Lipoprotein(a): Biology and Clinical Importance

Abstract

Lipoprotein(a) [Lp(a)] is a unique lipoprotein that has emerged as an independent risk factor for developing vascular disease. Plasma Lp(a) levels above the common cut-off level of 300 mg/L place individuals at risk of developing heart disease particularly if combined with other lipid and thrombogenic risk factors. Studies in humans have shown Lp(a) levels to be hugely variable and under strict genetic control, largely by the apolipoprotein(a) [apo(a)] gene. In general, Lp(a) levels have proven difficult to manipulate, although some factors have been identified that can influence levels. Research has shown that Lp(a) has a high affinity for the arterial wall and displays many athero-thrombogenic properties. While a definite function for Lp(a) has not been identified, the last two decades of research have provided much information on the biology and clinical importance of Lp(a).

Introduction

Lp(a) was first identified and classified as a “low density lipoprotein variant” over 40 years ago. Similar to low-density lipoprotein (LDL), but containing an additional protein, apo(a), Lp(a) has proven to be pathogenic in nature and involved in the development of coronary heart disease (CHD). Lp(a) has an unusual species distribution being present only in primates and hedgehogs. This makes animal models for the study of Lp(a) metabolism and pathogenicity scarce, although both transgenic rabbit and mouse models have recently been developed. To date, the only major influence on Lp(a) levels is a size polymorphism in the apo(a) gene. Traditional lipoprotein-lowering drugs do not significantly alter Lp(a) levels. This review highlights the latest findings on Lp(a) including new insights into its assembly, pathogenicity and growing clinical importance as a CHD risk factor. Factors affecting plasma Lp(a) levels and new possibilities for Lp(a)-lowering interventions will also be discussed.

Lp(a) Structure

Lp(a) was discovered in human serum in 1963 by Kåre Berg during a study of variation in LDL antigenicity.¹ Further characterisation of Lp(a) showed it to consist of a LDL covalently bound to a unique protein called apo(a).² Apo(a) is a homologue of plasminogen, containing multiple copies of plasminogen kringle 4, a single copy of plasminogen kringle 5, and an inactive protease domain.³ Past studies have shown that the number of kringle 4 (KIV) domains can vary from 12 to 51 giving rise to 34 different-sized apo(a) isoforms.⁴ Furthermore, within the repeated KIV domain exists 10 distinct types (KIV types 1–10) each present in single copy except for KIV type 2 which exists in varying number.³

Reducing agents were shown to dissociate apo(a) from LDL pointing to the existence of a disulphide linkage between apo(a) and the apolipoprotein (apo) B100 moiety of the LDL molecule.⁵ Indeed, mutagenesis studies have established that a disulphide bond exists between apo(a)Cys4057 ^6,7 and apoBCys4326.^8,9 Hydrodynamic studies of recombinant Lp(a) particles showed that a single copy of apo(a) attaches to the single copy of apoB on the LDL surface with minimal contact between the two proteins.¹⁰ More recent electron microscopy studies, however, have shown apo(a) to be wrapped around the LDL molecule suggesting multiple contacts between the apo(a) and apoB proteins.¹¹ This observation is supported by many studies that have identified multiple apo(a) sequences that interact noncovalently with apoB and vice versa. Some of these interactions are likely to be involved in the assembly of Lp(a) while others could be formed post-assembly to help stabilise the structure.

Lp(a) Assembly

The majority of evidence suggests that Lp(a) assembly occurs extracellularly, either in circulation or at the hepatocyte surface (reviewed in reference ¹²). There is, however, some evidence for intracellular assembly from kinetic studies in humans.¹³ Studies over the last decade have indicated that Lp(a) is assembled in a two-step manner involving noncovalent interactions between apo(a) and apoB that precede the formation of the disulphide bond between apo(a)Cys4057 and apoBCys4326.^6,14 Experiments showing that lysine analogues disrupt Lp(a) assembly have suggested that lysine binding domains in apo(a) and lysine residues in apoB are involved in the initial non-covalent interaction.^15,16 Mutagenesis studies documenting defective Lp(a) assembly after the removal of lysine binding sites in apo(a) KIV types 6–8 have provided further evidence for this.¹⁷ A recent study in which single point mutations were introduced into the lysine binding sites in KIV 6–8 has revealed that KIV 7 and 8, but not KIV 6, are required for efficient Lp(a) assembly.¹⁸

