REVIEWS Novel metabolic biomarkers of cardiovascular disease Majken K. Jensen, Monica L. Bertoia, Leah E. Cahill, Isha Agarwal, Eric B. Rimm and Kenneth J. Mukamal Abstract | Coronary heart disease (CHD) accounts for one in every six deaths in US individuals. Great advances have been made in identifying important risk factors for CHD, such as hypertension, diabetes mellitus, smoking and hypercholesterolaemia, which have led to major developments in therapy. In particular, statins represent one of the greatest successes in the prevention of CHD. While these standard risk factors are important, an obvious opportunity exists to take advantage of ongoing scientific research to better risk-stratify individuals and to identify new treatment targets. In this Review, we summarize ongoing scientific research in a number of metabolic molecules or features, including lipoproteins, homocysteine, calcium metabolism and glycaemic markers. We evaluate the current state of the research and the strength of evidence supporting each emerging biomarker. We also discuss whether the associations with CHD are strong and consistent enough to improve current risk stratification metrics, and whether these markers enhance our understanding of the underlying biology of CHD and thus point towards new treatment options. Jensen, M. K. et al. Nat. Rev. Endocrinol. advance online publication 2 September 2014; doi:10.1038/nrendo.2014.155
Introduction
Department of Nutrition, Harvard School of Public Health, 665 Huntington Avenue, 02115 Boston, MA, USA (M.K.J., M.L.B., L.E.C., I.A.). Channing Division of Network Medicine, Brigham and Women’s Hospital, Harvard Medical School, 181 Longwood Avenue, 02115 Boston, MA, USA (E.B.R.). Department of Medicine, Beth Israel Deaconess Medical Centre, 1309 Beacon Street, 02446 Brookline, MA, USA (K.J.M.). Correspondence to: M.K.J. mkjensen@ hsph.harvard.edu
One in six US individuals dies from coronary heart dis ease (CHD).1 Prospective cohort studies, which began in the 1950s, have contributed to considerable advances in the identification of important risk factors for CHD, such as hypertension, diabetes mellitus, smoking, and hyper cholesterolaemia. The identification of these risk factors has led to major advances in therapy, with statins, in par ticular, representing one of the biggest victories in the prevention of CHD.2 However, clinical decision-making is still largely founded on fairly crude risk categories, which are derived from the initial risk factors that were identified over 50 years ago.3 Although these traditional risk factors are important, ongoing scientific advances are presenting opportunities to develop new targets to better risk-stratify individuals and to develop new treatments. Successful new drugs to prevent cardiovascular disease have not been developed since the introduction of statins, despite ongoing advances in the development of novel anticoagulants, drug-eluting stents and other treatments for established cardiovascular disease. Markers of inflam mation, such as C‑reactive protein, are the group of newly recognized novel risk markers that have received the most attention, owing to their strong and robust associations with risk of cardiovascular disease.4 However, as treatment Competing interests M.K.J. and E.B.R. have received unrestricted research support from Roche to measure HDL subtypes in the Multi-Ethnic Study of Atherosclerosis. M.K.J. and E.B.R. are listed as co-inventors on a patent application filed by Harvard University for HDL ApoC‑III (US Patent Application 13/046,682, filed 11 March 2011: “Assay and prediction of cardiovascular risk based on HDL subtypes according to apoC-III”). M.L.B., L.E.C., I.A. and K.J.M. declare no competing interests.
of inflammation per se has not yet been shown to alter cardiovascular risk, markers of inflammation only have a limited role in risk stratification of individuals at inter mediate risk of cardiovascular disease; inflammation is not considered a specific target for therapy.5 In this Review, we discuss ongoing scientific research in a number of molecules and features of metabolism, includ ing lipoproteins, homocysteine, calcium metabolism and markers of glycaemia (Box 1). We review the current state of research and discuss whether or not these markers are important for clinical care. In particular, we focus on whether the associations with CHD are strong and consist ent enough to enhance current risk-stratification metrics and whether these markers contribute to elucidating the underlying biology of CHD and thus indicate potential novel treatment options.
Lipoprotein and lipid-related markers
Lipoproteins transport hydrophobic cholesterol through out the body. The simplest differentiation between lipo proteins is based on their density, which is strongly related to the abundance of either apolipoprotein B (ApoB) or apolipoprotein A‑I (ApoA‑I). High levels of cholesterol in lipoproteins that contain ApoB—namely VLDL and even more so LDL—are associated with increased risk of CHD (Figure 1). ApoA‑I-containing lipoproteins (typi cally HDL) have been shown to be inversely associated with CHD in large observational studies evaluating the concentration of ApoA‑I per se 6,7 or the concentration of cholesterol in the HDL fraction (HDL cholesterol).8 Although ApoB is present on both VLDL and LDL lipo proteins (90% is associated with LDL), the measurement of ApoB or non-HDL cholesterol levels captures both of
NATURE REVIEWS | ENDOCRINOLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS Key points ■■ Coronary heart disease accounts for one in six deaths in US individuals ■■ Standard cardiovascular risk factors, including age, diabetes mellitus, smoking, hypertension and hypercholesterolaemia, are responsible for the majority of the risk of coronary heart disease ■■ Ongoing scientific investigations in a number of areas of metabolism research have discovered several novel biomarkers that highlight the underlying biology of cardiovascular risk ■■ The identification of novel biomarkers could result in improved diagnosis, risk assessment and treatment of patients ■■ New therapies for cardiovascular disease could result from the development and use of novel biomarkers
Box 1 | Overview of novel cardiovascular biomarkers Lipoprotein metabolism ■■ Lp(a)—strong evidence for role as independent risk marker (however, associations are modest); large reductions necessary to affect risk of cardiovascular disease; possible therapies include niacin ■■ OxPL—seems to be marker of risk, but more data needed; currently no OxPL-based therapies exist ■■ Lp–PLA2—marker of risk, possibly owing to strong link with LDL; novel therapies at phase III stage of clinical development ■■ HDL function (reverse cholesterol transport)—cholesterol efflux measures require confirmation in prospective studies; assays not available for large-scale settings; HDL particle size not consistently associated with risk of CHD ■■ HDL function (HDL subtype)—MPO-generated dysfunctional HDL not yet evaluated in prospective studies; ApoC-III–HDL subtype results need further replication; ApoC-III antisense therapies in development to treat hypertriglyceridaemia, but not evaluated in relation to presence of ApoC-III on HDL and to risk of CHD Homocysteine ■■ No clear direct association; treatments widely available that effectively lower homocysteine levels, but not shown to affect risk of CHD Mineral metabolism ■■ Vitamin D—inverse association between 25-hydroxyvitamin D and risk of cardiovascular disease; vitamin D supplementation to reduce the risk of cardiovascular disease remains controversial; large scale trials (at adequate doses) underway in initially healthy individuals ■■ FGF23—risk marker in individuals with ESRD or severely impaired kidney function, but utility in risk stratification for cardiovascular disease in healthy individuals questioned Glycaemia ■■ Adiponectin—HMW adiponectin is inversely associated with cardiovascular disease (supported by genetic association studies); no novel drugs that increase levels of adiponectin are under development, but exercise and diet have been shown to affect adiponectin levels ■■ HbA1c—independent predictor of risk of cardiovascular disease in adults without diabetes mellitus, even at modestly elevated levels (well below the threshold for diabetes mellitus); unknown whether or not interventions to reduce levels of HbA1c in individuals without diabetes mellitus decrease risk of cardiovascular disease ■■ Haptoglobin phenotype—Hp2‑2 phenotype consistently associated with increased risk of cardiovascular disease in individuals with diabetes mellitus or with elevated blood glucose levels Abbreviations: ApoC-III, apolipoprotein C‑III; CHD, coronary heart disease; ESRD, end-stage renal disease; FGF23, fibroblast growth factor 23; HMW, high molecular weight; Lp(a), lipoprotein(a); Lp–PLA2, lipoprotein-associated phospholipase A2; MPO, myeloperoxidase; OxPL, oxidized phospholipid.
