Lipoprotein-Associated Phospholipase A₂ as a Novel Risk Marker for Cardiovascular Disease

Abstract

We sought to critically assess the role of lipoprotein-associated phospholipase A₂ (Lp-PLA₂) in the prediction of cardiovascular events in primary and secondary prevention settings. The inclusion criteria for our study included population-based epidemiologic studies and the presence of clinical outcomes of interest, including atherosclerotic disease, coronary events, stroke, and cardiovascular death. Studies that lacked clinical outcomes or that involved animals were excluded. We included primary and secondary prevention studies of subjects in all ethnic groups and of either sex, with no age limitation. We searched MEDLINE, Google Scholar, and the Cochrane Library for studies with publication dates from January 1970 through July 2009, and we searched major cardiology meeting abstracts from 2000 through 2009. From each study, we used predictive ability—including relative risk, hazard ratio, odds ratio, and prevalence of high Lp-PLA₂ levels, with adjustment—along with baseline population characteristics.

Of 33 studies that met our inclusion criteria, 30 showed a significant association between Lp-PLA₂ and cardiovascular events. Most of the studies had been adjusted for major Framingham risk factors and other variables that might influence the effect under question. After multivariate adjustments in cohort and nested case-control studies, increased levels of Lp-PLA₂ remained a significant predictor of cardiovascular events. The available body of evidence suggests that Lp-PLA₂ is a reliable marker of risk for cardiovascular events.

Traditional risk factors can predict the risk of cardiovascular (CV) disease in many, but not all, patients. About 10% to 20% of persons with coronary heart disease (CHD) have no identifiable risk factor,¹ and 35% of CHD deaths occur in patients who have total serum cholesterol levels of less than 200 mg/dL.²

Current risk-prediction methods (such as the Framingham risk table) are not able to accurately pinpoint which patient will develop a CV event, or at what time. Exhaustive research is under way to improve our ability to determine individuals' CV risks. Inflammation plays an important causal role in the development of atherosclerosis and its clinical complications³ (such as acute coronary syndrome [ACS], sudden cardiac death, acute myocardial infarction, unstable angina, and stroke). New diagnostic tests are being developed to identify individuals who are at high risk of inflammation.⁴ Several inflammatory markers, notably high-sensitivity C-reactive protein (hs-CRP)⁵ and white blood cells,⁶ have been studied as predictors of future CV events.

Lipoprotein-associated phospholipase A₂ (Lp-PLA₂) is a novel inflammatory marker that has been the recent focus of multiple epidemiologic studies involving primary and secondary prevention populations.⁷ A growing body of evidence suggests that Lp-PLA₂ is a CV risk marker independent of traditional risk factors like high-density-lipoprotein cholesterol (HDL-c), low-density-lipoprotein cholesterol (LDL-c), and hs-CRP. Lipoprotein-associated phospholipase A₂, a 45-kDa protein with 441 amino acids, was cloned in 1995.⁸ The gene for Lp-PLA₂ is located on chromosome 6p 21.2 to 12. Also known as platelet-activating factor acetylhydrolase (PAF-AH), Lp-PLA₂ hydrolyzes the short acyl group at the Sn-2 position of the phospholipids in oxidized LDL, which leads to production of the pro-inflammatory compounds lyso-phosphatidylcholine and oxidized nonesterified fatty acids.⁹ Lipoprotein-associated phospholipase A₂ is produced by several inflammatory cells (for example, monocytes, macrophages, T cells, hepatic Kupffer cells, and mast cells). In human plasma, it is associated mainly (70%–80%) with LDL particles, due to its affinity for the C-terminus of apolipoprotein B; the remaining 20%–30% is associated with HDL and other lipoproteins.¹⁰ Within LDL fractions, it shows more predilection for the denser LDL-4 and LDL-5, which are believed to be highly proatherogenic.¹¹ Studies have also shown that oxidized phospholipids are chemotactic and promote monocyte–endothelial-cell interaction.^12,13 A high level of circulating oxidized LDL has been shown to increase plaque vulnerability and mediate endothelial dysfunction, which can partly explain its higher levels in ACS patients.^14,15

Recent reports suggest that human beings and rabbits with elevated Lp-PLA₂ have higher plaque burden in the coronary arteries.¹⁶ The reduction of atherosclerotic lesions in hyperlipidemic rabbits, by inhibiting Lp-PLA₂, supports the notion that Lp-PLA₂ enhances atherosclerosis.¹⁶ Lipoprotein-associated phospholipase A₂ is also expressed within the necrotic core and apoptotic macrophages of vulnerable and ruptured plaque in human coronary atheroma, which suggests its active role in acute coronary syndrome.¹⁷ Moreover, high levels of Lp-PLA₂ mRNA have been noted in carotid plaque.

