidant effects of HDL

Oxidation of LDL in the arterial wall is one of the most crucial steps in the development of atherosclerotic lesions (Ross, 1999a). HDL is associated with several antioxidant enzymes, such as paraoxonase, platelet-activating factor acetylhydrolase, LCAT and reduced glutathione selenoperoxidase. In healthy subjects, HDL is able to inhibit the formation of mildly oxidized LDL (oxLDL). On the other hand, lipid hydroperoxides can be transfered from oxLDL to HDL by CETP. This oxidized HDL was shown to be cleared rapidly from circulation. Thus, HDL seems to be important for the physiological detoxification of oxLDL, as well (Garner et al., 1998). However, HDL of patients with CAD seems to lack this ability (Navab et al., 2000; 2001).

Anti-apoptotic effects of HDL

Cell death in response to endothelial injury is a constant process in atherosclerosis. The main stimulants are oxLDL, proinflammatory cytokines, as well as growth factor deprivation. In bovine aortic endothelial cells, Suc et al. demonstrated that HDL, especially ApoA-I, is able to prevent oxLDL-induced apoptosis by blocking intracellular signalling involved in apoptosis (Suc et al., 1997). In other studies, TNF-α-induced apoptosis, as well as apoptosis induced by growth factor deprivation, could be avoided by incubating the endothelial cells with HDL (Sugano et al., 2000; Nofer et al., 2001).

Anti-thrombotic effects of HDL

HDL seems to be implicated in haemostasis. On the one hand, it is able to regulate platelet adhesion by inhibiting platelet activation and aggregation (Conlan et al., 1993; Cockerill et al., 1999; Nofer et al., 2010). On the other hand, HDL influences plasmatic coagulation. It was shown to inhibit thrombin-induced tissue factor expression in human endothelial cells (Viswambharan et al., 2004), to inhibit the activation of factor X (Carson, 1981), and to enhance the activities of activated protein C and protein S, two important anticoagulants (Griffin et al., 1999).

HDL composition in health and in disease

Despite all the earlier-mentioned findings, it was recently shown that HDL may malfunction in chronic inflammation (Natarajan et al., 2010; Saemann et al., 2010). As HDL particles are heterogeneous in their composition, scientists have questioned whether the observed impairment is due to a difference in protein assembly from one individual to another. Using shotgun proteomics, Vaisar et al. identified 48 HDL-associated proteins of which only 22 were linked to cholesterol and lipoprotein metabolism whereas 23 were linked to acute phase-response proteins and 3 to complement activation. Interestingly, HDL particles of healthy individuals differed in their composition with those of patients with CAD. Furthermore, HDL particles isolated from atherosclerotic lesions were very similar to the ones isolated from CAD patients' plasma (Vaisar et al., 2007). Changes in HDL composition and function were also observed in other diseases. HDL of patients with diabetes, rheumatoid arthritis or chronic kidney disease displayed reduced antioxidant and anti-inflammatory properties as well as an altered protein cargo (Saemann et al., 2010; Holzer et al., 2011; Tolle et al., 2012).

A very recently published work by Kar et al. demonstrates the ability of oxidized phosphlipids (Ox-PL) to destabilize HDL particles and alter their function. Furthermore, they could show that reconstituted HDL particles with high amounts of Ox-PL had less capacity to stimulate paraoxonase 1 (PON1), a protein that plays an integral role in the stimulation of reverse cholesterol transport from macrophages to HDL particles (Berrougui et al., 2012; Kar et al., 2012).

Established drug therapies

Statins

Statins (hydroxymethyl-glutaryl-coenzyme A reductase inhibitors) are the standard therapy in primary and secondary cardiovascular prevention. According to the Treating to New Targets (TNT) trial, high-dose treatment along with treatment goals of LDL-C below 1000 mg·L⁻¹ are believed to achieve the best results in risk reduction (LaRosa et al., 2005). High-dose statins are even beneficial in patients with normal LDL-C levels at baseline (Ridker et al., 2009). In addition to effectively reducing LDL-C, statins are also able to increase HDL-C by 6–14.7% (Downs et al., 1998; Simes et al., 2002; Streja et al., 2002; Athyros et al., 2004; Nissen et al., 2006).

Recently, a meta-analysis of 37 randomized studies, including 32 258 dyslipidemic patients, investigated the effect of statins on HDL-C. It was shown that all statins significantly raise HDL-C levels, Athough in a dose-dependent manner and not each statin to the same extent. Interestingly, changes in LDL-C and HDL-C concentrations were statistically independent from each other, no matter which statin was used (Barter et al., 2010). However, the clinical importance of HDL-C concentrations in patients on statins is controversially discussed.

Some studies published data in favour of HDL-C and its role in disease prevention. In the GREACE study, a mean 3 year therapy with atorvastatin increased HDL-C by 7% and a hazard ratio of 0.85 for every 40 mg·L⁻¹ (0.1 mmol·L⁻¹) increase was calculated, independent of LDL-C lowering (Athyros et al., 2004). It is important to mention though that this study was planned in a ‘Real Life’ setting. Thus, the group receiving atorvastatin was treated in cardiology clinics whereas the control group was treated by general practitioners according to ‘usual’ medical care. In ASTEROID, rosuvastatin therapy led to an HDL-C increase of 14.7% accompanied by a significant regression in atherosclerosis. However, LDL-C was simultaneously reduced by 53.2% and no information was given whether LDL-C and HDL-C had an independent impact. (Nissen et al., 2006)

Conversely, results of the primary prevention trial JUPITER showed no predictive power of HDL-C for the end points first non-fatal myocardial infarction (MI) and stroke in patients already on a high-dose statin treatment. Consequently, the authors, although still supporting HDL-C measurements in the initial risk assessment, question the necessity of measurements during adequate statin therapy (Ridker et al., 2010). However, critics point out that HDL-C levels at baseline have been quite high in the JUPITER population. Referring to results from the VOYAGER meta-analysis, saying that the extent of HDL-C increase is dependent on HDL-C levels at baseline (low HDL-C levels are associated with a higher increase during statin treatment) the observations made in JUPITER could therefore lead to false conclusions (Barter et al., 2010).

Niacin

Niacin (nicotinic acid) is the oldest and most effective agent in increasing HDL-C. Discovered in the 1950s (Altschul et al., 1955), it caught interest primarily because of its ability to decrease LDL-C, but with the implementation of statin therapy in standard care, it slowly disappeared. Decades later, niacin reappeared in clinical use as a potential substance to increase HDL-C. Many clinical trials have been evaluating its benefit on atherosclerotic disease, in monotherapy or in combination with other drugs. However, these trials were rather small.

The Coronary Drug Project assessed niacin monotherapy in a randomized, placebo-controlled clinical trial in 3906 patients with previous MI. Niacin was shown to decrease the occurrence of MI at 6 years (The Coronary Drug Project Research Group, 1975) as well as total mortality at 15 years significantly (Canner et al., 1986).