these atherogenic particles and might add value beyond the measurement of LDL cholesterol levels in clinicalcare settings. Despite this appealing rationale, this addi tional value has not yet been convincingly demonstrated and high levels of LDL cholesterol continue to occupy an important place in clinical risk stratification.9
Lipoproteins in general, and LDL in particular, repre sent attractive targets for retarding and even regressing atherosclerosis.10 Nonetheless, while statins efficiently reduce blood levels of LDL cholesterol and concomi tantly reduce the risk of adverse cardiovascular events,11 other agents like ezetimibe, which decreases cholesterol absorption, have proven less successful.12 Even though the ongoing IMPROVE–IT trial13 is now planned to run to completion in the later part of 2014, whether the joint use of ezetimibe and statins, and the documented effects of this joint therapy in reducing LDL cholesterol levels, is efficacious for hard clinical end points remains uncertain. Such notable residual risk even among treated patients highlights the need for effective adjunctive therapies.14,15
Lipoprotein(a) Despite its similarity to both LDL and plasminogen— molecules whose biological activities are well understood— the function and purpose of lipoprotein Lp(a) is not yet clear. Lp(a) is an ApoB100 molecule that is covalently linked to Apo(a) and is synthesized by the liver.16,17 The ApoB100 component resembles LDL and the large Apo(a) glycoprotein has a sequence of Kringle IV repeats simi lar to the fibrinolytic proenzyme plasminogen.18 Similar to LDL, Lp(a) contributes to the development of athero sclerosis19–21 and foam-cell formation.22 Evidence suggests that Lp(a) is an atherosclerosis-specific marker but is not associated with the risk of thrombosis.21,23 A meta-analysis of 36 prospective studies that included 126,634 participants found that each SD increase in plasma levels of Lp(a) (~3.5-fold increase) was associ ated with a 13% increased risk of CHD (RR 1.13, 95% CI 1.09–1.18), even after adjustment for age, sex, systolic blood pressure, smoking status, history of diabetes mel litus, BMI, total cholesterol levels and history of stroke (RR 1.10, 95% CI 1.02–1.18). 24 No association with noncardiovascular-related mortality was found, lending support to the hypothesis that Lp(a) is an atherosclerosisspecific biomarker. Lp(a) only weakly correlates with traditional cardiovascular risk factors, such as levels of non-HDL cholesterol, ApoB100, fibrinogen and tri glycerides; the risk ratios reported in the meta-analysis were only slightly attenuated after adjusting for these common risk factors,24 adding credence to the hypoth esis that Lp(a) is an independent risk factor for CHD. On the basis of accumulating evidence, a 2013 consensus statement from the European Atherosclerosis Society recommended that primary care physicians should measure levels of Lp(a) in patients with familial hyper cholesterolaemia as a screening tool for the primary prevention of cardiovascular disease.25 Genetic studies support a potential causal relation ship between levels of Lp(a) and risk of cardiovascular disease. Variants at or near the Apo(a) genetic locus (LPA) strongly influence circulating levels of Lp(a) and Apo(a) size, which is inversely associated with levels of Lp(a).26,27 These genetic differences also translate into risk of CHD.28–30 However, compared with traditional cardio vascular disease risk factors such as non-HDL cholesterol levels, Lp(a) levels are only a modest risk factor and would
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrendo © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
Heart Artery
Phospholipid ApoA-I HDL
HDL
HDL
Cholesterol accumulation OxPL
CETP ApoB
Macrophage
Lp(a)
Potential plaque rupture Thrombus formation Acute event
Oxidation
LDL
CRP
LDL
IL-6
Progression of atherosclerosis
Reverse cholesterol transport
Liver
Figure 1 | Involvement of cholesterol and lipoproteins in atherosclerosis. Depicted is an example of an artery showing the progression of atherosclerosis. The process consists of deposition of cholesterol in the artery wall (by LDL), oxidation of LDL, inflammation (mediated by IL‑6 and CRP), attraction of monocytes and/or macrophages and the development of plaques. Atherosclerosis is a chronic process, which occurs in most people. Whether or not it results in an actual clinical manifestation of acute coronary events depends on the development of unstable plaques with a thin fibrous cap separating the lipid pool from the blood flow. Rupture of the plaques can lead to thrombosis. HDL size depends on the cholesterol concentration and ApoA‑I is the major apolipoprotein of HDL. HDL can take up small amounts of cholesterol from macrophages for transport back to the liver (reverse cholesterol transport), but this process depends on HDL functionality. Abbreviations: ApoA‑I, apolipoprotein A‑I; ApoB, apolipoprotein B; CETP, cholesteryl ester transfer protein; CRP, C‑reactive protein; Lp(a), lipoprotein(a); OxPL, oxidized phospholipids.
have to be dramatically reduced to have a substantial benefit in cardiovascular disease risk. To date, no therapies have proven successful in robustly modifying levels of Lp(a) in humans. Administration of niacin reduced the levels of Lp(a) by ~20%31–33 and the risk of cardiovascular disease in some trials,34 but not all,32 although interpretation of the results was complicated by the adverse effects of niacin on other metabolic param eters, such as glucose and uric acid levels. The HPS2– THRIVE niacin trial was terminated early owing to an increased risk of cardiomyopathy in patients receiving niacin plus laropiprant in combination with statins.35 As reported in previous trials, many patients in the inter vention arm could not tolerate niacin plus laropiprant owing to adverse effects on skin and the gastrointestinal tract. Statins can lower Lp(a) levels by ~10–20%,31,36 but whether the lowering of Lp(a) levels with statins reduces the risk of cardiovascular disease independently of the
beneficial effects of statins on reducing LDL cholesterol levels is uncertain.37,38 Several novel therapeutic targets with the potential to dramatically reduce levels of Lp(a) are currently being investigated. In mice, an antisense oligonucleotide tar geted at the gene encoding Apo(a) reduced plasma levels of Lp(a) by 25%39 and mipomersen, an antisense oligo nucleotide directed at the gene encoding ApoB100, reduced circulating levels of Lp(a) by 75%.40 Other potential thera peutic targets include the proprotein convertase subtilisin/ kexin type 9 (using monoclonal antibodies), the thyroid hormone receptor (using thyromimetics), the farnesoid X receptor–fibroblast growth factor axis and IL‑6.41 Overall, the results of the different therapeutic approaches are diffi cult to interpret because each therapy additionally changes levels of other plasma lipids, including LDL cholesterol, HDL cholesterol and triglycerides, although antisense technologies might be specific enough to help determine direct effects on levels of Lp(a). Therefore, although the association between levels of Lp(a) and the risk of cardio vascular disease is likely to be causal, existing approved therapies are not sufficiently specific to Lp(a) to target this intriguing molecule.41