During myocardial ischemia, membrane phospholipids may undergo hydrolysis under stress, leading to accumulation of arachidonic acid and lysophophatidylcholine.^18,19 Lipoprotein-associated phospholipase A₂ levels are shown to be up-regulated by endotoxins.^20,21 The proatherogenic contribution of Lp-PLA₂ has been much studied; however, its anti-inflammatory and antiatherogenic potential should not go unnoticed.⁷

Platelet-activating factor is prothrombotic and a mediator of inflammation; platelet-activating factor acetylhydrolase hydrolyzes PAF, thereby negating its proinflammatory and prothrombotic effects.²² The opposing actions of Lp-PLA₂ (PAF-AH) have led to a debate concerning which action is more potent and therefore warrants more attention. Initially, PAF-AH was thought to be atheroprotective by reducing platelet-activating factor.^23,24 However, subsequent studies have shown its role in inflammation: lysophosphatidyl choline stimulates macrophage proliferation, up--regulates cytokines and CD40 ligands, and increases the expression of vascular adhesion molecules.^25,26 This has led most researchers to believe that the proatherogenic effect of Lp-PLA₂ has a greater importance in CV disease prevention and management.

Because of its low biologic fluctuation, high vascular specificity, and indication (at elevated levels) of plaque inflammation and endothelial dysfunction, Lp-PLA₂ is useful in identifying patients at high risk of CHD. A recent report²⁷ suggests that an Lp-PLA₂ level >235 ng/mL in healthy populations and >225 ng/mL in clinical populations is a high-risk marker for CV events, and that Lp-PLA₂ should be measured in patients who are at moderate to very high risk of CV events. An elevated Lp-PLA₂ level also predicts increased risk of myocardial infarction and stroke in population studies that are fully adjusted for other CV risk factors. More epidemiologic and clinical studies in the future can help clarify the effect of Lp-PLA₂ on atherosclerosis-related CV risk.

Secretory PLA₂

Secretory PLA₂ (sPLA₂)—another enzyme from the phospholipase A₂ family—has been the focus of many studies for its role in predicting coronary artery disease (CAD).²⁸ Secretory PLA₂ is an acute-phase reactant, the activity of which is up-regulated by other inflammatory markers.²⁶ Unlike Lp-PLA₂, its levels correlate with hs-CRP. Secretory PLA₂ is not widely used in clinical practice; one explanation for this could be its nonspecific predictive ability. However, ongoing and future studies will help shed more light on the predictive potential of sPLA₂ and its potential in comparison with Lp-PLA₂.

The purpose of this particular review is to critically assess the role of Lp-PLA₂ in the prediction of CV events (with a focus on coronary heart disease) in primary and secondary prevention settings.

Methods

The protocol for this review of the literature was designed prospectively to define study objectives, search strategy, patient populations of interest, study-selection criteria and methods for determining study eligibility, outcomes of interest, data elements to be abstracted and methods for abstraction, and methods for the assessment of study quality. Our review included studies on subjects of either sex, with no age limitation. Studies in all ethnic groups are included. Primary and secondary prevention studies are included.

Inclusion Criteria. In order to be included, all studies had to be population based and epidemiologic; further, they had to manifest clinical outcomes of interest (that is, atherosclerotic disease, coronary events, stroke, and CV death). We excluded studies that did not meet these inclusion criteria, as well as animal studies.

MEDLINE®, Google™ Scholar, the Cochrane Library, and the abstract books of major cardiology meetings were searched. Additional studies were sought and identified by reviewing the references of included articles. Figure 1 depicts the flow chart of the study's manuscript identification and selection process.

Fig. 1 Flow chart of the study's manuscript identification and selection process.

The MEDLINE database was searched using PubMed (National Library of Medicine). The search included all articles published from January 1970 through July 2009 that included at least 1 of the following key words in their titles, abstracts, MeSH headings and subheadings (including single words and phrases), or keyword lists:

lipoprotein-associated phospholipase A2 or platelet-activating factor acetylhydrolase (alternatively: Lp-PLA2 or PAF-AH)

AND

cardiovascular diseases or coronary disease or myocardial infarction or stroke or sudden cardiac death

The PubMed search was limited to “humans” and to publication dates from 1970/01/01 to 2009/07/29.

