Atherosclerotic cardiovascular disease (CVD) is a global epidemic. As life expectancy has increased, chronic degenerative diseases have become a significant healthcare burden, affecting developed and developing countries alike,1 with atherosclerotic disease the leading cause of death according to the WHO.2 The development of effective strategies for the prevention of cardiovascular events at individual and population levels should reduce premature morbidity and mortality and have a significant global impact.1 To manage cardiovascular risk, it is prudent to understand the pathophysiological mechanisms of atherosclerosis and target the modifiable factors that contribute to CVD. Traditional risk factors for CVD include age, male gender, smoking, blood pressure, diabetes and cholesterol. These factors contribute to CVD risk through a complex interplay of many pathophysiological mechanisms, including endothelial injury, oxidative stress and induction of inflammatory pathways. Cholesterol is a key driving force involved in initiating vascular dysfunction and inflammation, an essential component of atherosclerotic plaque and thus an attractive and important target for disease prevention.3 The balance between cholesterol entry to, and removal from, the subintima is mediated by the interaction between cells in the vessel wall and lipid carriers called lipoproteins, which are highly relevant in the pathophysiology of atherosclerosis.
We hope this article will provide an up-to-date and comprehensive assessment of the role of high-density lipoprotein (HDL) in CVD for practising clinicians. This article will provide a summary, rather than a detailed biochemical examination, of the aspects of HDL biology that are relevant to atherogenesis. We will also focus on the relevance and utility of the measurement of HDL in patients with, and at risk of, CVD, particularly with regard to the estimation of cardiovascular risk and the current evidence base for HDL as a therapeutic target.
High-density Lipoprotein Biology
Atherosclerosis is a chronic inflammatory process affecting the vasculature, characterised by lipid deposition and formation of plaques within the arterial wall.4 Cholesterol transport to and from the arterial intima is key to this process; net deposition of cholesterol in excess of its removal from the arterial wall helps create a local environment that favours lipid oxidation and uptake into cells of the monocyte/macrophage lineage; this in turn drives inflammation and atherosclerosis via the release of inflammatory mediators and attraction of further inflammatory cells, together with endothelial activation and dysfunction.3
Lipoproteins are water-soluble structures that bind fat-soluble cholesterol and triglyceride (TG), thus enabling lipid carriage within the bloodstream. The lipoprotein family has been described and characterised by the size, density and electrophoretic mobility of its different subtypes.5 Those include large, TG-rich chylomicrons, very-low-density lipoproteins (VLDLs), intermediate-density lipoproteins (IDLs), low-density lipoproteins (LDLs) and HDLs.
LDL particles and, to a lesser extent, IDL, VLDL and chylomicron remnants provide the supply of cholesterol that enters the vascular wall and is ‘captured’ through interaction between apoprotein-B (Apo-B) and the scavenger receptor on the surface of monocytic lineage cells. These become activated and transform into macrophages and eventually into the cholesterol-laden foam cells, which further influence vascular inflammation and are the hallmark of the atherosclerotic plaque. In contrast, HDL mediates removal of cholesterol from peripheral tissues, primarily via interaction between apoprotein-A1 (Apo-A1) on smaller, discoid ‘pre-β’ HDL particles and the adenosine triphosphate-binding cassette transporter-A1 (ABCA1) cholesterol transporter on the cell surface, thus initiating the process of reverse cholesterol transport. Several other factors promote the flow of cholesterol through HDL in this process, including lecithin cholesterol acyltransferase (LCAT), phospholipase transfer protein (PLTP) and cholesteryl ester transfer protein (CETP), with factors such as scavenger receptor-B1 (SR-B1) and the adenosine triphosphate-binding cassette-G1 (ABCG1) transporter also facilitating transfer of additional cholesterol from peripheral tissues into more mature forms of HDL.4 Thus HDL promotes efflux of cholesterol from cholesterol-rich macrophages or foam cells, eventually allowing delivery of cholesterol to the liver for elimination. This limits oxidative damage caused by macrophage activation/foam cell formation and prevents atherogenic inflammatory responses induced by oxidised LDL cholesterol (LDL-C). Indeed, adding HDL to experimental arterial wall models reduces intimal LDL-C deposition, decreasing lipid oxidation and production of pro-inflammatory and pro-atherogenic cytokines.6 Some of the biological functions and effects of HDL are shown in Figures 1–3.