A number of apoB sequences that noncovalently bind apo(a) have been reported. Becker et al. have identified an apoB lysine residue in the N-terminus, apoBLys680, that mediates the noncovalent binding of an apoB18 fragment to apo(a).¹⁹ Subsequent studies have shown that a peptide containing the apoBLys680 residue binds to apo(a) KIV type 7.¹⁸ Sequences in the C-terminal region of apoB have also been implicated in the noncovalent interaction with apo(a). Experiments with truncated apoB molecules identified a region between apoB95 (4330 amino acids) and apoB97 (4397 amino acids) that was important for efficient Lp(a) assembly.²⁰ Further characterisation of the apoB4330 to 4397 region identified a 21 amino acid sequence (amino acids 4372–4392) containing four lysine residues (including a highly conserved lysine at position 4372) that posed a potential apo(a) binding site.²¹ A peptide spanning this sequence was shown to bind apo(a) and completely inhibit Lp(a) formation in vitro.²¹ Furthermore, mutation of lysine residues in the apoB4372–4392 sequence impairs Lp(a) assembly in transgenic mice expressing a full length apoB mutant.²²

All of the above studies suggest that the assembly of Lp(a) is complex and requires multiple apo(a)/apoB interactions for the efficient association of the two proteins before formation of the disulphide bond. A proposed model for the assembly of Lp(a) based on current data is shown in Figure 1. Future research in this area will aim to identify the exact amino acid interactions between apo(a) and apoB that are crucial for Lp(a) assembly and may provide new targets for the development of Lp(a)-lowering drugs. The crystal structures for some of the apo(a) KIV domains have recently been solved and will surely aid such an endeavour.^23,24

Figure 1.

Model of Lp(a) assembly. Lp(a) is formed through a two-step assembly mechanism involving initial non-covalent interactions between lysine residues in apoB (possibly Lys680 and Lys4372) and lysine binding domains in apo(a). These initial interactions precede the formation of a disulphide bond between apo(a)Cys4057 and apoBCys4326. N, N-terminus; C, C-terminus.

Lp(a) Metabolism

Despite years of research, the metabolic fate of Lp(a) has proven elusive. Because of its similarity to LDL, it was originally thought that Lp(a) would be cleared via the LDL receptor (LDLR). Some studies of familial hypercholesterolaemia (FH) individuals with mutant LDLRs have supported this hypothesis by showing that FH individuals have higher Lp(a) levels than control subjects.^25–27 In contrast with these findings, an in vivo kinetic study of FH subjects demonstrated that the LDLR was not required for Lp(a) catabolism.²⁸ Animal studies support a role for the LDLR in Lp(a) uptake since mice overexpressing the LDLR gene have an increased clearance of Lp(a)²⁹ and rabbits deficient in the LDLR gene, conversely show elevated Lp(a) levels.³⁰ One major body of evidence that does not support a role for the LDLR in Lp(a) clearance is large clinical trials that report no effect of the statins on Lp(a) levels.^{31, 32} Since the statins work by upregulating expression of the LDLR gene, one would have expected these drugs to be effective Lp(a)-lowering agents if the LDLR played a major role in Lp(a) clearance. In vitro evidence for Lp(a) binding to the LDLR is also lacking with cell culture studies showing that Lp(a) has a low affinity for the LDLR.^{33, 34}