OxPL on ApoB100-containing proteins Oxidized lipids have a central role in amplifying the inflammatory response and in the development of athero sclerosis and arterial plaques. Oxidized phospholipids (OxPL) on apoB100-containing lipoproteins (OxPL/ApoB) are an oxidation-specific biomarker transported by Lp(a), which circulates in plasma, deposits in the vascular wall and induces local inflammation.42 OxPL are also chaper oned by plasminogen in the circulation. Because OxPL lead to atherosclerosis even in animal models that lack Lp(a), it is possible that the adverse effect of Lp(a) origi nates from the role of Lp(a) in chaperoning these pro inflammatory phospholipids in the circulation. Levels of OxPL/ApoB are also independently associated with the risk of cardiovascular disease,42,43 although, like Lp(a), existing OxPL/ApoB-specific therapies do not yet exist. In general, antioxidants have not shown clear beneficial effects in preventing cardiovascular disease in humans despite promising data in animals.44 However, whether or not any of the tested agents actually reduce the levels of these highly inflammatory phospholipids is uncertain. Lipoprotein-associated phospholipase A2 Lipoprotein-associated phospholipase A2 (Lp–PLA2) is a secretory phospholipase and a member of one of the 15 groups that comprise the phospholipase A2 enzyme super family.45 In the circulation, Lp–PLA2 is bound to LDL (>80%), HDL (~10%) and negligibly to VLDL and Lp(a).46,47 Lp–PLA2 is produced by cells involved in athero sclerosis and inflammation, such as macrophages, T cells and mast cells,48 and cleaves the oxidized phosphatidyl choline (OxPC) component of oxidized LDL particles, generating proinflammatory and proatherogenic oxi dized fatty acids and lysophosphatidylcholine, which in turn activate inflammatory pathways in the vascular wall.49 Lp–PLA2 is highly upregulated in atherosclerotic plaques
NATURE REVIEWS | ENDOCRINOLOGY
ADVANCE ONLINE PUBLICATION | 3 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS and is linked to plaque rupture, and therefore might represent a potential marker of vascular inflammation. Physiologically, Lp–PLA2 has contradictory effects. The byproducts of the hydrolysis of OxPL, namely lysophosphatidylcholine and oxidized fatty acids, induce vascular smooth muscle cell migration, macrophage attraction and the expression of adhesion molecules and cytokines.49 However, this hydrolysis can also reduce the capacity of OxPL to further induce inflammation. Genetic variation in the gene encoding Lp–PLA2 (PLA2G7) also has contradictory effects and has been linked to both increased and decreased risk of cardiovascular disease,50 which is possibly explained by the binding of Lp–PLA2 to both proatherogenic LDL and antiatherogenic HDL.51 Lp–PLA2 might have a stronger link to vascular inflam mation than other more global markers of inflamma tion.52,53 Furthermore, Lp–PLA2 is associated with many traditional risk factors for cardiovascular disease51 and several studies have shown a robust association with risk of cardiovascular disease, even after adjustment in multi variable models.50 However, a large part of the positive association between activity and/or levels of Lp–PLA2 and risk of cardiovascular disease can be explained by its association with LDL.50 Moreover, studies in healthy individuals often failed to find an association between Lp–PLA2 and risk of cardiovascular disease.50,54 More evidence is needed to determine whether Lp–PLA2 is causally associated with cardiovascular disease. Many drugs commonly used for the primary or sec ondary prevention of cardiovascular disease lower Lp– PLA2 levels, including statins, lipid-lowering drugs, aspirin and β‑blockers, although none are specific to Lp–PLA2.51 Darapladib is an Lp–PLA2 inhibitor that has shown promising effects in patients with athero sclerosis in phase II clinical trials, specifically on plaque necrotic core lesion size, without altering levels of plasma lipids, lipoproteins or C‑reactive protein.55 However, the Stabilization of Atherosclerotic Plaque by Initiation of Darapladib Therapy trial (STABILITY) found that 4 years of darapladib therapy did not significantly reduce the risk of a composite end point of cardiovascular-related death, myocardial infarction or stroke (HR 0.89, 95% CI 0.77–1.03) in 15,828 participants with stable CHD, most of whom were receiving statins, β‑blockers, aspirin or angiotensin-converting-enzyme inhibitors.56 Nonetheless, a small reduction in the risk of total coronary events, including sudden cardiac death, nonfatal myocardial infarction, hospitalization for angina and any coronary revascularization procedure (HR 0.91, 95% CI 0.84–0.98) was observed.56 Even though these trial results suggest that darapladib does not noticeably reduce the risk of hard cardiovascular disease end points in patients with CHD, the achieved reductions in plasma levels of Lp– PLA2 suggest that the lack of response could be due to a minimal change in levels of Lp–PLA2.
HDL The strong inverse association between HDL cholesterol levels and risk of CHD has generated great interest in the concept of targeted therapy to raise total HDL cholesterol
levels to prevent CHD. This inverse relationship is pri marily due to the importance of HDL cholesterol in reverse cholesterol transport—the pathway by which peripherally deposited cholesterol is carried back to the liver for excretion; HDL cholesterol removes cholesterol from peripheral cells, suggesting that elevated levels of HDL cholesterol might reduce cardiovascular risk.57,58 However, the association between HDL cholesterol and CHD has proven to be more complex than initially appre ciated. Experimental studies of novel therapeutic agents, such as cholesteryl ester transfer protein inhibitors, have questioned the potential cardioprotective benefit of increasing HDL cholesterol. These agents dramatically raise levels of HDL cholesterol but have yet to show any beneficial effect on the clinical outcome of patients with CHD.59–61 Further, candidate and genome-wide predic tors of variation in HDL cholesterol have failed to dem onstrate an association with risk of CHD,62–65 as they have done for LDL-related genetic loci. Currently, a great unmet need exists for new assays and diagnostic tools that capture the functional proper ties of HDL, rather than merely its average cholesterol or ApoA‑I concentration. New assays of cholesterol efflux from macrophages have been developed that specifically measure the function of HDL in the reverse cholesterol pathway. However, accumulating evidence suggests that HDL has marked pleiotropic activities,57 implicating HDL in a wide range of biological pathways that extend beyond the classical view of HDL as a lipid-transport vehicle. HDL contains dozens of proteins with impor tant roles in oxidation, inflammation, haemostasis and innate immunity—pathways that are also implicated in cardiovascular disease.58 The heterogeneity in the com position and biological properties of HDL have made it increasingly apparent that a simple measure of total HDL cholesterol levels might not reflect the true underlying physiology and role of HDL cholesterol in atherosclerosis. For instance, modification of the ApoA‑I protein or the associated protein cargo could explain why HDL choles terol levels do not always linearly associate with a reduced risk of CHD in all individuals in every population.66–68 In the next section, we discuss aspects of HDL function, including sources of HDL heterogeneity and the progress made towards developing assays of HDL function for use as clinical biomarkers for CHD risk assessment.
