Lipoprotein(a) and cardiovascular disease Authors Robert S Rosenson, MD James H Stein, MD Section Editor Mason W Freeman, MD Deputy Editor

Gordon M Saperia, MD, FACC Disclosures

Last literature review version 19.3: Fri Sep 30 00:00:00 GMT 2011 | This topic last updated: Wed Jul 20 00:00:00 GMT 2011 (More) INTRODUCTION Most trials of lipid-lowering therapy for the prevention of cardiovascular disease (CVD) focused on patients with elevated LDL-cholesterol levels. (See "Treatment of lipids (including hypercholesterolemia) in primary prevention".)

Although other dyslipidemias, such as an elevated level of lipoprotein(a), also may promote atherosclerosis, interventions directed toward altering these have only infrequently been evaluated in controlled clinical trials [1]. Elevated serum lipoprotein (a), also referred to as Lp(a), is a risk factor for CVD. There is a causal

relationship between Lp(a) excess and risk for myocardial infarction (MI). (See 'Risk factor versus cause' below.)

This topic will review the genetics, structure and function of Lp(a), as well as its association with atherosclerotic cardiovascular disease. Possible indications for screening and therapy will also be addressed.

STRUCTURE AND FUNCTION Lp(a) is a modified form of low density lipoprotein (LDL) in which a large glycoprotein, apolipoprotein(a) [apo(a)] is covalently bound to apolipoprotein B by a disulfide bridge [2]. The apo(a) chain contains five cysteine rich domains known as "kringles" [3]. The fourth kringle is homologous with the fibrin-binding domain of plasminogen, a plasma protein that dissolves blood clots when activated. Because of this structural similarity to plasminogen, Lp(a) interferes with fibrinolysis by competing with plasminogen binding to molecules and cells. This impairs plasminogen activation, plasmin generation, and fibrinolysis [4,5]. Lp(a) also binds to macrophages via a high-affinity receptor that promotes foam cell formation and the deposition of cholesterol in atherosclerotic plaques [6].

GENETICS Serum Lp(a) levels are primarily genetically determined. In families without familial hypercholesterolemia, greater than 90 percent of the variability in Lp(a) levels can be explained by polymorphisms at the apo(a) gene locus (isoforms), also referred to as the LPA gene (Online Mendelian Inheritance in Man [MIM] 152200) [7]. One

important LPA polymorphism is the kringle IV type 2 size polymorphism, which results in a large number of differently sized isoforms of apolipoprotein(a) [8]. There is a strong inverse relationship between the size of the apo(a) isoforms and the Lp(a) concentrations [7,9-11]. A significant proportion (30 to 60 percent) of the population variation in Lp(a) levels is determined by this polymorphism [12].

The distribution of Lp(a) also varies between racial groups. Lp(a) levels are normally distributed in African-American populations; however, Caucasians, Eastern Asian, and Asian Indian populations have Lp(a) distributions that are skewed towards lower levels [11]. According to the Framingham Heart Study, the 90th percentile of Lp(a) levels is 39 mg/dL (1.39 mo/L) in men and 39.5 mg/dL (1.41 mo/L)

in women (units of mass) [13,14]. Blacks have somewhat higher values than whites [15]. (See 'Measurement of serum Lp(a) concentration' below.)

Measurement of serum Lp(a) concentration Lp(a) initially was identified as a "sinking" pre-beta lipoprotein band detected by gel electrophoresis [16]. The standard method for measuring Lp(a) is density gradient ultracentrifugation. ELISA techniques are now widely availabhttp://bolko.wordpress.com/le; however, several of these methods are insensitive, unable to distinguish between apo(a) isoforms, and crossreact with plasminogen, which can result in significant underestimation or overestimation of Lp(a) levels [15,17]. A state-of-the art,

commercially available assay uses a latex-enhanced immunoturbidimetric method that measures lipoprotein(a) independently of the apolipoprotein(a) size and number of kringle-IV repeats [18,19]. Lp(a) levels are stable in individuals with time [20].