HDL particles are complex in structure and composition. It is increasingly appreciated that HDL has a variety of physiological functions and is involved in the acute-phase response, complement regulation and cytokine-mediated pathways.7–9 HDL reduces expression of adhesion molecules, which has downstream consequences, including inhibition of neutrophil chemotaxis and reduction of reactive oxygen species generation within the arterial wall.8,9
HDL also acts to protect vascular endothelium in part through activation of nitric oxide-dependent pathways. Lysophospholipids carried by HDL activate endothelial nitric oxide synthase, which results in vasodilatation via nitric oxide release and relaxation of vascular smooth muscle cells within the arterial wall.3 Together with reverse cholesterol transport, these anti-inflammatory, antioxidant and vasodilatory properties of HDL may all contribute to an atheroprotective influence.3,6,8,9 The functions of HDL are summarised in Table 1.
Clinical Applications of High Density Lipoprotein
Considerable research effort has focused on establishing the mechanisms by which HDL might protect against atherosclerosis, and a large body of scientific evidence demonstrates several ways in which it may do so. However, HDL biology is not yet fully understood and the assays used to assess its multiple functions are not currently suitable for use outside the research setting. None the less, useful clinical evidence can be obtained simply by measuring HDL cholesterol (HDL-C) levels, particularly with regard to quantifying cardiovascular risk. The role of HDL as a treatment target in contemporary practice is more controversial, particularly in patients being treated with statins, as there is a relative paucity of outcome studies examining the incremental impact, on major clinical events, of the addition of agents that affect HDL to regimens that contain statins as background therapy.
Extensive research into factors associated with cardiovascular risk and the development of risk estimation systems has enabled us to identify high-risk populations who are most likely to benefit from evidence-based preventive interventions. This approach can help those responsible for developing and paying for systems of care to prioritise resource allocation most appropriately on the basis of both clinical effectiveness and cost-effectiveness; it can also limit the over-medicalisation of lower-risk individuals who are less likely to gain an appreciable clinical benefit from more intensive pharmacological approaches.10,11 Thus cardiovascular risk estimation has now become an essential component of the management of arterial disease.
High-density Lipoprotein in Cardiovascular Risk Assessment
Global Risk Assessment
Atherosclerotic CVD develops as a consequence of a complex interplay of multiple risk factors, with no single risk factor carrying enough weight in isolation for accurate prediction of risk. Therefore, models have been developed to estimate ‘global risk’ by accounting for the respective influence of those multiple risk factors. The output is typically presented as a patient’s 10-year risk of suffering a major clinical CVD event or CVD mortality, and is generally used to assess risk in individuals between the ages of 40 and 75 years. Risk estimation is not felt to be necessary in elderly (>75 years) or diabetic patients, nor in secondary prevention; these patient groups are all considered to be at high risk of cardiovascular events and should benefit from a more intensive approach to risk factor management.12
Much of our current knowledge of cardiovascular risk factors is derived from the Framingham Heart Study, which specifically identified low HDL-C as a CVD risk factor.13 Data from that study have been used to generate the Framingham Risk Score (FRS), the best-established and most widely used model, which uses systolic blood pressure, smoking status, total cholesterol (TC):HDL ratio, age, gender and the presence of diabetes mellitus and electrocardiographic left ventricular hypertrophy to estimate CVD risk.14,15
More recently, extensive European outcome data were used to generate the Systemic Coronary Risk Estimation (SCORE) system, a relative risk chart that incorporates similar variables to the Framingham model, but, in its initial iteration, only used TC levels and did not consider HDL.16 At the time, it was felt that the practical simplicity of using this single marker of lipid risk only outweighed any additional advantage conferred by knowledge of the patient’s more detailed lipid profile. However, a more recent analysis of the SCORE data suggests that the importance of HDL-C in risk estimation might have been underestimated. Inclusion of HDL values and use of the TC:HDL ratio commonly reclassifies patients into higher- or lower-risk categories, regardless of gender and baseline risk, and is likely to result in more accurate clinical decision-making; this approach has been recommended in the recent European Society of Cardiology/ European Atherosclerosis Society (ESC/EAS) lipid management guidelines and will be recommended in the forthcoming joint European CVD prevention guidance.17,18 Indeed, the US National Cholesterol Education Program (NCEP) and UK National Institute for Health and Clinical Excellence (NICE) recommend assessment of HDL-C values when calculating CVD risk.19Figure 4 illustrates the impact of variations in HDL level on the global cardiovascular mortality risk using the European SCORE model.