HDL function Reverse cholesterol transport Cellular cholesterol efflux activity A new measure of ex vivo cholesterol efflux from macro phages in ApoB-depleted serum has been described in a cross-sectional study of healthy volunteers and patients undergoing cardiac catheterization.69 By use of this novel measurement, cholesterol efflux capacity was found to have a strong inverse association with carotid intima–media thickness that is independent of levels of both HDL cholesterol and ApoA‑I. This new assay speci fically measures cholesterol efflux from macrophages, which contributes only minimally to the overall flux through the reverse cholesterol transport pathway and
4 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrendo © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS to the cholesterol content of HDL. The results of this study suggest that macrophage cholesterol efflux capacity could provide important clinical information in relation to cardiovascular disease beyond the measurement of levels of HDL cholesterol itself, as study participants with superior efflux capacity were less likely to develop angio graphic coronary artery disease (CAD) than participants with inferior efflux capacity.69 However, results from a second study of cholesterol efflux capacity in relation to myocardial infarction and stroke (in both cross-sectional analyses of prevalent CHD and prospectively in a cohort of 1,150 participants with stable CHD who were under going elective diagnostic coronary angiography) did not confirm these original findings.70 Although participants with the highest cholesterol efflux had the lowest odds of prevalent CAD, the prospective analysis showed that those free of CAD initially were at the highest risk of developing CAD during the next 3 years.70 These studies emphasize the need for further investigation of the cholesterol efflux assay and the underlying biology of cholesterol efflux to fully understand how reverse cholesterol transport can be measured and how the results relate to atherosclerosis. Of note, the participants in both studies were receiving highdose statins and antihypertensive medications and LDL cholesterol was not a predictor of CHD risk in the latter study.70 These observations highlight the need for more studies of general healthy populations (similar to those seen in primary care), with sufficient follow-up duration to determine long-term clinical outcomes. HDL subspecies according to size HDL particle size is another widely used surrogate meas ure of HDL function in the reverse cholesterol transport pathway. This approach is founded on the reasoning that large HDL (high cholesterol content) reflect healthy cholesterol transport from arteries to HDL, and that small HDL indicate less effective cholesterol transport, as HDL cannot acquire its full potential cholesterol load. However, associations between HDL and risk of cardio vascular disease vary depending on the method used to determine HDL particle size. Some studies have shown that levels of large HDL particles are inversely associated with CHD,71,72 but other studies have found that levels of both large and small HDL particles predict reduced CHD incidence73,74 or that HDL size determination does not improve CHD risk prediction beyond that provided by measurement of total HDL and ApoA‑I levels.75,76 Importantly, measurement of HDL size has also not been useful in understanding the response to treatment. For example, cholesteryl ester transfer protein inhibi tors, which are either not associated with CHD risk or are associated with increased risk, seem to increase levels of large HDL.60,77 Similarly, the Veterans Administration HDL Intervention Trial (VA–HIT) showed that gem fibrozil decreased coronary events and mortality in patients who had low levels of HDL cholesterol.78 The risk of CHD was lowest for participants with the highest baseline concentrations of large HDL and lowest con centrations of small HDL.79 However, the drug itself decreased the concentration of large HDL and increased
the concentration of small HDL, both actions opposite to those predicted by the baseline HDL size study.80 At this time, the size of HDL does not have convincing utility in either risk prediction or the evaluation of treat ment responses to therapies that might alter HDL size or concentration.
HDL subtypes In addition to ApoA‑I, which makes up about 70% of the protein found in HDL, HDL has long been known to contain other apolipoproteins that take part in lipid metabolism. In the 1960s, Petar Alaupovic proposed that HDL circulates in subtypes established by the presence of specific apolipoproteins that guide its function.81 These apolipoproteins include ApoA-II, ApoC‑I, ApoC-II, ApoC-III and ApoE. With the exception of ApoA-II, the relationship between the other apolipoproteins and CHD has, until recently, remained largely undetermined. Some HDL types contain both ApoA‑I and ApoA-II, whereas others contain ApoA‑I but not ApoA-II. Overall, this distinction has not translated into differences in CHD risk; high levels of both types are associated with reduced CHD.71,73,79 Nonetheless, unbiased proteomic studies using liquid chromatography–tandem mass spectrometry have identified >200 proteins that are associated with the HDL proteome.82 Of these proteins, 85 HDL-associated proteins have been reported by at least three independent groups.82 This tremendous compositional heterogeneity of the HDL proteome reflects our emerging understand ing of the diverse functions of HDL. It is increasingly clear that apolipoproteins are crucial functional components of lipoprotein particles that determine their class, lipid cargo and size, and mediate their downstream interactions with receptors, enzymes and other proteins. MPO-oxidized HDL HDL recovered from atherosclerotic lesions is extensively oxidized by myeloperoxidase.83 ApoA‑I, containing the immunogenic epitope of myeloperoxidase (2-OH group of Trp72) is a poor cholesterol acceptor and exerts pro inflammatory effects on endothelial cells. Although ApoA‑I with the myeloperoxidase epitope is present at very low levels in human plasma (~0.007%), elevated levels of this HDL subtype were strongly associated with prevalent cardiovascular disease in an outpatient popula tion.83 MPO-oxidized HDL appears to represent a new form of dysfunctional HDL that warrants further inves tigation as a potential diagnostic and therapeutic target for cardiovascular disease, although its rarity might limit its utility in clinical settings. ApoC-III-containing subclasses Novel laboratory assays that discriminate between HDL subclasses on the basis of their content in specific apolipo proteins have been developed. To date, these assays have largely focused on ApoC-III because of its importance in lipoprotein transport and uptake. Proinflammatory ApoC-III is an important modifier of the association between both HDL cholesterol and ApoA‑I and risk of CHD,84 and increases the risk of CHD associated with
NATURE REVIEWS | ENDOCRINOLOGY
ADVANCE ONLINE PUBLICATION | 5 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS LDL particles.85 In two prospective US studies, the con centration of HDL cholesterol containing ApoC-III was, surprisingly, positively associated with CHD, whereas HDL cholesterol lacking ApoC-III was strongly inversely associated with CHD.84 These findings have now been replicated in a third prospective case–control study.86 Why might the ApoC-III content so successfully dif ferentiate beneficial from potentially adverse HDL subtypes? HDL with ApoC-III has impaired protective ability. For example, HDL with ApoC-III does not inhibit early events in atherosclerosis in which monocytes adhere to endothelial cells; conversely, both total HDL and HDL without ApoC-III provide this protection.87 Consider ing its effects on LDL metabolism, it is plausible that ApoC-III also interferes with HDL reuptake by the liver. Cumulatively, these findings suggest that the apolipo protein cargo of HDL might modify the function of HDL and thereby enable the identification of functionally distinct HDL subtypes. Improved understanding of the biological importance of the apolipoprotein cargo might reveal new pathways that can be used for the early detection of individuals at high risk of cardiometabolic disease and identify novel therapeutic targets. Indeed, this approach has already been used to guide the development of small molecule inhibitors targeting HDL-associated proteins, with one already in development for ApoC-III.88
Homocysteine
Elevated plasma levels of the amino acid homocysteine are positively associated with several mechanisms related to the risk of cardiovascular disease, including endothe lial cell dysfunction, oxidation of LDL and monocyte adhesion.89 Homocysteine was first identified as a pos sible risk factor in individuals with homocystinuria, who have very high levels of homocysteine essentially from birth. However, individuals with more common genetic polymorphisms in genes encoding enzymes involved in the conversion of homocysteine to methionine (such as MTHFR 677C>T and MTR 2756A>G) also have ele vated levels of homocysteine and in some studies have an elevated risk of cardiovascular disease.90 Furthermore, numerous prospective studies have shown a positive association between plasma homocysteine levels and the risk of cardiovascular disease.91 Interestingly, folate fortifi cation of grain products (which lowers population levels of homocysteine) is associated with a reduced risk of stroke in the USA,92 which has further stimulated interest in the manipulation of homocysteine levels for cardiovascular disease prevention. Robust observational studies linking homocysteine to cardiovascular disease and the in vitro and in vivo effects of elevated homocysteine levels have led to a series of ran domized controlled trials of B vitamin (including folate) supplementation, mostly in patients with established cardiovascular disease and in whom the risk of coro nary events is highest. B vitamins promote homocysteine metabolism93 and are therefore inversely associated with plasma homocysteine levels. Although in these clinical trials levels of homocysteine were consistently lowered by
~25% (~3 μmol/l) for an average of 5 years, no effect of B vitamin supplementation on the risk of cardiovascular events was found.94 Some evidence suggests that lowering homocysteine levels is only effective in individuals without pre-existing disease. For example, folic acid supplementation can reduce the risk of stroke by 25% in individuals with no family history of stroke,95,96 indicating that homocysteine levels might have an effect on primary rather than secon dary prevention. However, how the mechanisms linking homocysteine to risk of cardiovascular disease would differ between individuals with and without pre-existing cardiovascular disease is unclear. A remaining possibil ity is the need to decrease homocysteine levels over very long periods of time, such as would occur in individu als with genetic variants associated with elevated levels of homocysteine; however, testing this hypothesis in formal clinical trials is difficult. Although B vitamin sup plements are an inexpensive and effective therapy for lowering homocysteine levels, homocysteine does not seem to have a clear causal association with the risk of cardiovascular disease.