In the Cholesterol Lowering Atherosclerosis Study, niacin was combined with colestipol, a bile-acid sequestrant, and change in atherosclerosis was quantified by coronary angiography in 162 men with previous coronary bypass surgery. After 2 years, drug-treated subjects showed a 37% increase in HDL-C levels, a 43% reduction in LDL-C and a 26% reduction in total cholesterol (TC). Atherosclerosis regression was seen in significantly more drug-treated than in placebo-treated subjects (Blankenhorn et al., 1987). At 4 years, these results remained significant. Furthermore, significantly fewer drug-treated subjects developed new lesions in native coronary arteries and bypass grafts. (Cashin-Hemphill et al., 1990)

The combination of niacin with statin therapy was evaluated in HATS and the ARBITER studies. The first included 180 patients with CAD, normal LDL-C and low HDL-C levels. After 3 years, the combination of simvastatin and niacin compared with placebo lead to a significant regression of coronary stenosis. LDL-C was reduced by 42% and HDL-C increased by 26% (Brown et al., 2001). The ARBITER 2 study included 167 patients with the same characteristics, but already on a current statin therapy and either niacin or placebo was added for 1 year. The primary endpoint was change in carotid intima-media thickness (IMT). In the niacin group, HDL-C increased by 21%. The placebo group had a significant increase in IMT, whereas the IMT of niacin-treated patients remained the same. However, there was no significant overall difference in IMT progression between the two groups (Taylor et al., 2004). As the ARBITER group did not want to dismiss on niacin, a subsequent study followed, in which a longer treatment period was tested (ARBITER 3). After 2 years of application, niacin-treated patients presented with a significant regression of IMT compared with the placebo group, and changes in HDL-C were independently associated with this arterial improvement (Taylor et al., 2006). Recently, the same group published the results of the ARBITER 6 study, which compared treatment with niacin versus ezetimibe added to a background statin therapy. Once again, niacin could prove its ability to significantly reduce IMT, whereas ezetimibe could not. (Villines et al., 2010)

The large-scale AIM-HIGH trial was scheduled to present further results regarding the effect of niacin plus statin on therapy of cardiovascular disease in 2012; however, the study was terminated prematurely due to a lack of effect (more on that in the conclusions section; Brown, 2005).

In 2010, a meta-analysis of 11 randomized controlled trials, also including the ones mentioned earlier, evaluated the influence of niacin therapy on cardiovascular outcome. A therapeutic strategy including niacin was shown to significantly reduce major coronary events by 25%, stroke by 26%, and all cardiovascular events by 27%. Unfortunately, due to lack of well-powered studies on niacin monotherapy, the effect of niacin alone could not be assessed (Bruckert et al., 2010).

While two receptors for niacin were discovered a few years ago (HM74 and HM74A), it is not quite clear how niacin leads to an increase in HDL-C levels. Various theories have been discussed (Soudijn et al., 2007). One rationale is based on niacin influencing CETP (Hernandez et al., 2007; van der Hoorn et al., 2008), supported by the observation of an inverse correlation between trigyleride (TG) levels and HDL-C levels (Geurian et al., 1992). Another one is based on niacin's inhibitory involvement in HDL-C catabolism (Shepherd et al., 1979; Bodor and Offermanns, 2008). At last, niacin was also shown to have pleiotropic effects apart from lipid modification, such as increasing the expression of PPAR-γ in macrophages and the endothelium (Vosper, 2009b).

Niacin has one significant flaw causing prostaglandin-mediated vasodilation, a phenomenon called flushing (Kamanna et al., 2009). This skin reaction makes the implementation of niacin in patient care very difficult as it leads to low compliance and treatment discontinuation. A promising strategy to avoid flushing without altering niacin's impact on lipid metabolism is the combination with an antagonist of the prostaglandin D2 receptor subtype 1 (PD1), that is laropiprant (Parhofer, 2009). In 2008, Merck's Tredaptive™, which combines extended-release niacin with laropiprant, has been approved for marketing in the European Union, Iceland and Norway, but failed to enter the US market (known as Cordaptive™ in the United States). The randomized, placebo-controlled, double-blind study HPS2-THRIVE due in 2013 will hopefully provide answers on the efficacy and safety of niacin plus laropiprant in cardiovascular disease (Armitage et al., 2007).

Fibrates

Fibrates (fibric acid derivates) are synthetic ligands for PPAR-α. The hallmarks of fibrate therapy are a substantial decrease of plasma TG levels ranging from 30 to 50% and a moderate increase of HDL-C levels ranging from 5 to 15% (Goldenberg et al., 2008).

Activation of PPAR-α leads to β-oxidation of free fatty acids (FFA) in the liver reducing the availability of fatty acids for VLDL synthesis and secretion. Furthermore, the expression of the gene coding for lipoprotein lipase is increased and apolipoprotein C-III (ApoC-III) expression in the liver is decreased. This leads to a reduced synthesis and simultaneously increased hydrolysis of triglyceride-rich lipoproteins. HDL-C increase results from raising the expression of apoliproteins A-I and A-II decreasing CETP-mediated transfer of cholesterol from HDL to VLDL and enhancing cell cholesterol efflux via the induction of ABCA 1 and the decrease of SR-B1 in the liver (Berger et al., 2005; Farnier, 2008).

Fibrates also reduce plasma LDL-C levels. This effect, however, is variable. A considerably more important effect on LDL is the ability to change particle size from smaller, more atherogenic particles to larger, less atherogenic particles. (Goldenberg et al., 2008)

Besides the effects on the lipid profile, fibrates also exert other pleiotropic functions via PPAR-α by modulating platelet aggregation and endothelial dysfunction as well as by acting anti-inflammatory through the inhibition of the transcription factor NF-κB, thereby reducing levels of IL-6 and C-reactive protein (CRP). Furthermore, fibrates have been shown to reduce levels of fibrinogen and to increase fibrinolysis (Goldenberg et al., 2008; Moutzouri et al., 2010).

Fibrates have been in use for over four decades, with clofibrate being the first fibrate used in the clinical setting. However, a trial conducted by the World Health Organization revealed increased mortality rates in patients treated with clofibrate (Report from the Committee of Principal Investigators, 1978).

Although all fibrates have been shown to increase HDL-C significantly, their beneficial effect on all-cause mortality and cardiac mortality remains controversial (Saha et al., 2007).

In the Helsinki Heart Study, a placebo-controlled study including over 4000 men, gemfibrozil was able to raise HDL-C levels by 11%. The incidence of coronary heart disease (CHD) after 5 years was reduced by 34%, non-fatal MI was reduced by 37%. All-cause mortality remained unchanged after 5 years as well as after 18 years of follow-up (Frick et al., 1987; Tenkanen et al., 2006). In the Veterans Affairs HDL Intervention Trial, 2531 men with established CHD and low HDL-C (<400 mg·L⁻¹) were randomized to gemfibrozil or placebo. HDL-C levels were increased by 6%, but more strikingly the risk for non-fatal MI or CHD mortality decreased by 22% (Rubins et al., 1999).