The major site of Lp(a) metabolism from animal studies appears to be the liver ^{35, 36} with some accumulation reported in muscle and spleen. Fragments of apo(a) are found in human urine suggesting that the kidney also plays a role in Lp(a) clearance although it is not a major route for Lp(a) catabolism.³⁷ Cell culture studies have documented that Lp(a) can interact with other members of the LDLR family including the megalin/gp330 and VLDL receptors (reviewed in reference ³⁸); however, the physiological importance of these interactions are unproven. Recently, Kostner et al. have identified an asialoglycoprotein receptor (ASGPR) that is highly expressed in the liver which binds and internalises Lp(a).³⁹ A series of in vivo experiments in hedgehogs and ASGPR knockout mice suggest that this receptor may constitute a major pathway for liver uptake of Lp(a).³⁹

Lp(a) as a Cardiovascular Risk Factor

Many large clinical trials have shown elevated levels of Lp(a) to be an independent risk factor for developing cardiovascular disease.^40–42 A meta-analysis of 5436 CHD subjects from 27 prospective studies concluded that individuals in the upper third of Lp(a) measurements were 70% more likely to develop CHD than those individuals in the bottom third.⁴³ The commonly cited cut-off value for Lp(a) becoming a risk factor is 300 mg/L. Most large cardiovascular studies have evaluated Lp(a) as a modest risk factor with odds ratios around 2 for those individuals above the cut-off level. There is, however, evidence from some studies that the risk of developing CHD from an elevated Lp(a) level is exacerbated in the presence of other lipid risk factors such as high LDL cholesterol or low HDL cholesterol levels.^41,44 This is also the case when elevated levels of Lp(a) are found in combination with thrombogenic risk factors such as Factor V Leiden, protein C deficiency and antithrombin III deficiency.⁴⁵ From this perspective, it may be prudent to aggressively treat modifiable risk factors in individuals that also have elevated Lp(a) levels.

As well as the clinical association of high levels of Lp(a) with CHD, there are multiple studies documenting the deposition of Lp(a) in human arteries affected by atherosclerosis. Immunohistochemical staining for Lp(a) has been documented in the aorta ⁴⁶ as well as coronary,⁴⁷ cerebral ⁴⁸ and peripheral vessels ⁴⁹ with the relative amount of Lp(a) deposition being related to the extent of atherosclerosis.⁴⁸ Studies in rabbits have also demonstrated accumulation of injected human Lp(a) in balloon-injured and atherosclerotic arteries.^50,51 The retention of Lp(a) in the arterial wall is likely due to apo(a)’s affinity for extracellular matrix proteins.⁵² Hughes et al. have shown that Lp(a)’s affinity for the extracellular matrix is mediated via lysine binding sites in apo(a) and that mutating these sites reduces Lp(a) deposition in the vascular wall.⁵³ Lp(a) can be deposited intact or as fragments of apo(a). The generation of apo(a) fragments is most likely from proteolytic cleavage by elastases or metalloproteinases secreted by cells in the arterial wall.⁵⁴ Once deposited, both intact Lp(a) and apo(a) fragments can elicit a range of biological activities (discussed below) that fuel the development of atherosclerosis.

Lp(a) in Animal Models

The use of animals to study Lp(a) is limited by the unusual species distribution of apo(a) which is found only in humans, old world monkeys and the hedgehog.² Interestingly, hedgehog apo(a) is different from the primate version with respect to the duplication of plasminogen kringle domains, indicating an independent evolution.⁵⁵ Both human apo(a) transgenic mice and rabbits have been generated to use as models to study the metabolism and pathogenicity of Lp(a). The first study of apo(a) transgenic mice by Lawn et al. reported significantly larger lesions in the transgenic mice compared to their nontransgenic littermates, when fed an atherogenic diet.⁵⁶ Interestingly, mouse apoB lacked the covalent linkage to human apo(a), yet was still found associated with apo(a) in the lesioned areas, suggesting a non-covalent association between the two proteins.⁵⁶ This association of mouse apoB with human apo(a) has been confirmed by other studies.^57,58 A recent study of apo(a) transgenic mice, using much older animals than previously studied by Lawn, showed that the majority of these animals developed lesions on a chow diet in contrast to non-transgenic littermates reiterating the finding that apo(a) does promote atherosclerosis in mice.⁵⁹