Mineral metabolism biomarkers Vitamin D An increasing number of healthy US adults take vitamin D supplements, primarily for bone health. 97 Howe ver, increasing evidence suggests that vitamin D is involved in multiple cardiometabolic processes, including the regula tion of pancreatic β-cell function, the renin–angiotensin system, vascular smooth muscle cell proliferation and myocardial hypertrophy.98 The major circulating form of vitamin D, 25-hydroxyvitamin D, reflects both intake and endogenous production of vitamin D.1 25-hydroxy vitamin D is highly protein-bound;99,100 only the pool of free 25-dihydroxyvitamin D, which represents
Introduction
Department of Nutrition, Harvard School of Public Health, 665 Huntington Avenue, 02115 Boston, MA, USA (M.K.J., M.L.B., L.E.C., I.A.). Channing Division of Network Medicine, Brigham and Women’s Hospital, Harvard Medical School, 181 Longwood Avenue, 02115 Boston, MA, USA (E.B.R.). Department of Medicine, Beth Israel Deaconess Medical Centre, 1309 Beacon Street, 02446 Brookline, MA, USA (K.J.M.). Correspondence to: M.K.J. mkjensen@ hsph.harvard.edu
One in six US individuals dies from coronary heart dis ease (CHD).1 Prospective cohort studies, which began in the 1950s, have contributed to considerable advances in the identification of important risk factors for CHD, such as hypertension, diabetes mellitus, smoking, and hyper cholesterolaemia. The identification of these risk factors has led to major advances in therapy, with statins, in par ticular, representing one of the biggest victories in the prevention of CHD.2 However, clinical decision-making is still largely founded on fairly crude risk categories, which are derived from the initial risk factors that were identified over 50 years ago.3 Although these traditional risk factors are important, ongoing scientific advances are presenting opportunities to develop new targets to better risk-stratify individuals and to develop new treatments. Successful new drugs to prevent cardiovascular disease have not been developed since the introduction of statins, despite ongoing advances in the development of novel anticoagulants, drug-eluting stents and other treatments for established cardiovascular disease. Markers of inflam mation, such as C‑reactive protein, are the group of newly recognized novel risk markers that have received the most attention, owing to their strong and robust associations with risk of cardiovascular disease.4 However, as treatment Competing interests M.K.J. and E.B.R. have received unrestricted research support from Roche to measure HDL subtypes in the Multi-Ethnic Study of Atherosclerosis. M.K.J. and E.B.R. are listed as co-inventors on a patent application filed by Harvard University for HDL ApoC‑III (US Patent Application 13/046,682, filed 11 March 2011: “Assay and prediction of cardiovascular risk based on HDL subtypes according to apoC-III”). M.L.B., L.E.C., I.A. and K.J.M. declare no competing interests.
of inflammation per se has not yet been shown to alter cardiovascular risk, markers of inflammation only have a limited role in risk stratification of individuals at inter mediate risk of cardiovascular disease; inflammation is not considered a specific target for therapy.5 In this Review, we discuss ongoing scientific research in a number of molecules and features of metabolism, includ ing lipoproteins, homocysteine, calcium metabolism and markers of glycaemia (Box 1). We review the current state of research and discuss whether or not these markers are important for clinical care. In particular, we focus on whether the associations with CHD are strong and consist ent enough to enhance current risk-stratification metrics and whether these markers contribute to elucidating the underlying biology of CHD and thus indicate potential novel treatment options.
Lipoprotein and lipid-related markers
Lipoproteins transport hydrophobic cholesterol through out the body. The simplest differentiation between lipo proteins is based on their density, which is strongly related to the abundance of either apolipoprotein B (ApoB) or apolipoprotein A‑I (ApoA‑I). High levels of cholesterol in lipoproteins that contain ApoB—namely VLDL and even more so LDL—are associated with increased risk of CHD (Figure 1). ApoA‑I-containing lipoproteins (typi cally HDL) have been shown to be inversely associated with CHD in large observational studies evaluating the concentration of ApoA‑I per se 6,7 or the concentration of cholesterol in the HDL fraction (HDL cholesterol).8 Although ApoB is present on both VLDL and LDL lipo proteins (90% is associated with LDL), the measurement of ApoB or non-HDL cholesterol levels captures both of
NATURE REVIEWS | ENDOCRINOLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS Key points ■■ Coronary heart disease accounts for one in six deaths in US individuals ■■ Standard cardiovascular risk factors, including age, diabetes mellitus, smoking, hypertension and hypercholesterolaemia, are responsible for the majority of the risk of coronary heart disease ■■ Ongoing scientific investigations in a number of areas of metabolism research have discovered several novel biomarkers that highlight the underlying biology of cardiovascular risk ■■ The identification of novel biomarkers could result in improved diagnosis, risk assessment and treatment of patients ■■ New therapies for cardiovascular disease could result from the development and use of novel biomarkers
Box 1 | Overview of novel cardiovascular biomarkers Lipoprotein metabolism ■■ Lp(a)—strong evidence for role as independent risk marker (however, associations are modest); large reductions necessary to affect risk of cardiovascular disease; possible therapies include niacin ■■ OxPL—seems to be marker of risk, but more data needed; currently no OxPL-based therapies exist ■■ Lp–PLA2—marker of risk, possibly owing to strong link with LDL; novel therapies at phase III stage of clinical development ■■ HDL function (reverse cholesterol transport)—cholesterol efflux measures require confirmation in prospective studies; assays not available for large-scale settings; HDL particle size not consistently associated with risk of CHD ■■ HDL function (HDL subtype)—MPO-generated dysfunctional HDL not yet evaluated in prospective studies; ApoC-III–HDL subtype results need further replication; ApoC-III antisense therapies in development to treat hypertriglyceridaemia, but not evaluated in relation to presence of ApoC-III on HDL and to risk of CHD Homocysteine ■■ No clear direct association; treatments widely available that effectively lower homocysteine levels, but not shown to affect risk of CHD Mineral metabolism ■■ Vitamin D—inverse association between 25-hydroxyvitamin D and risk of cardiovascular disease; vitamin D supplementation to reduce the risk of cardiovascular disease remains controversial; large scale trials (at adequate doses) underway in initially healthy individuals ■■ FGF23—risk marker in individuals with ESRD or severely impaired kidney function, but utility in risk stratification for cardiovascular disease in healthy individuals questioned Glycaemia ■■ Adiponectin—HMW adiponectin is inversely associated with cardiovascular disease (supported by genetic association studies); no novel drugs that increase levels of adiponectin are under development, but exercise and diet have been shown to affect adiponectin levels ■■ HbA1c—independent predictor of risk of cardiovascular disease in adults without diabetes mellitus, even at modestly elevated levels (well below the threshold for diabetes mellitus); unknown whether or not interventions to reduce levels of HbA1c in individuals without diabetes mellitus decrease risk of cardiovascular disease ■■ Haptoglobin phenotype—Hp2‑2 phenotype consistently associated with increased risk of cardiovascular disease in individuals with diabetes mellitus or with elevated blood glucose levels Abbreviations: ApoC-III, apolipoprotein C‑III; CHD, coronary heart disease; ESRD, end-stage renal disease; FGF23, fibroblast growth factor 23; HMW, high molecular weight; Lp(a), lipoprotein(a); Lp–PLA2, lipoprotein-associated phospholipase A2; MPO, myeloperoxidase; OxPL, oxidized phospholipid.