Risk Assessment beyond High-density Lipoprotein
Clinical practice favours the simplicity of both the Framingham and SCORE systems. However, these systems focus on traditional risk factors and do not formally allow for increased CVD risk in cases of positive family history, treated hypertension, social deprivation or obesity, which are all recognised to have an important influence on CVD risk. Obesity is becoming increasingly common, and the clustering of risk factors such as hypertension, dyslipidaemia and diabetes is commonly seen in obese individuals, particularly those with a central/visceral pattern of weight gain.21,22 This is an important factor contributing to CVD risk at individual and population levels, where an increasing prevalence of obesity is offsetting the benefits gained through improvements in CVD risk management elsewhere.23
Recently developed systems aim to improve the accuracy of risk estimation by incorporating additional risk factors. For example, the QRISK2 cardiovascular disease risk calculator developed from a large UK primary care dataset estimates CVD risk using the following additional risk factors: family history of early CVD, existing antihypertensive treatment, social deprivation, body mass index, ethnicity, presence of rheumatoid arthritis, chronic kidney disease and atrial fibrillation.24 The Reynolds Risk Score, developed in the US, estimates risk using a combination of high sensitivity C-reactive protein (CRP), non-HDL-C (TC-HDL), glycated haemoglobin (HbA1c) percentage and family history of premature myocardial infarction alongside traditional factors;25 it has been shown to provide incremental ability to predict risk compared with the FRS, but is not currently used in Europe, as it has not been validated in European populations, where routine CRP testing is not universally supported.
Although systems that incorporate additional risk factors demonstrate efficacy and can reclassify patients into higher- or lower-risk categories, they require information on a larger number of variables, or perhaps on newer variables. There are less extensive data supporting their use, somewhat limiting their utility in clinical practice. HDL is readily available on lipid profiling and can improve diagnostic accuracy without the practitioner having to request additional information, and it is likely to be both pragmatic and cost-effective to use risk scoring systems based on HDL levels.
Limitations of High-density Lipoprotein Cholesterol Levels in Risk Assessment
There has been much debate over the lipid parameters measured for the optimal assessment of CVD risk. Research has also focused on assessing the value of biomarkers for CVD, with the hope of elucidating which ones most accurately predict CVD risk.
Apo-A1 and Apo-B are the main surface proteins on LDL and HDL, respectively, and it has been argued that their levels provide a more reliable estimate of the amount of these lipoproteins than the levels of LDL-C and HDL-C and improve the accuracy of CVD risk prediction.26,27 However, an extensive recent meta-analysis suggests that lipid fractions are equally valuable in assessing risk in a general population. Clinical practice should be guided by which system is most pragmatic, cost-effective and relevant for local use.28
Lipid biology is complex, with much interplay between lipid fractions. Although low HDL and hypertriglyceridaemia are epidemiologically associated with CVD,13,29–33 debate has continued over whether both HDL and TG are independently associated with CVD following adjustment for other risk factors. Recent evidence from the Emerging Risk Factors Collaboration has reinforced information from previous studies,33 demonstrating that most of the predictive value of TG is lost following adjustment for other common risk factors. In contrast, the predictive value of HDL remains consistent following risk factor adjustment.34
It has been suggested that HDL-C may be less predictive of cardiovascular risk in patients achieving very low concentrations of LDL-C as a result of high-dose statin therapy,35,36 and recent data from the Justification for the use of statins in primary prevention: an intervention trial evaluating rosuvastatin (JUPITER) study have demonstrated that levels of HDL-C do not correlate with residual cardiovascular risk in patients with high CRP and low LDL-C receiving potent statin therapy.37 However, a recent meta-analysis of 20 randomised controlled trials, including the JUPITER study, has shown that low levels of HDL are clearly associated with CHD risk at all levels of LDL-C and that this relationship is not changed appreciably by statin therapy.38 Indeed, HDL remains an important predictor of progression of coronary atheroma, even in patients achieving very low levels of LDL on statin therapy.39,40
Currently, HDL-C together with non-HDL-C or TC within a multifactorial risk model appears to be sufficient for contemporary cardiovascular risk assessment without the need for TG values. Therefore, testing can be carried out without fasting, reducing the need for delayed or additional testing due to patient oversight.28 Apolipoproteins A and B demonstrate similar merit to HDL and non-HDL-C levels for risk prediction at a population level, although additional information might be provided by more detailed assessment of apolipoprotein levels in some individuals. Cost and practicality should guide lipid test selection for routine risk assessment, with conventional lipid profiling adequate for most individuals. Whether further functional assays of HDL might enhance the evaluation remains a subject of investigation.