CVD RISK An association between Lp(a) excess (Lp(a) level above the 95th percentile) and CVD, particularly coronary heart disease (CHD), was initially suggested by relatively small cross-sectional and retrospective epidemiologic studies [13,14,21-36]. The largest prospective analysis (2047 patients and 3921 controls) of the association between Lp(a) excess and incident CHD found an odds ratio of 1.60 (95% CI 1.38-1.85) between the upper and lower thirds of baseline Lp(a) levels after adjustment for traditional cardiovascular risk factors [37].

The best data that summarize the relationship between Lp(a) and CVD come from a 2009 meta-analysis of the individual patient records of over 120,000 participants in 36 prospective studies. In this study the association between Lp(a) and CHD risk was continuous [20]. In the 24 cohort studies in the meta-analysis, the rates of CHD were 5.6 and 4.4 per 1000 person-years in the top and bottom thirds of baseline Lp(a) distributions. After adjustment for multiple traditional risk factors, the risk ratio for CHD was 1.13 (95% CI 1.09-1.18); this is a lower risk ratio than found in the largest single study cited above [37]. The increase in risk with higher Lp(a) was continuous, as has been shown in prior studies [20,37].

Lp(a) excess is commonly detected in patients with premature CHD (figure 1). In one study, for example, Lp(a) excess was present in 18.6 percent of patients with premature CHD, 12.7 percent of whom had no other dyslipidemia [38].

Lp(a) is also associated with unstable angina and the presence of complex coronary lesions, suggesting a possible role in plaque rupture and coronary thrombosis [22]. In patients with an acute coronary syndrome, Lp(a) concentrations are predictive of an increased risk of cardiac death. One study found that among 266 patients with an acute myocardial infarction, a Lp(a) concentration 30 mg/dL (1.07 mo/L) was associated with a 62 percent increase in cardiac death during a three-year follow-up (29.8 versus 18.6 percent for concentrations <30 mg/dL [(1.07 mo/L]) [39]. Among 197 patients presenting with unstable angina, levels 7.9 mg/dL (0.28 mo/L) were predictive of subsequent cardiac death (relative risk 2.48).

In a systematic review of 40 studies involving 58,000 participants, the relative risk from CHD was 2.08 (95% CI 1.67-2.58) for those with small apo(a) isoforms (22 or fewer kringle IV type 2 repeats), compared to those with large isoforms [40].

Cerebrovascular disease Lp(a) levels are associated with cerebrovascular disease, a relationship that is stronger in men than in

women [24,28,41]. A meta-analysis that included data from 56,010 subjects with >4609 stroke events found that elevated Lp(a) was a risk factor for stroke, with the strongest evidence coming from prospective cohort studies [42]. In the two available meta-analyses, the range of risk comparing the highest to the lowest tertiles was 1.10 to 1.22 [20,42].

In a systematic review including six studies that evaluated ischemic stroke as an outcome, the relative risk was 2.14 (95% CI 1.85-2.97) [40].

Patients with hypertension Lp(a) may also play a role in the development of target-organ damage in patients with essential hypertension. In one study, 277 patients with untreated hypertension were compared with 102 healthy controls. Lp(a) levels were the best predictor of target-organ damage involving the kidney, heart, and arterial wall [43]. This relationship was independent of the level of blood pressure. Another independent factor was a higher frequency of low-molecular-weight apo(a) isoforms.

Risk factor versus cause The continuous positive associations between Lp(a) and CVD risk reviewed above suggest that Lp(a) is an independent risk factor. Evidence that it is causative comes from two gene association studies [8,44].

In the first of these, Lp(a) levels, myocardial infarctions, and the frequency of the number of 5.6-kilobase repeats determined by a polymorphism in the Lp(a) gene (which has been inversely related to Lp(a) levels in past studies) were compared in over 30,000 individuals enrolled in three cohorts of Danish citizens [8]. (See 'Structure and function' above and "Glossary of common biostatistical and epidemiological terms", section on 'Mendelian randomization'.)