A number of investigators have generated bona fide Lp(a) transgenic mice by crossing the apo(a) transgenic mice with human apoB transgenic mice.^60,61 These animals have higher LDL levels than apo(a) transgenic mice and contain covalently bound Lp(a) in their plasma; a situation more representative of the human one. Atherosclerosis studies of these animals, however, have provided conflicting results. An initial study by Callow et al. showed an 8-fold increase in lesion size in Lp(a) animals fed a high fat diet compared to nontransgenic littermates, with the presence of Lp(a) proving more atherogenic than expression of the apo(a) or apoB transgenes alone.⁶² Two subsequent studies of Lp(a) transgenic mice were at variance with this initial finding. Mancini et al. showed no difference in lesion area between Lp(a) and nontransgenic animals fed a high fat diet.⁵⁷ Sanan et al. reported no effect of apo(a) on atherosclerosis development when Lp(a) transgenic mice were compared to human apoB transgenic mice on a LDLR knockout background.⁶³

In contrast to the transgenic mouse model, expression of apo(a) in transgenic rabbits has consistently proven Lp(a) to be atherogenic. Fan et al. demonstrated that the expression of apo(a) promoted extensive fatty streak lesions in rabbits fed a cholesterol-rich diet.⁶⁴ Furthermore, lesions developed along the length of the aorta, unlike mice in which the lesions are mainly confined to the proximal aorta. The association of apo(a) with the development of atherosclerosis is even more striking when the apo(a) transgene is expressed in the Watanabe Heritable Hyperlipidemic (WHHL) rabbit.⁶⁵ These animals spontaneously develop lesions that are similar in composition to the human “vulnerable plaque” containing a necrotic core of foam cells, proliferated smooth muscle cells and a fibrous cap. The higher LDL levels in rabbits compared to mice and the ability of rabbit apoB to form a covalent linkage with human apo(a), may account for the tighter association between apo(a) expression and atherosclerosis in transgenic rabbits compared to transgenic mice.⁶⁶ Nevertheless, both models have provided useful insight into Lp(a) metabolism and mechanisms that may account for Lp(a)’s pathogenicity.

Pathogenicity of Lp(a)

The physiological role of Lp(a) remains unknown, although a number of possible functions have been proposed (reviewed in reference ⁶⁷). Lp(a) is certainly not essential, since a considerable number of individuals have no detectable Lp(a) with no apparent consequence. Probably the most likely function for Lp(a) is a role in wound healing. This makes biological sense given apo(a)’s affinity for fibrin, its ability to promote cell division and to carry a cholesterol load that could be used for wound repair. Most of the research on Lp(a) has centred around elucidating its role in promoting atherosclerosis. These studies have uncovered an array of biological activities produced by Lp(a) (see Table 1) that could explain its role in the development of CHD.

Table 1.

Pathogenic Activities of Lipoprotein(a)

Lp(a) has an affinity for many components of the subendothelial matrix including proteoglycans,⁶⁸ fibrinogen and fibronectin.⁵² Indeed, studies in rabbits have shown that Lp(a) can be retained in the arterial wall to a greater extent than LDL.⁵¹ Lp(a) also binds to triglyceride-rich lipoproteins, a property which may contribute to the accumulation of lipid in the arterial wall.⁶⁹ Lp(a), like LDL, is subject to oxidative modification to become a substrate for uptake by macrophages thereby fuelling the formation of foam cells.⁷⁰ Also like oxidised LDL, Lp(a) seems to have inflammatory properties promoting the chemotaxis of monocytes ⁷¹ and inducing the expression of vascular adhesion molecules.⁷²