In the Bezafibrate Infarction Prevention study, 3090 patients were randomized to bezafibrate or placebo. HDL-C increased by 14% in those receiving bezafibrate but the coronary event rate could not be reduced significantly. Post hoc analysis, however, revealed that in some subgroups bezafibrate therapy could influence the outcome beneficially: patients with high baseline triglyceride levels (>2000 mg·L⁻¹) showed a 39.2% reduction in the cumulative probability of the primary end points CHD mortality and non-fatal MI. In a long-term follow-up, cardiac mortality was significantly reduced dependent on the magnitude of change in HDL-C levels. Especially patients with metabolic syndrome benefited from bezafibrate showing a 56% reduction of cardiac mortality during an 8 year follow-up. In addition, bezafibrate was able to significantly delay onset of diabetes (Goldenberg et al., 2008).

The newest fibrate to be admitted to the market, fenofibrate, has been studied extensively. In the largest trial, theFIELD trial, 9795 diabetic patients were randomized to micronized fenofibrate or placebo. HDL-C increased only moderately in those treated with fenofibrate, albeit the difference to those receiving placebo was significant. The investigators of the FIELD study could detect a significant decrease of non-fatal MI in fenofibrate-treated patients, but the primary composite end point, CHD death, was not different in the two groups. However, total cardiovascular disease events decreased significantly in the fenofibrate treatment arm (Keech et al., 2005; Abourbih et al., 2009; Moutzouri et al., 2010).

Recently, results from the ACCORD lipid study led to similar conclusions; 5518 patients with type 2 diabetes, who were treated with open-label simvastatin, were assigned to either fenofibrate or placebo. After a mean follow-up of 4.7 years, no significant difference in secondary outcome (non-fatal MI, non-fatal stroke, cardiovascular death) could be observed. Consequently, the routine use of fenofibrate in combination with simvastatin to reduce cardiovascular risk in high-risk patients with type 2 diabetes could not be recommended by the ACCORD study group. (Ginsberg et al., 2010; Tenenbaum and Fisman, 2010)

In conclusion, different types of fibrates have been under evaluation up until now. While some were shown to be potent agents in the treatment for dyslipidaemia in cardiovascular disease, others failed to do so and cannot be recommended in standard care. In general, fibrates were shown to be more potent in triglyceride reduction than in alteration of HDL-C and LDL-C levels. Therefore, fibrates should mostly be used in treatment of patients with hypertriglyceridaemia and low HDL-C levels, which is a common feature in type 2 diabetes and the metabolic syndrome (Figure 1).

Figure 1.

An overview of the HDL life cycle and where the therapeutic options mentioned earlier come into action. Statins raise HDL-C levels mainly by inhibiting CETP. Fibrates induce RCT via ABCA1 and ABCG1 up-regulation, inhibit CETP and induce ApoA-I production. Niacin induces RCT via ABCA1 and ABCG1, block CETP and inhibit clearance of HDL particles in the liver. PPARα/γ/δ agonists mainly raise HDL-C through ABCA1 and ABCG1 and consecutive induction of RCT. LXR agonists activate RCT via ABCA1 and ABCG1, promote transport of cholesterol esters from the liver to the intestines, inhibit intestinal cholesterol absorption and up-regulate excretion of cholesterol.

Future therapeutic options

PPAR agonists

PPAR-α, -γ and -δ play essential roles in glucose and lipid metabolism. Regarded as ‘master’ transcriptional regulators of nutrient metabolism, these three nuclear receptors form heterodimers with the retinoid X-receptor (RXR) and bind to peroxisome proliferator responsive elements (Berger et al., 2005; Ogata et al., 2009).

PPAR-α is found mainly in the liver, kidney, heart and skeletal muscle. It up-regulates genes responsible for the oxidation of fatty acids, lipolysis and HDL metabolism while down-regulating VLDL synthesis and cholesterol esterification and inhibiting inflammation (Ogata et al., 2009).

PPAR-γ is present in adipocytes, skeletal muscle, the heart and monocytes. It plays an important role in adipocyte differentiation and lipid storage and improves insulin sensitivity. PPAR-γ activation increases lipid uptake into the adipose tissue, thus attenuating lipolysis and FFA release leading to a decline in circulating FFAs. Activation also leads to reduced pro-inflammatory cytokine and chemokine levels. In macrophages activation induces lipid efflux while in vitro vascular smooth muscle cell proliferation was blocked and apoptosis was increased. (Berger et al., 2005; Ogata et al., 2009)

The mechanism of action for PPAR-δ is less elucidated. Expressed ubiquitously, this receptor seems to protect against hyperlipidaemia and obesity while also acting anti-inflammatorily (Berger et al., 2005; Ogata et al., 2009).

All three receptors increase transcription of ABCA1, which mediates the efflux of cholesterol from cells to nascent HDL particles (Ogata et al., 2009).

Interestingly, all PPARs are dependent on the PPAR-α promoter activity and the liver X-receptor (LXR) to exert their actions. siRNA mediated downregulation of LXRα was shown to attenuate any effects of PPAR agonists. In PPAR-α (–/–) cells, the effects of PPAR agonists were reduced while the effect was completely abolished when the PPAR-α promoter was mutated (Ogata et al., 2009).

There are several drugs acting as PPAR agonists to increase HDL-C. Some of them are in use for decades such as fibric acid derivatives while the effect of others such as thiazolidinediones (TZDs) or dual PPAR-α/γ agonists, PPAR-δ agonists and novel PPAR-α agonists are controversial or are currently under investigation, respectively.

PPAR-α agonists

In recent years the mechanisms of action and the binding specificities of fibrates to PPARs have been uncovered. While fibrates mainly bind to PPAR-α, most fibrates were also found to bind to at least PPAR-γ as well, which might explain the delay of onset of diabetes caused by bezafibrate in the BIP study (via the insulin sensitizing ability of PPAR-γ binding; Goldenberg et al., 2009; Tenenbaum and Fisman, 2010).

Fibrates have been in use for several decades now. Therefore, numerous studies dealt with the mechanisms of action and the effects of fibrate therapy. Thus the effects of fibrates are discussed in a separate chapter of this review (see earlier).

Several novel PPAR-α agonists were discovered or designed that exhibit a greater PPAR-α binding potency and specificity than fibrates. Potent agonists such as LY518674 or CP-778 875 could lower TG levels and raise HDL-C levels in humans in lower doses than fibrates. The highly selective agonists BMS-687453 and BMS-711939 were also able to raise HDL-C and ApoA-I levels in human ApoA-I transgenic mice as well as lower TG and LDL-C levels in fat-fed hamsters. While the LDL-C lowering ability was greater than that of fenofibrate, the influences on TG and HDL-C levels were comparable with fenofibrate (Mukherjee et al., 2008).