The following findings were noted:

Multivariable-adjusted hazard ratios (HR) for MI with increasing Lp(a) levels were 1.2, 1.6, 1.9, and 2.6 events per 10,000 person-years between the 22nd and 66th, 67th and 89th, 90th to 95th percentiles, and greater than the 95th percentile respectively, compared to less than the 22nd percentile. This trend was statistically significant. Mean Lp(a) levels were 56, 31, 20 and 15 mg/dL (1.99, 1.10, 0.71, and 0.53 mo/L) for the first, second, third, and fourth quartiles of the number of 5.6 kilobase repeats. This trend was statistically significant. The multivariable-adjusted HRs for MI increased significantly as the frequency (by quartile) of the number of 5.6 kilobase repeats. For example, in one of three Danish cohorts, the HR for MI were 1.5, 1.3, and 1.1 events per 10,000 person-years for individuals in the first, second, and third quartiles respectively, compared with individuals in the fourth quartile.

A doubling of Lp(a) levels was associated with a 2.2 percent increase in the risk of MI, using a sophisticated statistical analysis. A second gene association study evaluated the associations between Lp(a) levels, two common variant chromosomal regions (single nucleotide polymorphisms) near the gene locus that code for apolipoprotein (a), and the presence or absence of coronary artery disease (3145 cases and 3352 controls) [44]. The following findings support a causal role for Lp(a):

The two single nucleotide polymorphisms explained approximately 36 percent of the overall variance in plasma Lp(a) levels. Both variants were strongly and significantly associated with an increased level of Lp(a) Both variants were strongly and significantly associated with an increased risk of coronary artery disease (CAD). Using a meta-analysis technique, the odds ratio for CAD with one variant present was 1.51, 95% CI 1.38-1.66 and for both variants 2.57, 95% CI 1.80-3.67. After adjustment for the Lp(a) level, the association between the presence of the variants and the risk of CAD was no longer present. The observations from these two gene associate studies are suggestive of a causative relationship between higher Lp(a) and an increased risk of coronary artery disease events. Causality could be proven with a randomized trial showing benefit from a lowering in the risk of events with therapy targeted at reducing Lp(a) levels.

MECHANISMS OF CVD RISK

Atherothrombosis Lp(a) excess may promote atherosclerosis by the following mechanisms:

The VLDL receptor found on the macrophages present in atherosclerotic lesions can bind to and mediate the catabolism of Lp(a) by endocytosis, leading to its degradation within lysosomes [45]. This could lead to cellular accumulation of lipid within macrophages. Supporting this hypothesis is the observation that Lp(a) is ubiquitous in human coronary atheroma, colocalizes with plaque macrophages, and is detected in larger amounts in tissue from culprit lesions in patients with unstable compared to stable coronary disease [46]. Binding to the endothelium and components of the extracellular matrix [47] as well as endothelial dysfunction due to selective impairment of vasodilator capacity of receptor-mediated endothelial stimuli [48]. The importance of the latter effect is uncertain since Lp(a) may not impair nitric oxide-mediated vasodilation, in contrast to the demonstrated adverse effect of oxidized LDL [49]. Increased expression of intercellular adhesion molecule-1, resulting in the recruitment of monocytes to the vessel wall and binding to macrophages [50]. This can promote foam cell formation and the localization of Lp(a) in atherosclerotic plaques [6,51,52] Enhancing the susceptibility of LDL-cholesterol to oxidative modification [53,54]

Interaction with the fibrinolytic and coagulation systems leading to increased tissue-factor mediated thrombosis and inhibition of clot lysis [55]. There may also be an association between fibrinogen, apolipoprotein B concentrations, and Lp(a) levels [56]. Also, another study found that the atherogenic effects of Lp(a) may be enhanced or partially mediated through the formation of circulating immune complexes containing Chlamydia pneumoniae-specific IgG antibodies [57]. (See "Chlamydophila (Chlamydia) pneumoniae infection as a potential etiologic factor in atherosclerosis".)

Not all studies have found an association between Lp(a) and atherogenesis. In a cohort consisting primarily of hypertensive patients, Lp(a) levels did not predict coronary artery calcification either in the cohort as a whole or within groups stratified by conventional CHD risk factors [58].

Restenosis The effect of Lp(a) on lipid metabolism, the coagulation and fibrinolytic systems, and the stimulation of smooth muscle cell proliferation suggests that Lp(a) may play a role in restenosis after angioplasty or stenting. However, the data are conflicting [59-61].