these atherogenic particles and might add value beyond the measurement of LDL cholesterol levels in clinicalcare settings. Despite this appealing rationale, this addi tional value has not yet been convincingly demonstrated and high levels of LDL cholesterol continue to occupy an important place in clinical risk stratification.9
Lipoproteins in general, and LDL in particular, repre sent attractive targets for retarding and even regressing atherosclerosis.10 Nonetheless, while statins efficiently reduce blood levels of LDL cholesterol and concomi tantly reduce the risk of adverse cardiovascular events,11 other agents like ezetimibe, which decreases cholesterol absorption, have proven less successful.12 Even though the ongoing IMPROVE–IT trial13 is now planned to run to completion in the later part of 2014, whether the joint use of ezetimibe and statins, and the documented effects of this joint therapy in reducing LDL cholesterol levels, is efficacious for hard clinical end points remains uncertain. Such notable residual risk even among treated patients highlights the need for effective adjunctive therapies.14,15
Lipoprotein(a) Despite its similarity to both LDL and plasminogen— molecules whose biological activities are well understood— the function and purpose of lipoprotein Lp(a) is not yet clear. Lp(a) is an ApoB100 molecule that is covalently linked to Apo(a) and is synthesized by the liver.16,17 The ApoB100 component resembles LDL and the large Apo(a) glycoprotein has a sequence of Kringle IV repeats simi lar to the fibrinolytic proenzyme plasminogen.18 Similar to LDL, Lp(a) contributes to the development of athero sclerosis19–21 and foam-cell formation.22 Evidence suggests that Lp(a) is an atherosclerosis-specific marker but is not associated with the risk of thrombosis.21,23 A meta-analysis of 36 prospective studies that included 126,634 participants found that each SD increase in plasma levels of Lp(a) (~3.5-fold increase) was associ ated with a 13% increased risk of CHD (RR 1.13, 95% CI 1.09–1.18), even after adjustment for age, sex, systolic blood pressure, smoking status, history of diabetes mel litus, BMI, total cholesterol levels and history of stroke (RR 1.10, 95% CI 1.02–1.18). 24 No association with noncardiovascular-related mortality was found, lending support to the hypothesis that Lp(a) is an atherosclerosisspecific biomarker. Lp(a) only weakly correlates with traditional cardiovascular risk factors, such as levels of non-HDL cholesterol, ApoB100, fibrinogen and tri glycerides; the risk ratios reported in the meta-analysis were only slightly attenuated after adjusting for these common risk factors,24 adding credence to the hypoth esis that Lp(a) is an independent risk factor for CHD. On the basis of accumulating evidence, a 2013 consensus statement from the European Atherosclerosis Society recommended that primary care physicians should measure levels of Lp(a) in patients with familial hyper cholesterolaemia as a screening tool for the primary prevention of cardiovascular disease.25 Genetic studies support a potential causal relation ship between levels of Lp(a) and risk of cardiovascular disease. Variants at or near the Apo(a) genetic locus (LPA) strongly influence circulating levels of Lp(a) and Apo(a) size, which is inversely associated with levels of Lp(a).26,27 These genetic differences also translate into risk of CHD.28–30 However, compared with traditional cardio vascular disease risk factors such as non-HDL cholesterol levels, Lp(a) levels are only a modest risk factor and would
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrendo © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
Heart Artery
Phospholipid ApoA-I HDL
HDL
HDL
Cholesterol accumulation OxPL
CETP ApoB
Macrophage
Lp(a)
Potential plaque rupture Thrombus formation Acute event
Oxidation
LDL
CRP
LDL
IL-6
Progression of atherosclerosis
Reverse cholesterol transport
Liver
Figure 1 | Involvement of cholesterol and lipoproteins in atherosclerosis. Depicted is an example of an artery showing the progression of atherosclerosis. The process consists of deposition of cholesterol in the artery wall (by LDL), oxidation of LDL, inflammation (mediated by IL‑6 and CRP), attraction of monocytes and/or macrophages and the development of plaques. Atherosclerosis is a chronic process, which occurs in most people. Whether or not it results in an actual clinical manifestation of acute coronary events depends on the development of unstable plaques with a thin fibrous cap separating the lipid pool from the blood flow. Rupture of the plaques can lead to thrombosis. HDL size depends on the cholesterol concentration and ApoA‑I is the major apolipoprotein of HDL. HDL can take up small amounts of cholesterol from macrophages for transport back to the liver (reverse cholesterol transport), but this process depends on HDL functionality. Abbreviations: ApoA‑I, apolipoprotein A‑I; ApoB, apolipoprotein B; CETP, cholesteryl ester transfer protein; CRP, C‑reactive protein; Lp(a), lipoprotein(a); OxPL, oxidized phospholipids.
have to be dramatically reduced to have a substantial benefit in cardiovascular disease risk. To date, no therapies have proven successful in robustly modifying levels of Lp(a) in humans. Administration of niacin reduced the levels of Lp(a) by ~20%31–33 and the risk of cardiovascular disease in some trials,34 but not all,32 although interpretation of the results was complicated by the adverse effects of niacin on other metabolic param eters, such as glucose and uric acid levels. The HPS2– THRIVE niacin trial was terminated early owing to an increased risk of cardiomyopathy in patients receiving niacin plus laropiprant in combination with statins.35 As reported in previous trials, many patients in the inter vention arm could not tolerate niacin plus laropiprant owing to adverse effects on skin and the gastrointestinal tract. Statins can lower Lp(a) levels by ~10–20%,31,36 but whether the lowering of Lp(a) levels with statins reduces the risk of cardiovascular disease independently of the
beneficial effects of statins on reducing LDL cholesterol levels is uncertain.37,38 Several novel therapeutic targets with the potential to dramatically reduce levels of Lp(a) are currently being investigated. In mice, an antisense oligonucleotide tar geted at the gene encoding Apo(a) reduced plasma levels of Lp(a) by 25%39 and mipomersen, an antisense oligo nucleotide directed at the gene encoding ApoB100, reduced circulating levels of Lp(a) by 75%.40 Other potential thera peutic targets include the proprotein convertase subtilisin/ kexin type 9 (using monoclonal antibodies), the thyroid hormone receptor (using thyromimetics), the farnesoid X receptor–fibroblast growth factor axis and IL‑6.41 Overall, the results of the different therapeutic approaches are diffi cult to interpret because each therapy additionally changes levels of other plasma lipids, including LDL cholesterol, HDL cholesterol and triglycerides, although antisense technologies might be specific enough to help determine direct effects on levels of Lp(a). Therefore, although the association between levels of Lp(a) and the risk of cardio vascular disease is likely to be causal, existing approved therapies are not sufficiently specific to Lp(a) to target this intriguing molecule.41
OxPL on ApoB100-containing proteins Oxidized lipids have a central role in amplifying the inflammatory response and in the development of athero sclerosis and arterial plaques. Oxidized phospholipids (OxPL) on apoB100-containing lipoproteins (OxPL/ApoB) are an oxidation-specific biomarker transported by Lp(a), which circulates in plasma, deposits in the vascular wall and induces local inflammation.42 OxPL are also chaper oned by plasminogen in the circulation. Because OxPL lead to atherosclerosis even in animal models that lack Lp(a), it is possible that the adverse effect of Lp(a) origi nates from the role of Lp(a) in chaperoning these pro inflammatory phospholipids in the circulation. Levels of OxPL/ApoB are also independently associated with the risk of cardiovascular disease,42,43 although, like Lp(a), existing OxPL/ApoB-specific therapies do not yet exist. In general, antioxidants have not shown clear beneficial effects in preventing cardiovascular disease in humans despite promising data in animals.44 However, whether or not any of the tested agents actually reduce the levels of these highly inflammatory phospholipids is uncertain. Lipoprotein-associated phospholipase A2 Lipoprotein-associated phospholipase A2 (Lp–PLA2) is a secretory phospholipase and a member of one of the 15 groups that comprise the phospholipase A2 enzyme super family.45 In the circulation, Lp–PLA2 is bound to LDL (>80%), HDL (~10%) and negligibly to VLDL and Lp(a).46,47 Lp–PLA2 is produced by cells involved in athero sclerosis and inflammation, such as macrophages, T cells and mast cells,48 and cleaves the oxidized phosphatidyl choline (OxPC) component of oxidized LDL particles, generating proinflammatory and proatherogenic oxi dized fatty acids and lysophosphatidylcholine, which in turn activate inflammatory pathways in the vascular wall.49 Lp–PLA2 is highly upregulated in atherosclerotic plaques
NATURE REVIEWS | ENDOCRINOLOGY