In diabetic cynomolgus monkeys, the potent agonist CP-900691 not only improved plasma lipid levels significantly by lowering TG, VLDL-C and LDL-C levels, increasing HDL-C and ApoA-I levels, but also ameliorated insulin sensitivity, reduced body weight and CRP levels and raised adiponectin levels. While all other aforementioned novel agonists were not superior to fibrate controls in animals and humans, CP-900691 could be an exciting alternative by tackling both, the glycaemic and the lipidaemic status simultaneously (Wagner et al., 2010).

TZDs

The two TZDs currently on the market, rosiglitazone (except in Europe) and pioglitazone, are mainly agonists of PPAR-γ, although pioglitazone and even rosiglitazone are said to bind to PPAR-α as well, and are mostly recognized for their ability to increase insulin sensitivity. Rosiglitazone has been taken from the European market because its use was associated with an increased risk for MI (Nissen et al., 2007a; Shah et al., 2010).

The beneficial effect of TZDs on the lipid profile and on HDL-C levels is controversial. While there are studies showing an HDL-C increase of 14–18% in diabetic populations treated with TZDs, other studies only show a marginal or even adverse effect on lipid parameters. Pioglitazone has been shown to increase ApoA-I levels and rosiglitazone increases ApoA-II levels, whereby an increase of ApoA-I levels is controversial. However, the mechanisms by which these glitazones are able to influence HDL-C levels are not completely understood yet. (van Wijk et al., 2003; Carreon-Torres et al., 2009; Millar et al., 2010)

Dual PPAR-α/γ agonists

PPAR-α agonists are known to influence the lipid profile beneficially and PPAR-γ agonists improve insulin sensitivity. Thus, there have been attempts to design new, dual PPAR agonists that combine both effects to treat patients more effectively, because many diabetic patients also suffer from dyslipidaemia and dyslipidemic patients frequently develop type 2 diabetes or at least an impaired glucose tolerance.

New glitazars like tesaglitazar and muraglitazar bind to both PPAR-α and PPAR-γ and did show favourable effects in animal models increasing insulin sensitivity and improving dyslipidaemia. Both drugs even entered phase III trials, but both were withdrawn due to serious patient safety reasons as the risk for serious adverse cardiovascular events was increased in the respective studies. Other dual PPAR-α/γ agonists are being tested for efficacy and safety in animals and a new glitazone, namely aleglitazar, is even being tested in a phase II trial. However, it seems as if the activation of PPARs and in particular the balance between PPAR-α and PPAR-γ binding, is more delicate than expected (Nissen et al., 2005; Tenenbaum et al., 2005; Kendall et al., 2006; Zadelaar et al., 2006; Kim et al., 2008; Casimiro-Garcia et al., 2009; Henry et al., 2009; van der Hoorn et al., 2009; Hansen et al., 2011).

PPAR-δ agonists

One of the key components of PPAR-δ activation is the induction of skeletal muscle fatty acid oxidation, which can also be observed in response to physical exercise. Physical exercise leads to PPAR-δ up-regulation and subsequently to increased lipolysis and activity of SR-B1 and ABCA1. Thus, it can be hypothesized that upon PPAR-δ activation TG should decrease while HDL-C levels should increase. Several PPAR-δ agonists were designed and tested in recent years but the effect on HDL-C levels remained controversial. In a study with human subjects, the agonist GW501516 was shown to decrease TG and increase HDL-C levels. These effects were comparable with those seen in animal models with this agonist (Sprecher et al., 2007).

In another study with moderately overweight individuals, the same agonist could not raise HDL-C levels significantly; however, fasting and postprandial plasma TG, LDL-C and ApoB levels were decreased significantly and liver fat content could also be lowered. (Riserus et al., 2008)

The agonist GW0742 did influence neither HDL-C production nor catabolism but promoted macrophage-to-feces reverse cholesterol transport via the reduction of Niemann–Pick C1 like protein 1 (NPC1L1) expression (Briand et al., 2009).

CETP Inhibitors

CETP is a high molecular weight protein secreted by the liver and adipose tissue and is mainly found in association with HDL. CETP stimulates the transfer of cholesterol ester from HDL to TG-rich lipoproteins (TRL), such as VLDL and LDL. It is hypothesized that this cholesterol increase in TRL leads to enhanced cholesterol deposition into the peripheral arterial wall. Furthermore, CETP transfers triglycerides from VLDL or LDL to HDL. (Tall, 1993) TG-rich HDL particles get more easily hydrolysed by hepatic lipase, which leads to a destabilization of HDL and decreasing plasma concentrations due to enhanced renal catabolism. (Lamarche et al., 1999) Consequently, many scientists have considered CETP to be a culprit in dyslipidaemia and atherogenesis. This hypothesis was further supported by findings regarding altered CETP activity and CETP serum levels in these diseases.

CETP activity was shown to be elevated in CAD, in diabetes and in the metabolic syndrome. (Hibino et al., 1996; Riemens et al., 1998; Gomez Rosso et al., 2008; Park et al., 2010) Gene mutations in CETP leading to CETP deficiency were shown to result in increased HDL-C and ApoA-I lipoprotein as well as decreased LDL-C levels. (Inazu et al., 1990) The Framingham Offspring Study examined the TaqI polymorphism leading to decreased CETP activity. The phenotype leading to significantly lower CETP activity, lower CETP levels and higher HDL-C levels, could be associated with a decreased risk for CHD, although statistical significance could not be sustained after adjustment for common cardiovascular risk factors and was only seen in men (Ordovas et al., 2000). Two years later, VA-HIT presented similar results (Brousseau et al., 2002). CETP serum levels were also associated with atherosclerosis and were able to predict future CAD in apparently healthy men and women with high TG concentrations. (Boekholdt et al., 2004) In subjects with familial hypercholesterolaemia, CETP levels were positively associated with progression of atherosclerosis, determined by IMT. In this same study, statins were shown to decrease CETP and atherogenic lipid levels. Still, in a subgroup of subjects with high CETP, statins were not able to sufficiently normalize lipid abnormalities (de Grooth et al., 2004). These results illustrate the need for medication affecting CETP activity specifically.

Several strategies have been pursued in order to counteract CETP, including vaccination(Rittershaus et al., 2000), antisense deoxynucleotides (Sugano et al., 1998) and small molecule inhibitors of CETP. Three compounds have found their way into clinical trials so far, namely torcetrapib, dalcetrapib and anacetrapib.