Myocardial infarction due to vasospasm The incidence of a myocardial infarction in patients with coronary artery vasospasm is low; the mechanism may be the triggering of thrombus formation and its

slow removal due to impaired fibrinolysis. (See "Variant angina".) A possible role of Lp(a) in the occurrence of MI in patients with coronary vasospasm was evaluated in one report of 77 patients with coronary spasm, including 16 with a prior MI [62]. The Lp(a) concentration in these patients was higher than in 81 normal controls, but lower than in 177 patients with coronary artery disease. The incidence of high concentrations of Lp(a), defined as >25 mg/dL (0.89 mo/L), was higher in the patients with coronary spasm who had a MI (56 versus 21 percent in those without an infarction); patients with higher Lp(a) levels had a higher incidence of prior infarction compared to those without high Lp(a) levels (41 versus 13 percent).

SCREENING At present, screening and treatment for Lp(a) excess levels should only be considered for:

Patients with CHD and no other identifiable dyslipidemia Patients with a strong family history of CHD and no other dyslipidemia Patients with hypercholesterolemia refractory to therapy with LDL cholesterol lowering therapies [63] The last recommendation is based upon the observation that in some patients, Lp(a) excess may account for significant component of the LDL-cholesterol level calculated by the Friedewald formula, and that standard LDL-cholesterol lowering therapies do not lower Lp(a) levels appreciably.

TREATMENT Clinical trials related to the prevention of CHD have not evaluated the effects of specific Lp(a)-lowering therapies, thus all recommendations for treatment are speculative. The primary goal of therapy for Lp(a) excess (Lp(a) concentration above the 95th percentile) is LDL-C reduction to the patient's target LDL-C concentration based upon risk factors (table 1) [63]. (See "ATP III guidelines for treatment of high blood cholesterol".)

Aggressive LDL-C reduction Some clinicians recommend treating LDLC more aggressively in the presence of Lp(a) excess, however the effectiveness of this approach has not been formally studied. In a posthoc analysis of the Familial Atherosclerosis Treatment Study, elevated Lp(a) levels were associated with progression of coronary atherosclerosis and CHD events ONLY IF the LDL-C level was not reduced by more than ten percent [64]. In a post-hoc analysis of the Familial Hypercholesterolemia Regression Study of LDL apheresis, Lp(a) reduction did not provide additional angiographic benefit if LDL-C lowering therapy reduced the LDL-C to <130 mg/dL [65]. These studies, although small and not designed to address treatment of Lp(a) excess, provide some evidence that Lp(a) is less important of a CHD predictor when LDL-C has been treated.

Lp(a) reduction If the LDL-C concentration cannot be reduced to the optimal LDL-C level for an individual patient, use of medications that lower Lp(a) (usually with nicotinic acid) may be initiated. This approach will lower Lp(a) levels and further reduce LDL-C, but this approach has

not been shown to improve important outcomes such as death or myocardial infarction.

As reviewed below, Lp(a)-lowering is usually accomplished using nicotinic acid, which also reduces LDL-cholesterol and has salutary effects on other aspects of the lipoprotein profile, in addition to Lp(a) reduction.

If the decision is made to lower Lp(a) levels, one possible therapeutic goal is to lower Lp(a) cholesterol levels measured by ultracentrifugation to less than 30 mg/dL (1.07 mo/L) in white patients, a level previously associated with increased CHD risk [63,66]; a goal of less than 20 (0.71 mo/L) mg/dL may be appropriate when Lp(a) protein levels are measured by ELISA. (See 'Measurement of serum Lp(a) concentration' above.)

Lipid-lowering drugs With respect to lipid-lowering drugs:

Statins and bile acid sequestrates do not reduce Lp(a) levels [35,63]. Lp(a) levels may increase slightly on statins, however this has not been shown to mitigate their beneficial effects on CHD event reduction. Most fibric acid derivatives do not lower Lp(a) levels, except for bezafibrate which may produce up to a 39 percent reduction [67-70]. Bezafibrate is not approved for use in the United States.