Although individuals with low HDL-C frequently have dysfunctional HDL, recent evidence has demonstrated that levels alone do not always correlate with HDL function,3,41 emphasising the complex relationship that exists between HDL and atherosclerosis. Genetic and environmental host factors can influence the function of HDL – for example, certain inflammatory states reduce reverse cholesterol transport and anti-inflammatory function without altering the levels of HDL-C.42
A small number of individuals originating from the Italian village of Limone sul Garda were found to express a variant of Apo-A1 known as Apo-A1 Milano. Those with this mutation demonstrate a very low risk of ischaemic heart disease despite very low plasma levels of HDL-C.43 Although the mechanism of benefit has not yet been fully elucidated, Apo-A1 Milano appears to promote very rapid cholesterol efflux and reverse cholesterol transport in experimental models. It also has enhanced antioxidant and anti-inflammatory properties, with parenteral infusion of Apo-A1 Milano mediating plaque regression within weeks in patients with coronary atheroma.44,45
HDL function is affected by a variety of host factors, including the acute-phase response to stress and infection, rheumatological disease, diabetes mellitus and impaired glucose tolerance.42 Importantly, individuals with abdominal obesity commonly have altered HDL function46 and it appears that intensive dietary and lifestyle modification can modulate HDL-C from a pro-inflammatory to an anti-inflammatory state.47
It has been proposed that the pharmaceutical industry ought to focus on developing agents that improve HDL function and assays to measure this functionality, in addition to simply focusing on increasing levels of HDL-C.48 Given the strong independent relationship between HDL-C levels and CVD, and the beneficial impact of HDL on disease progression in animal models, therapeutics that target HDL-C ought to provide clinical benefit.28
Low High-density Lipoprotein as a Therapeutic Target – High-density Lipoprotein-modifying agents
Individuals with low levels of HDL-C, who often have elevated TG and dysfunctional HDL, have increased CVD risk even when receiving statins, particularly those with diabetes mellitus.49–51 Indeed, statins confer only a modest benefit once LDL-C levels have been reduced to very low levels, have only a modest impact on HDL levels (3–10 %); and some agents reduce serum HDL-C, suggesting that combination therapies may be of benefit.32,52
Niacin (nicotinic acid) has been used in clinical practice for many years and has notable effects on all aspects of atherogenic dyslipidaemia, with effects on HDL-C, LDL-C, TG and lipoprotein(a) [Lp(a)]. It increases HDL-C levels by 15–35 %53,54 and protects HDL from metabolic consumption by reducing the breakdown of its main structural component, Apo-A1. Despite a good safety profile, niacin commonly causes flushing in the initial stages of therapy. This may indicate good response to therapy55 and tends to settle over time, but remains problematic for many patients and is a common reason for non-compliance.53,56,57 It is otherwise well tolerated, significantly increases serum HDL-C levels and appears to result in improved patient outcomes and recovery of the atheroprotective effects of HDL.19,57
Recent evidence suggests that niacin therapy demonstrates both quantitative and qualitative effects on HDL, increasing plasma levels and improving molecule function. Trials have demonstrated that the additional use of extended-release niacin alongside conventional statin therapy improves several lipid fractions – LDL-C, non-HDL-C, TG, HDL-C, Lp(a), CRP – when compared with simvastatin monotherapy53,54,56 and, in a recent study, the administration of niacin to diabetic subjects has been shown to increase the protective effects of HDL on vascular endothelium, with improved endothelial repair, nitric oxide production and resultant endothelial-dependent vasodilatation.58
Effects of Niacin on Outcomes
In addition to improvements in lipid fractions and functionality, combination therapy using niacin has shown favourable results regarding patient outcomes. Niacin therapy (in combination) appears to improve coronary atherosclerosis on angiography (0.4 % regression) and reduces long-term mortality (11 %), cardiovascular morbidity (27 % reduced non-fatal re-infarction) and the need for surgical revascularisation (60 %) when compared with placebo.57,59–63
Niacin seems to be both efficacious and safe and is frequently considered for use in high-risk patients with dyslipidaemia (in particular low HDL-C) refractory to statin monotherapy.19 Unfortunately, the majority of studies so far have focused on statin–niacin combination therapy versus placebo. Two large trials are addressing whether niacin has additional benefits on top of statin use: the Atherothrombosis intervention in metabolic syndrome with low HDL cholesterol/high triglyceride and impact on global health outcomes (AIM-HIGH) and the Heart protection study-2 treatment of HDL to reduce the incidence of vascular events (HPS-2 THRIVE) trials.