Abstracts presented at the 2000–2009 meetings of the American Heart Association, American College of Cardiology, and European Society of Cardiology were also searched using the keywords lipoprotein-associated phospholipase A2 and platelet-activating factor acetylhydrolase; then the searched results were screened, if they included CV events.

Each study was initially evaluated for its potential to cover any of the topics of interest and was studied in regard to all potentially relevant study-selection criteria. We included prospective and retrospective cohort studies, case-control studies, and nested case-control and case-cohort studies.

Description of Studies

From each study, we abstracted the predictive ability (for example, relative risk [RR], hazard ratio [HR], and odds ratio [OR]) and the prevalence of high Lp-PLA₂ levels, along with baseline population characteristics and the effects of adjustment for established coronary risk factors.

Critical Appraisal of Selected Studies

The quality and validity of the studies were evaluated using the Critical Appraisal Tools guidelines developed by the Public Health Resource Unit of the United Kingdom National Health Services.²⁹ The same critical appraisal tools were used to assess the cohort studies, the nested case-control studies, and the case-cohort studies.

Results

Total retrieval from our PubMed search was 301 references at the time of the study. After searching all described databases regarding the relationship between Lp-PLA₂ and CV disease, we were left with 33 articles (18 cohort, 6 nested case-control or case-cohort, and 9 case-control studies) that met the study-selection criteria for inclusion in this systematic review.

A total of 52,995 subjects participated in 33 studies. Most of the participants were men, with an approximate age range of 21 to 84 years. The participants represented multiethnic backgrounds, and the studies were conducted in multinational centers. The follow-up period ranged from as short as 42 days to as long as 16 years. Sixteen studies assessed Lp-PLA₂ mass assay, 11 studies assessed Lp-PLA₂ activity assay, and 6 studies assessed both.

For the cohort and nested case-control studies, we abstracted data on the number of enrolled subjects, number of cases and controls, demographics of study subjects, definition of cases, list of Framingham risk factors for which the analyses were adjusted, list of other variables included in multivariate analyses, duration of follow-up, and major results after full adjustment for the described variables. After adjustment in cohort and nested case-control studies, the risk ratio and ORs ranged roughly between 1.40 and 2.00, provided that the relationship was statistically significant.

For case-control studies, we abstracted data on the number of cases and controls, demographics of study subjects, definition of cases and controls, list of Framingham risk factors for which the analyses were adjusted, list of other variables included in multivariate analyses, and major statistical findings (after full adjustment, when applicable).

These studies are abstracted in Tables I, II, and III.^30–62

TABLE I. Prospective Cohort Studies of Lipoprotein-Associated Phospholipase A₂ (Lp-PLA₂) and Cardiovascular Events

TABLE I. continued

TABLE I. continued

TABLE I. continued

TABLE II. Nested Case-Control and Case-Cohort Studies of Lipoprotein-Associated Phospholipase A₂ (Lp-PLA₂) and Cardiovascular Events

TABLE III. Case-Control Studies of Lipoprotein-Associated Phospholipase A₂ (Lp-PLA₂) and Cardiovascular Events

TABLE III. continued

Discussion

Although our case-control studies support a relationship between Lp-PLA₂ and CV disease, the evidence from cohort and nested case-control studies is more reliable and provides more robust evidence in support of such a relationship. Because data from a good number of cohort and nested case-control studies are available, we focus our discussion on the data from these studies.

Most of the studies showed a positive association between Lp-PLA₂ and CV events, but a few did not. One such study, by Blake and colleagues,⁴⁹ failed to show a relationship between Lp-PLA₂ levels and CV incidents; this may have been due to lack of power to detect such a relationship, because this study included the lowest number of subjects, and no power calculation was provided by the authors. The study included only middle-aged women, with the confounding effect of hormonal therapy, and the participants were few in number (123 cases, 123 controls). Lower-risk women were also included in this study. However, women with very high levels of Lp-PLA₂ had increased risk of future CV events, as noted in the other trials. Oldgren and colleagues,⁵⁹ in FRISC II and GUSTO IV, failed to find an association, in ACS patients, between Lp-PLA₂ mass measurement and future CV events, including death, within 6 months to 1 year of follow-up. Similar results were found when measured early after an ACS event in the PROVE IT–TIMI 22 trial³¹; but at 30 days after the event, Lp-PLA₂ activity in the highest quintile was associated with increased risk of additional CV events, when compared with the lowest quintile (P=0.002). Finally, a recent case-control study failed to show any difference in Lp-PLA₂ activity between patients with CAD and without CAD.⁶²