Estrogen replacement therapy (ERT) reduces Lp(a) levels by up to 50 percent [71,72], an effect that was somewhat mitigated by concomitant progesterone therapy in some reports [71-73] but not the PEPI trial (figure 2) [74]. However, the clinical role for HRT is uncertain and it is not recommended for CVD risk reduction. (See "Postmenopausal hormone therapy and cardiovascular risk".) Nicotinic acid (2 to 4 g/day) is the most effective therapy for reducing Lp(a) levels [75,76]. In addition to lowering Lp(a) levels, it has multiple, salutary effects on the lipoprotein profile including reduction in LDLcholesterol, apo B-100, small LDL, and triglycerides, it raises HDLcholesterol levels and in combination with statins, appears to reduce CHD and CHD risk. Nicotinic acid reduces Lp(a) levels by as much as 38 percent [37,77]. Neomycin (2 to 3 g/day) also lowers Lp(a) levels by as much as 24 percent, but has many side effects and in general should not be used to lower Lp(a) [78]. Combined therapy with neomycin and niacin lowers Lp(a) and LDL-C levels by 45 percent [78]. Apheresis The efficacy of Lp(a) apheresis has only been investigated in small observational studies. A 2007 report reviewed many of these studies [77]. The authors suggested that a treatment goal of less than 30 mg/dL (1.07 mo/L) was reasonable if a patient was receiving apheresis. (See "Primary disorders of LDL-cholesterol metabolism", section on 'Management' and "Treatment of drug-resistant hypercholesterolemia", section on 'LDL apheresis'.)

Aspirin A report from the Womens Health Initiative demonstrated that carriers of a minor allele of apo(a) (rs3798220) had elevated Lp(a), doubled cardiovascular disease risk, and appeared to benefit more from use of aspirin than noncarriers [79]. After 9.9 years of follow-up, individuals with this minor allele had a nearly doubled risk of major cardiovascular disease events (hazard ratio 2.11, 95% CI 1.39-2.52). However, the risk was more than halved by use of aspirin (HR=0.44, 95% CI 0.20-0.94). The risk was not significantly reduced among noncarriers, and the interaction between carrier status and aspirin use was statistically significant [79].

SUMMARY AND RECOMMENDATIONS Lipoprotein(a), also referred to as Lp(a), is a modest, independent risk factor for atherosclerotic CVD events, especially MI. (See 'Risk factor versus cause' above.)

There are no clinical trials that have adequately tested the hypothesis that Lp(a) reduction reduces the incidence of first or recurrent CVD events. Therefore, widespread screening for Lp(a) excess is not indicated and treatment of Lp(a) excess should only be considered in specific circumstances, when the treating clinician believes that the possible CVD benefits of Lp(a) reduction outweigh the potential risks of therapy.

Based upon the available evidence we suggest targeted screening of the following types of individuals:

Patients with CVD and no other identifiable and treatable dyslipidemia Patients with recurrent CVD events despite adequate risk factor treatment Patients with a strong family history of CHD and no other dyslipidemia When Lp(a) excess (Lp(a) concentration above the 95th percentile) is identified, more aggressive LDL-C targets than those recommended for the general population may be considered. This is based upon the knowledge that there may be modest, excess residual CVD risk associated with high Lp(a) levels. Therapy with a statin or nicotinic acid or both may be needed to achieve these goals. (See "Intensity of lipid lowering therapy in secondary prevention of coronary heart disease", section on 'Summary and recommendations'.)

The following suggestions for treatment of Lp(a) excess are speculative and await confirmation in clinical trials:

We suggest the addition of nicotinic acid for individuals on maximal dose statin therapy who have not achieved LDL-C targets (Grade 2C). If such an approach is chosen, we titrate niacin to the maximally tolerated recommended dose in an attempt to lower LDL-C to target. (See "Lipid lowering with drugs other than statins and fibrates", section on 'Nicotinic acid (Niacin)' and 'Treatment' above.)

We suggest the use LDL-C lowering therapy for individuals with a family history of premature cardiovascular disease whose LDL-C does not make them candidates for a statin based on national guidelines (Grade 2C). In this situation, we support an optional target of LDL-C <70 mg/dL. A statin is preferred, and niacin may be added if necessary to reach this lower goal. (See 'Treatment' above.) Use of UpToDate is subject to the Subscription and License Agreement. REFERENCES