Apo(a)’s similarity to plasminogen has initiated much research into the effect of Lp(a) on fibrinolysis with the hypothesis that it could interfere with plasminogen action and promote thrombosis. Indeed, numerous studies have confirmed that Lp(a) does interfere with plasminogen activation by multiple mechanisms. Early cell culture studies showed that both Lp(a) and apo(a) bound to fibrin with high affinity and that Lp(a) could compete with plasminogen for fibrin binding to impair plasminogen activation.⁷³ Interestingly, smaller apo(a) isoforms have a higher affinity for fibrin than larger isoforms suggesting that smaller isoforms may be associated with a greater risk of thrombosis.⁷⁴ A recent study suggests that as well as binding to fibrin, apo(a) also interacts with tissue plasminogen activator (tPA) and plasminogen within the fibrinolytic complex to alter the kinetics and slow plasmin generation.⁷⁵ Studies of cultured endothelial cells have also shown that Lp(a) can upregulate the expression of plasminogen activator inhibitor-1 (PAI-1), which in turn reduces the amount of tPA available for plasminogen activation.⁷⁶ Other thrombogenic properties reported for Lp(a) include an ability to promote platelet aggregation⁷⁷ and the ability to inactivate tissue factor pathway inhibitor, a major regulator of the coagulation cascade.⁷⁸ Both animal and some human studies have provided evidence that Lp(a) is thrombotic in nature (reviewed in reference ⁷⁹).

Another feature of Lp(a) is its growth-factor-like properties, which is unsurprising considering apo(a) belongs to a growth factor family.⁸⁰ In vitro studies have shown that Lp(a) stimulates the growth of endothelial and smooth muscle cells.^81,82 Lp(a)’s capacity to promote smooth muscle cell proliferation has been a feature in animal studies. Ichikawa et al. documented an increase in smooth muscle cell proliferation in areas of Lp(a) accumulation in transgenic rabbits.⁸³ Furthermore, immunohistochemical analysis of the smooth muscle cell layer showed expression of markers relating to proliferation and de-differentiation. Grainger et al. identified the mechanism by which Lp(a) stimulates smooth cell growth, after demonstrating that Lp(a) inhibited the activation of transforming growth factor-β, a negative regulator of smooth muscle cell proliferation.⁸²

A recent study by Pelligrino et al. has identified yet another mechanism for Lp(a)’s involvement in atherogenesis, by showing that the apo(a) component of Lp(a) induces a rearrangement of actin fibres in cultured endothelial cells.⁸⁴ This leads to a loss of cell to cell contact which may contribute to the initial damage and increased endothelial layer permeability that precedes the development of atherosclerotic lesions.

Lp(a) Measurement

Lp(a) is currently measured by a range of commercially available immunoassays including enzyme-linked-immunoabsorbant (ELISA), immunoturbidometric and immunonephelometric assays that use antibodies specific to the apo(a) moiety of Lp(a). A dilemma exists with most of these methods with respect to the affinity of the antibodies to different apo(a) size isoforms. A considerable number of methods are based on antibodies that have epitopes within the repetitive KIV type 2 kringle. This results in a varying immunoreactivity of the antibody to the samples being measured depending on the number of KIV type 2 repeats. Furthermore, because the calibrator used to standardise the assay will not be representative of the majority of apo(a) isoforms being measured, those samples containing smaller isoforms than the calibrator will be underestimated and conversely those having larger isoforms will be overestimated. This impact of apo(a) size on Lp(a) measurement has been well documented by Marcovina et al. who developed two ELISA assays for Lp(a), one using an antibody, MAb a5, specific to the repetitive KIV type 2 kringle and one using an antibody, MAb a40, specific to the unique KIV type 9 kringle.⁸⁵ Both assays were calibrated with an apo(a) standard containing 21 KIV repeats and then used to measure Lp(a) in a large number of individuals. As expected, identical values were obtained by the two ELISA methods when samples containing apo(a) isoforms with 21 KIV repeats were measured. However, samples with apo(a) isoforms >21 KIV repeats were overestimated, and samples with apo(a) isoforms <21 KIV repeats were underestimated using the MAb a5 assay compared to the MAb a40 assay. The MAb a40 assay has been validated in studies with a large number of individuals.⁸⁶ It has also been used to demonstrate that the apo(a) size dependent bias exists with the majority of commercially available assays.⁸⁷