Limitations of the aforementioned negative studies include retrospective analysis of patients, reduction in Lp-PLA₂ activity during the 1st few days after ACS,⁶³ decrease in Lp-PLA₂ activity by statin use, Lp-PLA₂ measurement in frozen plasma, and effect of long-term storage on the level of Lp-PLA₂. We should keep in mind, though, that a bias toward publishing positive findings and toward withholding negative findings might have prevented the publication of other studies with negative findings. Limitations in some of the positive-association studies include small sample size, retrospective nature, and the study of stored, frozen samples. However, recent studies have shown stability of Lp-PLA₂ in properly stored samples.

All other published studies that we included in our analysis showed a significant association between Lp-PLA₂ serum levels and CV events. These studies described the size of the effect by different measures of association, such as comparing HRs for the top tertiles or quartiles with those for the reference bottom tertiles or quartiles, or comparing HRs for 1 standard deviation⁶⁴ increase in Lp-PLA₂. Most of the studies did adjust for major Framingham risk factors and other variables that are believed to influence the effect in question, especially LDL-c and CRP. After investigators fully adjusted for potential confounders in cohort and nested case--control studies, the RRs and ORs ranged roughly between 1.40 and 2.00, provided that the relationship was statistically significant. This range of effect size is in accordance with the effect size observed in studies of newly emerging coronary risk markers. Lipoprotein-associated phospholipase A₂ significantly retained its independence for prediction of CAD and also, in some cases, for prediction of ACS, after adjusting for LDL-c, CRP, and other variables; in some studies, however, the association weakened after adjustment. Lipoprotein-associated phospholipase A₂ levels were in positive correlation with total cholesterol and LDL-c, while most of the studies showed a negative correlation with HDL. Lipoprotein--associated phospholipase A₂ showed no correlation with CRP levels, except for 2 studies that showed a weakly positive correlation.^40,44 Folsom and colleagues⁶⁵ studied the influence of inflammatory markers, in addition to traditional risk factors, in the prediction of CAD. The traditional risk-factor model showed CAD prediction with a receiver operating characteristic area under curve of 0.8, but the addition of Lp-PLA₂ showed a statistically significant increase in the area under curve. Notably, hs-CRP—an acute-phase reactant that increases both in obesity and in acute and chronic systemic inflammation—did not significantly increase the area under curve. Recently, it has been suggested that hs-CRP may be more than a simple marker of atherosclerosis. It can bind to oxidized LDL and activate the complement.^66,67 When compared with hs-CRP, Lp-PLA₂ in some studies is suggested to be a more promising marker of risk prediction. Winkler and associates⁴¹ showed that, in moderate-risk patients (Framingham risk, 10%–20%) with hs-CRP <3 mg/L, increased Lp-PLA₂ levels doubled the HR for cardiac death. Similarly, the WOSCOPS study⁴⁸ showed that in multivariate analysis, Lp-PLA₂ was significantly associated with risk, compared with hs-CRP (P=0.005 vs P=0.09). However, further trials are needed to confirm an incremental prognostic value of Lp-PLA₂ over hs-CRP. While Lp-PLA₂ at this time could be speculated to have better specificity than CRP, evidence of better risk prediction is increasing when the two complement one another. The ARIC trial⁵¹ showed that low-to--moderate levels of both markers in combination significantly increase the risk, in comparison with a combination either of high Lp-PLA₂ and low CRP or of high CRP and low Lp-PLA₂. The MONICA study⁵⁰ also showed better risk prediction when high Lp-PLA₂ and high CRP were combined, when compared with each used alone (HR, 1.93; 95% confidence interval, 1.09–3.4). The IHCS trial⁴³ revealed better prediction of CAD death with combinations of high CRP/low Lp-PLA₂ and high CRP/high Lp-PLA₂—but not with combinations of low CRP/low Lp-PLA₂ and low CRP/high Lp-PLA₂.