ADVANCE ONLINE PUBLICATION | 3 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS and is linked to plaque rupture, and therefore might represent a potential marker of vascular inflammation. Physiologically, Lp–PLA2 has contradictory effects. The byproducts of the hydrolysis of OxPL, namely lysophosphatidylcholine and oxidized fatty acids, induce vascular smooth muscle cell migration, macrophage attraction and the expression of adhesion molecules and cytokines.49 However, this hydrolysis can also reduce the capacity of OxPL to further induce inflammation. Genetic variation in the gene encoding Lp–PLA2 (PLA2G7) also has contradictory effects and has been linked to both increased and decreased risk of cardiovascular disease,50 which is possibly explained by the binding of Lp–PLA2 to both proatherogenic LDL and antiatherogenic HDL.51 Lp–PLA2 might have a stronger link to vascular inflam mation than other more global markers of inflamma tion.52,53 Furthermore, Lp–PLA2 is associated with many traditional risk factors for cardiovascular disease51 and several studies have shown a robust association with risk of cardiovascular disease, even after adjustment in multi variable models.50 However, a large part of the positive association between activity and/or levels of Lp–PLA2 and risk of cardiovascular disease can be explained by its association with LDL.50 Moreover, studies in healthy individuals often failed to find an association between Lp–PLA2 and risk of cardiovascular disease.50,54 More evidence is needed to determine whether Lp–PLA2 is causally associated with cardiovascular disease. Many drugs commonly used for the primary or sec ondary prevention of cardiovascular disease lower Lp– PLA2 levels, including statins, lipid-lowering drugs, aspirin and β‑blockers, although none are specific to Lp–PLA2.51 Darapladib is an Lp–PLA2 inhibitor that has shown promising effects in patients with athero sclerosis in phase II clinical trials, specifically on plaque necrotic core lesion size, without altering levels of plasma lipids, lipoproteins or C‑reactive protein.55 However, the Stabilization of Atherosclerotic Plaque by Initiation of Darapladib Therapy trial (STABILITY) found that 4 years of darapladib therapy did not significantly reduce the risk of a composite end point of cardiovascular-related death, myocardial infarction or stroke (HR 0.89, 95% CI 0.77–1.03) in 15,828 participants with stable CHD, most of whom were receiving statins, β‑blockers, aspirin or angiotensin-converting-enzyme inhibitors.56 Nonetheless, a small reduction in the risk of total coronary events, including sudden cardiac death, nonfatal myocardial infarction, hospitalization for angina and any coronary revascularization procedure (HR 0.91, 95% CI 0.84–0.98) was observed.56 Even though these trial results suggest that darapladib does not noticeably reduce the risk of hard cardiovascular disease end points in patients with CHD, the achieved reductions in plasma levels of Lp– PLA2 suggest that the lack of response could be due to a minimal change in levels of Lp–PLA2.
HDL The strong inverse association between HDL cholesterol levels and risk of CHD has generated great interest in the concept of targeted therapy to raise total HDL cholesterol
levels to prevent CHD. This inverse relationship is pri marily due to the importance of HDL cholesterol in reverse cholesterol transport—the pathway by which peripherally deposited cholesterol is carried back to the liver for excretion; HDL cholesterol removes cholesterol from peripheral cells, suggesting that elevated levels of HDL cholesterol might reduce cardiovascular risk.57,58 However, the association between HDL cholesterol and CHD has proven to be more complex than initially appre ciated. Experimental studies of novel therapeutic agents, such as cholesteryl ester transfer protein inhibitors, have questioned the potential cardioprotective benefit of increasing HDL cholesterol. These agents dramatically raise levels of HDL cholesterol but have yet to show any beneficial effect on the clinical outcome of patients with CHD.59–61 Further, candidate and genome-wide predic tors of variation in HDL cholesterol have failed to dem onstrate an association with risk of CHD,62–65 as they have done for LDL-related genetic loci. Currently, a great unmet need exists for new assays and diagnostic tools that capture the functional proper ties of HDL, rather than merely its average cholesterol or ApoA‑I concentration. New assays of cholesterol efflux from macrophages have been developed that specifically measure the function of HDL in the reverse cholesterol pathway. However, accumulating evidence suggests that HDL has marked pleiotropic activities,57 implicating HDL in a wide range of biological pathways that extend beyond the classical view of HDL as a lipid-transport vehicle. HDL contains dozens of proteins with impor tant roles in oxidation, inflammation, haemostasis and innate immunity—pathways that are also implicated in cardiovascular disease.58 The heterogeneity in the com position and biological properties of HDL have made it increasingly apparent that a simple measure of total HDL cholesterol levels might not reflect the true underlying physiology and role of HDL cholesterol in atherosclerosis. For instance, modification of the ApoA‑I protein or the associated protein cargo could explain why HDL choles terol levels do not always linearly associate with a reduced risk of CHD in all individuals in every population.66–68 In the next section, we discuss aspects of HDL function, including sources of HDL heterogeneity and the progress made towards developing assays of HDL function for use as clinical biomarkers for CHD risk assessment.
HDL function Reverse cholesterol transport Cellular cholesterol efflux activity A new measure of ex vivo cholesterol efflux from macro phages in ApoB-depleted serum has been described in a cross-sectional study of healthy volunteers and patients undergoing cardiac catheterization.69 By use of this novel measurement, cholesterol efflux capacity was found to have a strong inverse association with carotid intima–media thickness that is independent of levels of both HDL cholesterol and ApoA‑I. This new assay speci fically measures cholesterol efflux from macrophages, which contributes only minimally to the overall flux through the reverse cholesterol transport pathway and
4 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrendo © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS to the cholesterol content of HDL. The results of this study suggest that macrophage cholesterol efflux capacity could provide important clinical information in relation to cardiovascular disease beyond the measurement of levels of HDL cholesterol itself, as study participants with superior efflux capacity were less likely to develop angio graphic coronary artery disease (CAD) than participants with inferior efflux capacity.69 However, results from a second study of cholesterol efflux capacity in relation to myocardial infarction and stroke (in both cross-sectional analyses of prevalent CHD and prospectively in a cohort of 1,150 participants with stable CHD who were under going elective diagnostic coronary angiography) did not confirm these original findings.70 Although participants with the highest cholesterol efflux had the lowest odds of prevalent CAD, the prospective analysis showed that those free of CAD initially were at the highest risk of developing CAD during the next 3 years.70 These studies emphasize the need for further investigation of the cholesterol efflux assay and the underlying biology of cholesterol efflux to fully understand how reverse cholesterol transport can be measured and how the results relate to atherosclerosis. Of note, the participants in both studies were receiving highdose statins and antihypertensive medications and LDL cholesterol was not a predictor of CHD risk in the latter study.70 These observations highlight the need for more studies of general healthy populations (similar to those seen in primary care), with sufficient follow-up duration to determine long-term clinical outcomes. HDL subspecies according to size HDL particle size is another widely used surrogate meas ure of HDL function in the reverse cholesterol transport pathway. This approach is founded on the reasoning that large HDL (high cholesterol content) reflect healthy cholesterol transport from arteries to HDL, and that small HDL indicate less effective cholesterol transport, as HDL cannot acquire its full potential cholesterol load. However, associations between HDL and risk of cardio vascular disease vary depending on the method used to determine HDL particle size. Some studies have shown that levels of large HDL particles are inversely associated with CHD,71,72 but other studies have found that levels of both large and small HDL particles predict reduced CHD incidence73,74 or that HDL size determination does not improve CHD risk prediction beyond that provided by measurement of total HDL and ApoA‑I levels.75,76 Importantly, measurement of HDL size has also not been useful in understanding the response to treatment. For example, cholesteryl ester transfer protein inhibi tors, which are either not associated with CHD risk or are associated with increased risk, seem to increase levels of large HDL.60,77 Similarly, the Veterans Administration HDL Intervention Trial (VA–HIT) showed that gem fibrozil decreased coronary events and mortality in patients who had low levels of HDL cholesterol.78 The risk of CHD was lowest for participants with the highest baseline concentrations of large HDL and lowest con centrations of small HDL.79 However, the drug itself decreased the concentration of large HDL and increased
the concentration of small HDL, both actions opposite to those predicted by the baseline HDL size study.80 At this time, the size of HDL does not have convincing utility in either risk prediction or the evaluation of treat ment responses to therapies that might alter HDL size or concentration.