With such a variety of assays being used and no common calibrator, standardisation of Lp(a) measurement between laboratories is difficult. Storage of samples for Lp(a) measurement can also be an issue, with prolonged storage of samples at sub-optimal temperatures leading to the physical breakdown of the Lp(a) particle and changes in its immunoreactivity.⁸⁸While it is not possible to eliminate all of these differences, a recent report from a NHLBI workshop on Lp(a) provides recommendations on how to reduce Lp(a) measurement inaccuracies.⁸⁹

Lp(a) Levels: Genetic Influences

Plasma concentrations of Lp(a) are highly heritable, hugely variable and largely unaltered by environmental factors. Differences in Lp(a) levels exist between populations, suggesting ethnic differences in the control of Lp(a) levels. For example, African populations have higher levels compared to Caucasian populations.⁹⁰ Within populations, Lp(a) levels can vary over 1000-fold between individuals.⁴ The highly heterogeneous apo(a) gene accounts for much of this variation. The apo(a) size polymorphism clearly has a large influence on Lp(a) levels, as first discovered by Utermann who showed an inverse relationship between the number of KIV repeats and plasma Lp(a) levels.⁵ Cell culture studies have since shown that this relationship is due to a lowered efficiency of secretion of the larger isoforms.⁹¹

Multiple studies have confirmed the inverse relationship between apo(a) size and Lp(a) level suggesting that the apo(a) size polymorphism accounts for the majority of variation in plasma Lp(a) levels depending on the population studied.⁴ However, even subjects with the same-sized apo(a) alleles, can have Lp(a) concentrations that vary by up to 200 fold.⁹² Other sequence variations in the apo(a) gene have been reported to influence Lp(a) levels. These include a pentanucleotide repeat polymorphism in the promoter,⁹³ a C/T polymorphism in the 5′ untranslated region⁹³ and a G to A substitution in the +1 donor site of the KIV type 8 intron, which generates a null allele.⁹⁴ While these sequence variations may account for some of the remaining differences in Lp(a) levels, it is highly likely that other genes involved in the assembly and metabolism of Lp(a) also contribute to the control of Lp(a) levels. Indeed, a recent linkage study identified a significant linkage between plasma Lp(a) levels and a region on chromosome 19 (LOD score 3.8) in a non-Hispanic American population.⁹⁵ Interestingly, the LDLR gene lies in the approximate region identified. Another linkage study of Western European families identified a region on chromosome 1 as having a significant influence on Lp(a) levels.⁹⁶ Future studies will undoubtedly identify other genes that control Lp(a) levels, although their influence is likely to be minor compared to that exerted by the apo(a) gene.

Lp(a) Levels: Other influences

Lp(a) levels can be increased in certain disease conditions such as renal disease.⁹⁷ There is also evidence that Lp(a) behaves as an acute phase reactant, with some studies showing significant increases in Lp(a) levels following tissue injury.⁹⁸ A number of studies have reported associations between Lp(a) and inflammatory cytokines including tumour necrosis factor-α (TNF-α), transforming growth factor-β (TGF-β), interleukin 6 (IL6) and monocyte chemoattractant protein 1 (MCP-1).^99,100 Interestingly, the apo(a) gene contains multiple IL6 response elements¹⁰¹ and cell culture studies have shown that apo(a) gene expression is up-regulated by IL6, leading to an accumulation of Lp(a) particles.¹⁰² Yano et al. using immunohistochemical techniques, showed accumulation of Lp(a) in inflamed tissues, providing support for the notion that Lp(a) is involved in the wound healing process.¹⁰³ An increase in Lp(a) levels with inflammation, combined with Lp(a)’s affinity for extracellular matrix proteins, may well underlie the accumulation of Lp(a) in the artery in the early phases of atherosclerosis.