The association of Lp-PLA₂ and CV events was consistent across a wide variety of subjects of both sexes, from different ethnic backgrounds and different countries. This relationship was observed in both primary and secondary prevention studies (that is, in subjects without and with clinically apparent disease at baseline). A dose–response relationship was observed in multiple studies, with higher risks associated with higher tertiles or quartiles of Lp-PLA₂, after adjustment for the full range of known confounders. The absolute cutoff value of Lp-PLA₂ is not clear at this time. It is different in women, smokers, and the elderly, compared with men, nonsmokers, and the young. Because these studies used different cutoff points, it is impossible (and outside the scope of this study) for us to determine the best cutoff value for Lp-PLA₂ for the purpose of CV risk prediction. A consensus panel has recently proposed a set of cutoff points for interpreting Lp-PLA₂ levels in clinical settings.⁶⁸ According to the recommendations, a cutoff point of 235 ng/mL for Lp-PLA₂ mass concentration should be lowered to 200 ng/mL, values above which would place patients in a high-risk category. Whether lowering the cutoff level of Lp-PLA₂ will accord with the actual risk in these patients has not been proved in clinical trials.

Lipoprotein-associated phospholipase A₂ is an enzyme that can be measured by either its mass or its activity.⁷ Sixteen studies included in this review calculated mass and 12 studies calculated activity, while 6 calculated both mass and activity. There is still controversy about the method of estimating Lp-PLA₂ level. Only modest correlation has been found between Lp-PLA₂ activity and Lp-PLA₂ mass. Koenig and colleagues³² showed that Lp-PLA₂ mass in multivariate analysis predicted increased risk of future CV events, when compared with Lp-PLA₂ activity (4-yr follow-up, 1,051 patients). Persson and associates³⁸ reported that Lp-PLA₂ activity, but not mass, was associated with increased risk of CV disease. Jenny and co-authors,³⁴ in their study, showed no difference in Lp-PLA₂ activity and mass with respect to risk prediction. O'Donoghue and the PROVE IT–TIMI 22 trial investigators³¹ showed a moderate correlation between Lp-PLA₂ activity and mass, both at baseline (r=0.35, P <0.001) and at 30-day follow-up (r=0.36, P <0.001). Recently, the United States Food and Drug Administration has approved the Immunoassay (PLAC test), which measures Lp-PLA₂ mass, for screening patients who are at high risk for CV disease. A large-scale, randomized, multicenter trial is needed for a head-to-head comparison, to determine which assay method is the better predictor of future events.

On the basis of our evaluation of the studies included in this review, we can state that Lp-PLA₂ fulfills most, but not all, of the criteria⁶⁹ set forth by the American Heart Association for the evaluation of a novel risk marker. None of these studies, however, has evaluated the accuracy of Lp-PLA₂, its ability to reclassify risk, the net effect that such reclassification would have, the fit of the model, or the cost-effectiveness. This needs to be done in future studies specifically designed to examine these issues.

The identification of Lp-PLA₂ as a risk marker for CVD is consistent with our current understanding of atherosclerosis as an inflammatory disease. The available body of evidence suggests that Lp-PLA₂ is a reliable marker of risk for CV events. A fundamental question remains: Is Lp-PLA₂ a simple marker or does it have a causal impact on atherosclerosis? The relationship between Lp-PLA₂ and CV events is consistent, strong, graded, and biologically plausible, thereby meeting a few (but not all) of several criteria proposed by Sir Bradford Hill to evaluate a cause–effect relationship.⁷⁰ Obviously, it will be crucial to show that inhibition of Lp-PLA₂ causes stabilization or regression of atherosclerosis and, preferably, that it prevents CV events. Recently, 2 randomized, placebo-controlled clinical trials studied the effects, in the vascular wall and systemic circulation, of the Lp-PLA₂-specific inhibitor darapladib. Serruys and coworkers,⁷¹ in their trial, showed a 59% reduction in Lp-PLA₂ levels in the darapladib-treated group (P <0.001 vs placebo) and a significant increase in necrotic core volume in the placebo group (P=0.009), when compared with the darapladib-treated group (P=0.71). Mohler and colleagues⁷² found a 66% decrease in Lp-PLA₂ levels in the darapladib-treated group (P <0.001 vs placebo). Statins and fibrates are also shown to decrease Lp-PLA₂ activity by 20% to 30% without an effect on de novo synthesis and secretion of Lp-PLA₂ by macrophages.⁷³ Whether inhibiting or decreasing its activity by the use of these drugs will lead to a decrease in CV events or death remains to be proved in clinical trials. Recently, a consensus panel has recommended decreasing LDL-c by 30 mg/dL in patients with high levels of Lp-PLA₂, regardless of baseline LDL-c.⁶⁸

If the reversibility criterion is met by future studies, a cause–effect relationship can be considered for Lp-PLA₂ and CV disease, which is a multicausal disease with a complex pathogenesis.⁷⁴