Despite promising results in phase I and II studies (Brousseau et al., 2004; Davidson et al., 2006), in the large-scale phase III study, ILLUMINATE, torcetrapib therapy led to an increase in all-cause mortality and cardiovascular events, although it was able to increase HDL-C by 72% and decrease LDL-C by 25% (Barter et al., 2007). Correspondent results came from several imaging studies, in which torcetrapib failed to halt atherosclerosis progression in patients with mixed dyslipidaemia as well as in patients with coronary disease. (Bots et al., 2007; Nissen et al., 2007b) In patients with familial hypercholesterolaemia, torcetrapib therapy even led to a significant progression in IMT of the common carotid artery compared with the control group (Kastelein et al., 2007). As increases in systolic blood pressure, a decrease in serum potassium and an increase in serum sodium, bicarbonate and aldosterone were observed in these trials, subsequent analyses addressed the question whether the negative outcome results were due to CETP inhibition or compound-specific off-target effects. The conclusion was drawn that the off-target toxicity contributed to the observed adverse events and that CETP inhibition, by means of raising HDL-C and lowering LDL-C levels, had no impact (Kastelein, 2007; Barter, 2009). Finally, the concept of CETP inhibition was not dropped, and new compounds have been designed that differ from torcetrapib in certain molecular structures, thus avoiding off-target effects.

Dalcetrapib (JTT-705) was tested in patients with dyslipidaemia, CAD or CAD risk equivalent in several phase II studies. Dalcetrapib was generally well tolerated and successfully increased HDL-C with no clinically relevant changes in blood pressure, laboratory parameters including aldosterone, or electrocardiograms (Stein et al., 2009; 2010). Results from Dal-PLAQUE were published in September 2011. In this phase IIb, study with a follow-up of 24 months, 130 patients with CAD or at high risk for CAD were given either 600 mg of dalcetrapib or placebo. Primary end points were MRI-assessed structural changes in the arterial wall or (¹⁸FDG)PET/CT-assessed changes in arterial inflammation. Under dalcetrapib, HDL-C increased by 31% and CETP activity decreased. In contrast to placebo use, neither significant plaque progression (at 24 months) nor an increased inflammatory response in the vessel wall (at 6 months) was seen in the dalcetrapib group. There were no significant differences in adverse events between the two groups.

Interestingly, high-sensitive CRP increased by 33% in the dalcetrapib group whereas no change was seen in the placebo group. According to the authors, this disparity between local and systemic inflammation was already observed in other studies and raises the question whether general blood biomarkers provide the same information as local imaging biomarkers. (Fayad et al., 2011)

Phase III studies are ongoing to elucidate whether dalcetrapib is able to prevent cardiovascular events and mortality. Dal-PLAQUE2 started in January 2010 and is designed to evaluate dalcetrapib's power to prevent atherosclerotic progression using IMT and intravascular ultrasound (IVUS) techniques. Follow-up is planned at 24 months (Tardif et al., 2011).

Dal-OUTCOMES, testing a dosage of 600 mg dalcetrapib in patients with recent ACS, is scheduled to be complete in 2013 and will give information about dalcetrapib's clinical efficacy to reduce coronary events in secondary prevention. (Schwartz et al., 2009)

Anacetrapib (MK-0859) was shown to be even more potent than the other CETP inhibitors. Eight weeks of anacetrapib therapy increased HDL-C by 139% and decreased LDL-C by 40%. (Bloomfield et al., 2009) There was no significant difference in blood pressure compared with placebo, and anacetrapib was well tolerated. The phase III study DEFINE will examine anacetrapib's effect in patients with CAD or CAD risk equivalents on a background statin therapy (Cannon et al., 2009; 2010).

In addition, new compounds are constantly developed and are being tested in their ability to inhibit CETP (Kuo et al., 2009; Schmeck et al., 2010; Wang et al., 2010).

HDL infusions and HDL mimetics

In contrast to increasing HDL-C indirectly by interfering in the HDL metabolism, the concept was developed to directly increase HDL-C, by infusing reconstituted or recombinant HDL particles into the circulation.

One of these compounds is recombinant ApoA-IMilano. ApoA-IMilano is an apolipoprotein variant that was naturally found in inhabitants of a little village in Northern Italy. Despite strikingly low HDL-C concentrations, it was quite suprising that the carriers showed no increased risk for atherosclerosis (Franceschini et al., 1985; Sirtori et al., 2001). After recombinant ApoA-IMilano has proved itself safe and efficient in several in vitro and in vivo models (Shah et al., 1998; 2001; Chiesa et al., 2002), results of a randomized, placebo-controlled IVUS trial on a recombinant ApoA-IMilano/phospholipid complex, called ETC-216, in patients with recent ACS were published in 2003. Although the authors reported a significant regression of coronary atherosclerosis, it is noteworthy that this significance was only calculated within groups, without comparing the atheroma changes in the treated versus the placebo group (Nissen et al., 2003). A few years later Nicholls et al. (2006) presented similar results. Another reconstituted HDL compound is CSL-111, consisting of ApoA-I from human plasma combined with soybean phosphatidylcholine. Four weekly infusions of CSL-111 in 111 patients proved to be well tolerated and, compared with baseline, there was significant lesion regression. However, there was no significant reduction in atheroma volume, as measured by IVUS, compared with the placebo group. (Tardif et al., 2007)

Soon after the ILLUMINATE results were published, the scientific community started to wonder whether the sole therapeutic increase in HDL-C was enough to prevent atherosclerosis. As torcetrapib failed to have a beneficial impact on lesions despite significant changes in HDL-C concentrations, it was suggested that HDL quality, that is function, was more important than quantity. Human ApoA-I consists of 243 amino acids and is organized in a helix bundle domain comprising the N-terminal, central α-helices and a strongly lipid binding C-terminal domain (Vedhachalam et al., 2004). Laboratories have designed short peptides that do not share sequence homology, but mimic ApoA-I function. As the first mimetics were able to bind lipids, but failed to change atherosclerotic lesions, emphasis was set on finding new mimetics with preserved anti-inflammatory properties (Navab et al., 2010). The peptide 4F is the most promising mimetic so far. The 4F peptide synthesized from d-amino acids (D-4F) proved stable in the circulation in vivo. In ApoE-null mice, oral D-4F rendered HDL anti-inflammatory, enhanced reverse cholesterol transport from macrophages as much as the formation of pre-β HDL (Navab et al., 2004). A dose-dependent formation of pre-β HDL particles could also be confirmed in human plasma (Troutt et al., 2008). Recently, an in vitro study confirmed 4F's anti-inflammatory properties in human cells. 4F was shown to induce cholesterol efflux in human monocyte-derived macrophages, significantly reduced LPS-triggered production of pro-inflammatory cytokines, and significantly decreased monocyte adhesion to human endothelial cells as well as transendothelial migration (Smythies et al., 2010). It was, however, not evident if these anti-inflammatory actions would also affect the development of atherosclerotic lesions. Navab et al. administered 4F orally to LDL receptor-null mice on a Western diet. Lesions decreased by 79% and LDL was protected from oxidation (Navab et al., 2002). In rabbits, D-4F also induced a significant regression of atherosclerotic lesions and improved the HDL-inflammatory index. Interestingly, these changes were independent of TC or HDL-C levels (Van Lenten et al., 2007). Subsequently, D-4F was tested in a phase I study in patients with CHD or at equivalent risk. Five groups of 8 people received 30, 100, 300, 500 mg of unformulated D-4F, or placebo, respectively. There were no alterations in lipid or lipoportein concentrations, but 4 h after administration of 300 mg and 2 h after administration of 500 mg, the HDL-inflammatory index improved significantly, suggesting an anti-inflammatory effect of D-4F in humans. (Bloedon et al., 2008)