It is disappointing to note the recent discontinuation of the AIM-HIGH trial in coronary disease patients with low HDL, due to a safety analysis suggesting clinical futility with regard to the likelihood of achieving the primary endpoint and perhaps even a small increased risk of stroke.64,65 It is most likely that the AIM-HIGH trial was too small and therefore underpowered to address the main outcome measure(s) in the light of the observed event rate in the study population, but further insights will be provided when the full manuscript is published. The considerably larger HPS-2 THRIVE study is ongoing and should hopefully provide a more definitive answer.
Peroxisome Proliferator-activated Receptor-α Agonists
Fibrates activate peroxisome proliferator-activated receptor-α (PPAR-α), which alters expression of apolipoproteins and has downstream effects on many aspects of dyslipidaemia. Substantial reductions in TG are seen, alongside reductions in LDL-C and moderately increased HDL-C (10 %).66 Fibrate monotherapy has improved outcomes in patients with coronary disease and low HDL,67 but not in those with diabetes mellitus, although some subgroups may benefit.68,69 There is some reluctance to use fibrates in combination with statins, due to worries regarding increased adverse events (muscle aches, rhabdomyolysis). While genuine concerns exist for gemfibrozil, which inhibits statin metabolism and may increase the adverse effects of statins on muscle tissue,70 fenofibrate does not interact with statins and combination therapies demonstrate a good safety profile,71 with similar efficacy to niacin combination therapies.
The Action to control cardiovascular risk in diabetes (ACCORD) trial evaluated the effect of statin monotherapy versus combination therapy with fenofibrate on clinical outcomes in patients with type 2 diabetes mellitus without selecting for those with low HDL and/or elevated TG. Overall, patient outcomes were no better with the additional use of fenofibric acid, but subgroup analysis suggested a markedly beneficial effect on patient outcomes in those with high baseline TG and low HDL-C. Thus it might be appropriate to consider fenofibric acid–statin combination therapy to address residual lipid risk in diabetic patients with this characteristic dyslipidaemic lipid profile who are already on statin therapy.72
Cholesteryl Ester Transfer Protein Antagonists
CETP antagonists are a novel therapeutic class in HDL modification. CETP activity promotes neutral lipid exchange, in particular the delivery of cholesterol into Apo-B-containing atherogenic lipoproteins, and it also promotes HDL remodelling, which regenerates the pre-β HDL that has major importance in the initial stage of reverse cholesterol transport. CETP antagonists reduce HDL-C clearance and are the most potent therapeutic agents that increase the levels of HDL-C.73,74
Despite marked improvements in serum HDL-C, the CETP antagonist torcetrapib demonstrated increased risk of morbidity and mortality, which resulted in the discontinuation of the Investigation of lipid level management to understand its impact in atherosclerotic events (ILLUMINATE) trial.75 It is postulated that the failure of torcetrapib may be due to significant off-target effects, resulting in activation of the renin–angiotensin–aldosterone system and increased blood pressure in treated subjects. However, much controversy exists regarding the mechanisms underlying the adverse effects demonstrated, and the reasons for higher mortality and morbidity remain unclear.75 It is now hoped that the adverse effects of torcetrapib are not due to a ‘class’ effect.76 Indeed, these off-target effects are not seen with the more selective agent anacetrapib. Recent data from the Determining the efficacy and tolerability of CETP inhibition with anacetrapib (DEFINE) trial are reassuring, pointing towards an impressive effect on HDL values (138 % increase versus placebo) alongside a good safety profile, with no significant effect on morbidity and mortality.77,78 Dalcetrapib also shows promise, despite less potent CETP modulation and smaller increases in serum HDL-C levels (36 %) when compared with torcetrapib and anacetrapib. Unlike anacetrapib and torcetrapib, this molecule does not inhibit CETP-induced pre-β HDL formation,79 which may have clinical importance, as pre-β HDL is a key component of reverse cholesterol transport.79