Lp(a) levels are generally very resistant to changes in diet, although there is evidence that dietary fat lowers Lp(a) levels. Hornstra et al. documented a lowering of plasma Lp(a) levels in individuals placed on diets rich in saturated fat.¹⁰⁴ In keeping with this, Ginsberg et al. reported an increase in Lp(a) levels in individuals after they reduced their saturated fat intake.¹⁰⁵ Monosaturated fats also seem to reduce Lp(a) levels, as shown by a recent study that reported a significant decrease in Lp(a) levels in individuals whose diets were supplemented with almonds.¹⁰⁶ A high intake of polyunsaturated fats may also lower Lp(a). Marcovina et al. documented significantly lower levels of Lp(a) in a population of Bantu fisherman compared to a nearby Bantu population existing on a vegetarian-based diet.¹⁰⁷ This finding was supported by a significant inverse relationship between plasma omega-3 polyunsaturated fatty acids and Lp(a) levels. Calorie restricted diets may favourably influence Lp(a) levels. One study of obese women with high Lp(a) levels did report a significant decrease in levels after 6 weeks on a calorie restricted diet.¹⁰⁸ Finally, the lipid-lowering effect of alcohol also seems to apply to Lp(a).^109,110

Lp(a) Lowering Interventions

Lp(a) levels are generally unresponsive to traditional lipid-lowering drugs, such as the statins or fibrates. One exception is niacin, which has routinely been shown to effectively lower Lp(a) levels, when given in high doses (2 to 3 g/day).¹¹¹ The action of niacin appears to be through reducing Lp(a) production rates. Unfortunately, high doses of niacin can be associated with headaches, flushing and liver toxicity; side effects that can lead to non-compliance. Two less traditional lipid-modifying therapies that effectively lower Lp(a) levels are LDL apheresis and hormone replacement therapy. LDL apheresis has been reported to lower Lp(a) levels by greater than 50%.¹¹² However, this procedure is very costly, time consuming, and usually reserved for people with extreme forms of hypercholesterolaemia, such as FH homozygotes. Trials of post-menopausal women receiving various forms of estrogen replacement therapy have shown quite large decreases in Lp(a) levels particularly in those individuals with high baseline levels.¹¹³ The action of estrogen is likely through a reduction in apo(a) secretion from the liver, since the apo(a) gene contains an estrogen receptor response element and studies in HepG2 cells have shown that estrogen reduces apo(a) expression.¹¹⁴ While estrogen therapy may be an effective Lp(a)-lowering agent, its controversial effect on cardiovascular risk and safety with respect to increasing the risk of certain cancers, will no doubt prevent its use as a routine means to lower Lp(a).

Other agents reported to lower Lp(a) levels include 2 g per day L-carnitine,¹¹⁵ a combination of L-lysine and ascorbate (3g/day of each)¹¹⁶ and the cholestin extract, Xuezhikang (1.2 g/day).¹¹⁷ A recent trial in 37 Japanese patients with high Lp(a) levels (>300 mg/L) has shown that low doses of aspirin (81 mg/day) decrease Lp(a) levels by approximately 20%.¹¹⁸ The aspirin affect is intriguing given Lp(a)’s possible role as an acute phase response protein. Interestingly, aspirin has been shown to reduce apo(a) transcriptional activity in human hepatocytes and interferes with the IL6 enhancement of apo(a) transcription.¹¹⁹ It will be of interest to see if the Lp(a)-lowering activity of aspirin can be repeated in larger scale clinical trials as this may be a cheap and effective means of lowering Lp(a) while reaping the anti-thrombogenic benefits of aspirin therapy. Also worthy of mention as possible Lp(a)-lowering therapies, although still in the experimental stage, are the use of apo(a) anti-sense RNA to reduce apo(a) expression¹²⁰ and the use of synthetic peptides to inhibit Lp(a) assembly.²¹

Conclusions

Four decades of research on Lp(a) have seen it emerge as a clinically important molecule. Evidence has been gained for Lp(a)'s involvement in the development of CHD to a point where routine measurement of Lp(a) in patients at risk must be recommended. Much information has also been gained regarding the genetic control, metabolism and biological activity of Lp(a); however, some questions remain in relation to its assembly, catabolism and interactions with other risk factors. One major challenge that still remains is the development of a therapeutic agent to specifically lower Lp(a) levels. The development of a safe and effective means of lowering Lp(a) will provide the opportunity to conduct intervention trials to further decipher Lp(a)’s contribution to the development of CHD.