Another ApoA-I mimetic peptide, 5A, was recently shown to inhibit inflammation and oxidative stress in a carotid artery stenosis model in rabbits. These effects were also seen in human coronary artery endothelial cells in vitro and were shown to be dependent on ABCA1 (Tabet et al., 2010).

ApoA-I Up-regulators

A novel approach to increase HDL and to improve its function is to stimulate the synthesis of ApoA-I, thus, stimulating the first step in HDL life cycle. RVX-208, a stimulator of ApoA-I gene expression is a promising compound in that respect. RVX-208 was shown to induce ApoA-I synthesis in HepG2 cells. In monkeys, it also led to an increase in serum ApoA-I and HDL-C concentrations (60 and 97%, respectively), as well as to ABCA1-, ABCG1- and SR-B1-dependent cholesterol efflux. In humans, administered orally for 1 week, RVX-208 significantly increased ApoA-I, pre-β HDL and HDL functionality (Bailey et al., 2010). However, further trials are needed to address whether the upregulation of ApoA-I by compounds, such as RVX-208, have atheroprotective effects in cardiovascular diseases.

LXR agonists

Belonging to a large family of nuclear receptors, LXRs bind to the regulatory regions of target genes and stimulate their transcription. Initially, LXRs were isolated from the cDNA library of a human liver as orphan receptors because their ligands were not identified at the time of characterization. It is now evident that oxidized cholesterol derivatives, also known as oxysterols, are specific ligands for LXRs. Currently, two LXR isoforms, LXRα and LXRβ, which share 80% homology have been characterized. The former is expressed mainly in the liver, intestine, kidney, spleen and adipose tissue, while the latter is expressed ubiquitously (Wojcicka et al., 2007).

Both receptor types form heterodimers with the RXR, which can then be activated by either LXR agonists or 9-cis retinoic acid (a specific RXR ligand) to bind to an LXR response element (LXRE) in the promoter regions of target genes (Wojcicka et al., 2007).

The primary role of LXRs is to act as intracellular cholesterol sensors. LXRs sense excess cholesterol (cholesterol and oxysterol levels correlate with each other, thus if oxysterol levels are high cholesterol levels will be high as well) and activate several mechanisms to protect the cell from cholesterol overload. LXRs are therefore able to stimulate the reverse cholesterol transport, from the removal of cholesterol from the cell, to the transport of cholesterol to the liver to its biliary excretion. Of lesser importance is the inhibition of intestinal cholesterol uptake as well as cholesterol synthesis (Wojcicka et al., 2007).

In macrophages LXRs up-regulate the expression of ABCA1 and ABCG1. ABCA1 is also involved in the formation of nascent HDL particles in the liver and small intestine. LXRs also stimulate the expression of Niemann–Pick C1 (NPC1) and C2 (NPC2), which redistribute cholesterol from the endosomal compartment to the plasma membrane. Furthermore, hepatic SR-B1, essential for the delivery of cholesterol from HDL particles to hepatocytes, is up-regulated by LXRs in vitro (Wojcicka et al., 2007; Beltowski, 2008).

LXRs also decrease the expression NPC1L1, which is present in the apical membrane of enterocytes to absorb cholesterol from the intestinal lumen. Simultaneously, ABCA1 in the small intestine is up-regulated leading to the formation of HDL particles and the transport of cholesterol to ApoA-I. Hepatic ABCG5 and ABCG8, responsible for the cholesterol transport to the bile, thus leading to an increased faecal cholesterol extraction are up-regulated by LXRs as well (Wojcicka et al., 2007; Beltowski, 2008).

However, LXRs stimulate hepatic lipogenesis leading to hepatic and plasma hypertriglyceridaemia, leading to partially severe liver steatosis and dysfunction. This effect is mediated by the sterol regulatory element binding protein-1c (SREBP-1c) Knockout studies have revealed that SREBP-1c is influenced by LXRα rather than LXRβ (Jamroz-Wisniewska et al., 2007; Wojcicka et al., 2007; Beltowski, 2008).

As of now, LXR agonists are being developed and tested in animal models for their effects. Several synthetically engineered LXR agonists have been tested extensively, namely T0901317, GW3965, AZ876 and ATI-111. They all show a high potency regarding the interaction with the receptors, but none of them show an exclusive selectivity for either of the two variants (Groot et al., 2005; Honzumi et al., 2010; Peng et al., 2011; Srivastava, 2011; van der Hoorn et al., 2011).

Based on the fact that LXRs are potent activators of RCT, one might suppose that agonists like T0901317 should increase HDL-C levels. However, LXRs and their agonists not only activate RCT, but also CETP leading to a cholesterol transfer from HDL to LDL and VLDL particles. Therefore, studies regarding the influence of LXR agonists on the lipid profile have to be interpreted with care because several studies were performed in mice, which apparently lack CETP. Thus, T0901317 is able to elevate HDL-C levels in wild-type mice while CETP transgenic C57BL/6J mice and cynomolgus monkeys show an increase in non-HDL cholesterol levels and decreased HDL-C levels upon treatment with T0901317 (Jamroz-Wisniewska et al., 2007; Wojcicka et al., 2007; Beltowski, 2008; Honzumi et al., 2010).

An earlier study showed similar results in male Syrian gold hamsters treated with either GW3965 or SB742881, respectively. In such treated hamsters, LDL-C and VLDL-C levels rose significantly while HDL-C levels decreased significantly. Also, TG levels increased considerably (Groot et al., 2005).

ATI-111 an LXR agonist showing preference for the LXRα isoform and higher potency than T0901317 was shown to not increase TG, VLDL-C or LDL-C levels in male LDLR–/– mice after 8 weeks of treatment. However, HDL-C levels also did not rise significantly (Peng et al., 2011).