Reassuring study results point towards improved safety profiles of the newer CETP agents anacetrapib and dalcetrapib.80 It is hoped that the encouraging early results will translate into improved outcomes in the large clinical trials currently in progress. Several pivotal trials will assess the efficacy of dalcetrapib: dal-PLAQUE will assess atherosclerosis and inflammation, dal-VESSEL endothelial function and blood pressure, and dal-ACUTE and dal-OUTCOMES the effect of dalcetrapibdalcetrapib on adverse clinical events in patients with acute coronary syndromes. Similarly, the Randomised evaluation of the effects of anacetrapib through lipid modification (REVEAL) HPS-3 TIMI-55 trial will assess the impact of anacetrapib on morbidity and mortality in 30,000 patients with known CVD.81–83
Novel HDL agents show early promise and aim to optimise HDL function in addition to improving serum HDL-C values. Apo-A1 mimetic compounds appear to confer atheroprotection via antioxidant properties in addition to increasing HDL-C formation, particle maturation and reverse cholesterol transport. Liver X receptor (LXR) agonists are thought to promote reverse cholesterol transport by modulating expression of membrane transporter ABCA1, which promotes delivery of lipids to nascent HDL particles. Although experimental animal and early clinical models have demonstrated the efficacy of these novel therapeutic agents,84–87 considerable further clinical research is needed to confirm their efficacy and safety.
There is growing evidence in favour of the use of additional therapeutic agents in combination with statins for the treatment of atherogenic dyslipidaemia. Niacin and fibrates have a good safety profile when used appropriately and lead to significantly improved levels of HDL-C. Current guidelines suggest that the additional use of a fibrate or niacin to statin therapy may be appropriate in high-risk patients with low HDL-C and/or hypertriglyceridaemia.32 Newer pharmacological therapies, such as CETP antagonists, Apo-A1 analogues and HDL mimetics, show early promise, but further investigation into their efficacy, safety and impact on cardiovascular events is needed before they can be licensed and recommended for use in clinical practice.
Non-pharmacological treatment remains fundamental to the management of cardiovascular risk. Healthy dietary intake and regular exercise have antiatherogenic effects through reduction in blood pressure and atherogenic dyslipidaemia,88 improved insulin sensitivity and glycaemic control.89 Recent data also suggest that dietary and lifestyle modification alone can convert HDL-C from a pro-inflammatory to an anti-inflammatory state.47 However, it is particularly difficult for patients to make long-term changes to lifestyle, and cost-effective support systems are elusive.90 Population-level interventions impacting dietary and other lifestyle choices may hold greater promise. Current recommendations – e.g., those of the NCEP-Adult Treatment Panel III – suggest that practitioners continue to advocate non-pharmacological strategies, including healthy diet, moderate daily exercise and smoking cessation, alongside the use of pharmacotherapeutics in certain high-risk individuals.32
Considerable evidence supports HDL as an atheroprotective agent and, as such, HDL is highly relevant to our understanding of CVD. HDL testing demonstrates value in cardiovascular risk estimation by accurately reclassifying patients into higher- or lower-risk categories, and its use is currently recommended by national and European guidelines. However, the clinical impact of medications addressing HDL as a treatment target remains controversial, and outcome data from large randomised clinical trials demonstrating an incremental impact of HDL-modifying agents in combination with background statin therapies are essential to justify their routine use. The results of these trials are eagerly awaited.