HDL subtypes In addition to ApoA‑I, which makes up about 70% of the protein found in HDL, HDL has long been known to contain other apolipoproteins that take part in lipid metabolism. In the 1960s, Petar Alaupovic proposed that HDL circulates in subtypes established by the presence of specific apolipoproteins that guide its function.81 These apolipoproteins include ApoA-II, ApoC‑I, ApoC-II, ApoC-III and ApoE. With the exception of ApoA-II, the relationship between the other apolipoproteins and CHD has, until recently, remained largely undetermined. Some HDL types contain both ApoA‑I and ApoA-II, whereas others contain ApoA‑I but not ApoA-II. Overall, this distinction has not translated into differences in CHD risk; high levels of both types are associated with reduced CHD.71,73,79 Nonetheless, unbiased proteomic studies using liquid chromatography–tandem mass spectrometry have identified >200 proteins that are associated with the HDL proteome.82 Of these proteins, 85 HDL-associated proteins have been reported by at least three independent groups.82 This tremendous compositional heterogeneity of the HDL proteome reflects our emerging understand ing of the diverse functions of HDL. It is increasingly clear that apolipoproteins are crucial functional components of lipoprotein particles that determine their class, lipid cargo and size, and mediate their downstream interactions with receptors, enzymes and other proteins. MPO-oxidized HDL HDL recovered from atherosclerotic lesions is extensively oxidized by myeloperoxidase.83 ApoA‑I, containing the immunogenic epitope of myeloperoxidase (2-OH group of Trp72) is a poor cholesterol acceptor and exerts pro inflammatory effects on endothelial cells. Although ApoA‑I with the myeloperoxidase epitope is present at very low levels in human plasma (~0.007%), elevated levels of this HDL subtype were strongly associated with prevalent cardiovascular disease in an outpatient popula tion.83 MPO-oxidized HDL appears to represent a new form of dysfunctional HDL that warrants further inves tigation as a potential diagnostic and therapeutic target for cardiovascular disease, although its rarity might limit its utility in clinical settings. ApoC-III-containing subclasses Novel laboratory assays that discriminate between HDL subclasses on the basis of their content in specific apolipo proteins have been developed. To date, these assays have largely focused on ApoC-III because of its importance in lipoprotein transport and uptake. Proinflammatory ApoC-III is an important modifier of the association between both HDL cholesterol and ApoA‑I and risk of CHD,84 and increases the risk of CHD associated with
NATURE REVIEWS | ENDOCRINOLOGY
ADVANCE ONLINE PUBLICATION | 5 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS LDL particles.85 In two prospective US studies, the con centration of HDL cholesterol containing ApoC-III was, surprisingly, positively associated with CHD, whereas HDL cholesterol lacking ApoC-III was strongly inversely associated with CHD.84 These findings have now been replicated in a third prospective case–control study.86 Why might the ApoC-III content so successfully dif ferentiate beneficial from potentially adverse HDL subtypes? HDL with ApoC-III has impaired protective ability. For example, HDL with ApoC-III does not inhibit early events in atherosclerosis in which monocytes adhere to endothelial cells; conversely, both total HDL and HDL without ApoC-III provide this protection.87 Consider ing its effects on LDL metabolism, it is plausible that ApoC-III also interferes with HDL reuptake by the liver. Cumulatively, these findings suggest that the apolipo protein cargo of HDL might modify the function of HDL and thereby enable the identification of functionally distinct HDL subtypes. Improved understanding of the biological importance of the apolipoprotein cargo might reveal new pathways that can be used for the early detection of individuals at high risk of cardiometabolic disease and identify novel therapeutic targets. Indeed, this approach has already been used to guide the development of small molecule inhibitors targeting HDL-associated proteins, with one already in development for ApoC-III.88
Homocysteine
Elevated plasma levels of the amino acid homocysteine are positively associated with several mechanisms related to the risk of cardiovascular disease, including endothe lial cell dysfunction, oxidation of LDL and monocyte adhesion.89 Homocysteine was first identified as a pos sible risk factor in individuals with homocystinuria, who have very high levels of homocysteine essentially from birth. However, individuals with more common genetic polymorphisms in genes encoding enzymes involved in the conversion of homocysteine to methionine (such as MTHFR 677C>T and MTR 2756A>G) also have ele vated levels of homocysteine and in some studies have an elevated risk of cardiovascular disease.90 Furthermore, numerous prospective studies have shown a positive association between plasma homocysteine levels and the risk of cardiovascular disease.91 Interestingly, folate fortifi cation of grain products (which lowers population levels of homocysteine) is associated with a reduced risk of stroke in the USA,92 which has further stimulated interest in the manipulation of homocysteine levels for cardiovascular disease prevention. Robust observational studies linking homocysteine to cardiovascular disease and the in vitro and in vivo effects of elevated homocysteine levels have led to a series of ran domized controlled trials of B vitamin (including folate) supplementation, mostly in patients with established cardiovascular disease and in whom the risk of coro nary events is highest. B vitamins promote homocysteine metabolism93 and are therefore inversely associated with plasma homocysteine levels. Although in these clinical trials levels of homocysteine were consistently lowered by
~25% (~3 μmol/l) for an average of 5 years, no effect of B vitamin supplementation on the risk of cardiovascular events was found.94 Some evidence suggests that lowering homocysteine levels is only effective in individuals without pre-existing disease. For example, folic acid supplementation can reduce the risk of stroke by 25% in individuals with no family history of stroke,95,96 indicating that homocysteine levels might have an effect on primary rather than secon dary prevention. However, how the mechanisms linking homocysteine to risk of cardiovascular disease would differ between individuals with and without pre-existing cardiovascular disease is unclear. A remaining possibil ity is the need to decrease homocysteine levels over very long periods of time, such as would occur in individu als with genetic variants associated with elevated levels of homocysteine; however, testing this hypothesis in formal clinical trials is difficult. Although B vitamin sup plements are an inexpensive and effective therapy for lowering homocysteine levels, homocysteine does not seem to have a clear causal association with the risk of cardiovascular disease.
Mineral metabolism biomarkers Vitamin D An increasing number of healthy US adults take vitamin D supplements, primarily for bone health. 97 Howe ver, increasing evidence suggests that vitamin D is involved in multiple cardiometabolic processes, including the regula tion of pancreatic β-cell function, the renin–angiotensin system, vascular smooth muscle cell proliferation and myocardial hypertrophy.98 The major circulating form of vitamin D, 25-hydroxyvitamin D, reflects both intake and endogenous production of vitamin D.1 25-hydroxy vitamin D is highly protein-bound;99,100 only the pool of free 25-dihydroxyvitamin D, which represents