A study in APOE*3 Leiden mice evaluating the effects of the LXR agonist AZ876 could show a decreased TC level while HDL-C levels increased. However, TG levels also rose considerably leading to a steatosis of the liver and significant liver dysfunction. (van der Hoorn et al., 2011)

Considering all these results, the effect of LXR agonists on HDL-C plasma levels seems quite disappointing. Moreover, considering the dramatic rise of TG levels after treatment with some LXR agonists their usefulness regarding the lipid profile modifying effect has to be evaluated quite critically. However, the same studies showing no or a negligible positive effect on plasma lipid levels in general and HDL-C plasma levels could specifically show a clear anti-atherogenic and anti-inflammatory effect.

LXR agonists were shown to decrease atherosclerotic plaque size in mouse models (T0901317, GW3965, AZ876, ATI-111) significantly. Some of them were able to reduce the number of lesions (AZ876, GW3965), alter the composition of plaques towards more stable lesions (T09013117) or even prevent lesion development (AZ876, GW3965) (Jamroz-Wisniewska et al., 2007; Beltowski, 2008; Peng et al., 2011; Srivastava, 2011; van der Hoorn et al., 2011).

In mice and rats administration of T0901317 or GW3965 led to decreased levels of inflammatory cytokines and other key mediators in the development of atherosclerosis such as TNF-α, IL-1β, IL-6, IFN-γ, MMP-9 and ICAM-1 (Jamroz-Wisniewska et al., 2007; Beltowski, 2008; van der Hoorn et al., 2011).

Interestingly, tri-butyltin chloride, an organotin widely used in agricultural and chemical industries and known to induce adipogenesis and reduce fertility in mammals, was shown to induce ABCA1 expression and ApoA-I-mediated cellular cholesterol efflux via the activation of the LXRα/RXR heterodimer (Cui et al., 2011).

While LXR agonists do not seem to be suitable to increase HDL-C levels they could represent an exciting class of new anti-atherosclerotic drugs as they were shown to in part lower atherosclerotic plaque size dramatically. Some of these molecules were even able to completely prevent the formation of atherosclerotic lesions. However, these studies were conducted in animal models. The next step has to be to try to translate these highly interesting findings into the clinical setting. It should also be noted that LXRα and LXRβ use different signalling pathways (Jamroz-Wisniewska et al., 2007; Wojcicka et al., 2007). Currently, no exclusively specific agonists for either isoform of the receptor are available so that unwanted side effects of LXR signalling such as liver steatosis jeopardize the beneficial anti-inflammatory and anti-atherosclerotic effects of these compounds. Thus, the design of drugs activating either LXR isoform exclusively or drugs displaying tissue selectivity would provide exciting opportunities for the treatment of patients (Figure 1).

Conclusion

The Framingham Study was the first to suggest that positively influencing HDL-C levels may lead to a decrease in cardiovascular mortality (Gordon et al., 1977). Since then, other studies have confirmed this statement. Currently, optimal statin treatment manages to achieve a risk reduction for cardiovascular mortality of about 25–35% (Boden et al., 2011; Brooks et al., 2010; Davidson and Rosenson, 2009). Considering that cardiovascular disease can be held accountable for about 50% of all deaths a risk reduction of 25–35% cannot be enough (Greenow et al., 2005).

Raising HDL-C was seen as a viable and promising way to further reduce the risk of cardiovascular mortality. However, things are not as easy as they seem with HDL-C (Asztalos et al., 2006; Freund et al., 1993; Gordon et al., 1977; 1989; Kannel and McGee, 1985; Wilson et al., 1988).

Firstly, raising HDL-C was proven to be a daunting task. Of the currently available drugs only niacin has the potential to raise HDL-C levels substantially (approximately 25%) (Carlson, 2005; Brooks et al., 2010). Dietary and lifestyle changes as well as statin therapy can help raise HDL-C levels by 5–10% (Downs et al., 1998; Streja et al., 2002; Varady and Jones, 2005; Nissen et al., 2006; Hausenloy and Yellon, 2009). Fibrate treatment manages to raise HDL-C by 10–15% (Frick et al., 1987; Goldenberg et al., 2008). But studies examining the effect of fibrates on cardiovascular mortality yielded disappointing results (Keech et al., 2005; Abourbih et al., 2009; Ginsberg et al., 2010; Tenenbaum and Fisman, 2010).

Niacin seems to be neglected in dyslipidaemia treatment because it frequently provokes flushing. This side effect is certainly stressful for patients, but it is harmless. However, while earlier studies indicated that niacin had a favourable effect on cardiovascular mortality, the recently published results of AIM-HIGH were sobering at first. AIM-HIGH could not show any benefit for niacin treatment in combination with statin therapy. But, as outlined by Nicholls, the results of this study have to be interpreted with caution (Nicholls, 2012).

The verdict on niacin is still open at least until the results of the HPS2-THRIVE study are published.

Secondly, there has been increasing evidence that the quality and not the quantity of HDL is important. HDL particles are in large part responsible for reverse cholesterol transport, the mechanism that can reverse atheroma formation. HDL particles take up cholesterol from foam cells and thus reduce the risk of cholesterol oxidation and further foam cell formation. This way, the vicious circle can be broken and arterial wall inflammation is halted. HDL particles are also able to normalize endothelial cell function, inducing nitric oxide (NO) production (leading to vasodilation) and inhibiting inflammation, chemotaxis and thrombosis. So, to influence atherosclerosis proper HDL particle function is of utmost importance. (Tall, 2008; Singh et al., 2010)

Studies showed that in a state of chronic inflammation (as is the case with atherosclerosis, chronic renal disease, diabetes or several forms of arthritis), reverse cholesterol transport is interrupted. Furthermore, HDL particles do not act as anti-oxidative agents, they rather support oxidation of cholesterol, fuelling the inflammatory process. Moreover, other pleiotropic effects of HDL are hampered (Natarajan et al., 2010; Saemann et al., 2010).

One study could show that when the protease chymase is present, reverse cholesterol transport via ABCA1 is severely impaired, thus indicating that the altered environment in an atherosclerotic vessel wall may severely impact the functionality of HDL particles (Favari et al., 2004).

ApoA-I is a specific target for myeloperoxidase (MPO) an agent found mainly in neutrophils and monocytes and playing an important role in microorganism eradication. MPO, which is found in high amounts in human atheroma lesions is able to alter the protein structure of HDL particles rendering them less able to remove cholesterol from macrophages. MPO was even able to alter the functionality of reconstituted HDL (Smith, 2010a,b). Considering the fact that statins (rosuvastatin in particular) are able to lower MPO levels in vivo their role in HDL metabolism seems to be preserving HDL function by creating a less inflamed environment rather than to simply increase HDL-C levels (Andreou et al., 2010).

Drugs like torcetrapib (ILLUSTRATE, ILLUMINATE, RADIANCE) and niacin (AIM-HIGH) could raise HDL-C levels by 25–60%. (Barter, 2009) Still, no change in cardiovascular mortality was detected. In addition the atheroma size was not affected which could support the notion that raising HDL-C levels may not be as beneficial as once thought. (Singh et al., 2010) One meta-regression analysis, studying the association between change in HDL-C levels and CAD morbidity and mortality could not even detect a significant reduction of CAD risk upon elevation of HDL-C levels while the reduction of LDL-C levels had significant beneficial effect (Briel et al., 2009).

The proper function of HDL particles seems to be the crucial factor leading to the favourable effects. Supporting this hypothesis is the fact that people expressing ApoA-IMilano have low levels of HDL-C, yet are not more at risk of suffering from cardiovascular disease than the normal population. ApoA-IMilano infusions much rather helped lower the atheroma burden via reverse cholesterol transport. (Joy and Hegele, 2008; Tall, 2008; Vergeer et al., 2010)

Thirdly, HDL-C may be a suboptimal parameter for assessing cardiovascular risk. Several studies have shown that a low HDL-C level is a risk factor for cardiovascular disease. But HDL-C levels, which are acquired by ultracentrifugation determining the amount of cholesterol in HDL particles per 100 mL of plasma, give no hint on the composition of HDL particles nor their functionality (deGoma et al., 2008). HDL particles differ in their ways of action (e.g. large cholesterol rich particles activate reverse cholesterol transport via SR-B1, while the smaller preβ-HDL particles use ABCA1). Simply measuring HDL-C levels does not provide enough information on an otherwise highly complex and dynamic system. New, inexpensive and easily applicable assays are needed to assess the functional capacity of HDL particles in cardiovascular disease patients to better understand the pathophysiology of atherosclerosis and to be able to identify the specific therapeutic needs for every patient (deGoma et al., 2008; Vergeer et al., 2010). Interestingly, study groups were recently able to show how drug treatment is able to alter HDL composition. Sorrentino et al. published results of the effect of extended-release niacin therapy on HDL in diabetic patients. Three months of niacin (1500 mg daily) versus placebo led to a significant improvement of HDL function in the drug-treated group. Niacin stimulated endothelial NO production, endothelial progenitor cell-mediated endothelial repair and reduced superoxide production, all properties which are impaired in diabetics compared with healthy controls (Sorrentino et al., 2010). In CAD subjects, 1 year of extended-release niacin (2000 mg daily) plus atorvastatin therapy modified HDL in a way that protein composition resembled more closely that of healthy control subjects indicating improved RCT (Green et al., 2008).

Page et al. found out that mifepristone, a glucocorticoid and progestin antagonist currently under investigation for the treatment of Cushing's syndrome, did lower HDL-C levels in 30 healthy postmenopausal females by approximately 25%, but that HDL-C mediated cholesterol efflux from cultured macrophages did only decrease by 12%, thus supporting the hypothesis that HDL-C levels are not necessarily directly correlated to HDL particle function (Page et al., 2012).

With the predictive ability of HDL-C levels and the efficacy of HDL-C raising drugs being critically discussed some new parameters were presented that might be better suited to predict cardiovascular mortality (Table 1).

Table 1.

List of the relevant established HDL-C raising substance classes and their therapeutic effects

On the one hand, the ratio of ApoB/ApoA-I has been proposed to correlate well, even better than LDL-C levels, with cardiovascular risk. Several studies (AMORIS, INTERHEART) could show that the ratio of ApoB/ApoA-I is better suited to predict cardiovascular risk than LDL-C levels with the risk increasing the higher the ratio is (Yusuf et al., 2004; Walldius and Jungner, 2006).

Another molecule worth noting as a potential biomarker and risk factor in cardiovascular disease is lipoprotein (a) [Lp(a) ]. Consisting of a cholesterol rich LDL particle with one molecule of apolipoprotein B100 and one molecule of apolipoprotein A this molecule acts prothrombotic as it shares structural homology with plasminogen and plasmin without their fibrinolytic activity. Additionally, it nurtures atherosclerotic plaques with the deposition of Lp(a)-cholesterol. Elevated Lp(a) levels have been shown to correlate with increased cardiovascular risk being independent from levels of other lipid molecules or age. Statins and niacin are able to lower Lp(a) levels significantly, however, no randomized, controlled intervention trials selectively aiming at reducing CAD via the reduction of Lp(a) levels have been designed. Such studies will be necessary to recognize and accept Lp(a) as a potential risk factor which can be targeted in the treatment of cardiovascular disease. (Nordestgaard et al., 2010)

Finally, the quest for further lowering the risk for cardiovascular mortality beyond the use of statins and the reduction of LDL-C levels has lead to an intense examination of the role of HDL particles in the biology of atherosclerosis. As a result, the mechanisms of established drugs, which induce an increase of HDL-C levels, were discovered and new drugs that aim for higher HDL-C levels were designed.

In recent years, the importance of HDL-C levels and HDL particles in atherosclerosis has often been questioned. Increase of HDL-C levels has led to conflicting results regarding cardiovascular risk. New therapeutic options are being developed and analysed for their therapeutic potential.

Additionally, there are new biomarkers and risk factors being proposed for future use and targeting.

So, while raising HDL-C has not yet met the initially lofty expectations it still should be a viable therapeutic option for those remaining at a high cardiovascular risk especially because the results of the Framingham study and others that followed cannot be ignored.

Glossary

ABCA

ATP-binding cassette transporter A

ABCG

ATP-binding cassette transporter G

ApoA-I

apolipoprotein A-I

ApoB

apolipoprotein B

ApoC

apolipoprotein C

ApoC-III

apolipoprotein C-III

ApoE

apolipoprotein E

CAD

coronary artery disease

CETP

cholesterol ester transfer protein

CHD

coronary heart disease

CRP

C-reactive protein

FFA

free fatty acid

HDL(-C)

high-density lipoprotein(-cholesterol)

ICAM-1

intercellular cell adhesion molecule-1

IMT

intima-media thickness

IVUS

intravascular ultrasound

LCAT

lecithin-cholesterol acyltransferase

LDL(-C)

low-density lipoprotein (-cholesterol)

Lp(a)

lipoprotein (a)

LXR

liver X-receptor

LXRE

liver X-receptor response element

MCP-1

monocyte chemoattractant protein-1

MI

myocardial infarction

MPO

myeloperoxidase

NPC1

Niemann–Pick C1

NPC1L1

Niemann–Pick C1 like protein 1

oxLDL

oxidized low-density lipoprotein

PD1.prostaglandin

D2 receptor subtype 1

PON1

paraoxonase-1

RCT

reverse cholesterol transfer

rHDL

reconstituted high-density lipoprotein

RXR

retinoid X-receptor

SR-BI

scavenger-receptor B-I

TC

total cholesterol

TG

triglyceride

TRL.TG-rich

lipoprotein

TZD

thiazolidinediones

VCAM-1

vascular cell adhesion molecule-1

VLDL(-C)

very low-density lipoprotein (-cholesterol)

Conflicts of interest

None.