| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Special Report |
1 Department of Medicine, University of Washington, Northwest Lipid Research Laboratories, 2121 N. 35th St., Seattle, WA 98103.
2 Department of Biochemistry, Queen’s University, Room A208 Botterell Hall, Kingston, ON K7L 3N6, Canada.
3 Vascular Biology Program, NIH/NHLBI/DHVD, 6701 Rockledge Dr., Room 10198, Bethesda, MD 20892.
aAuthor for correspondence. Fax 206-685-3279; e-mail smm{at}u.washington.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionDiscussion HighlightsRecommendationsReferences |
|---|
37% of the US population judged to be at high risk for developing coronary artery disease (CAD), based on the National Cholesterol Education Program guidelines, have increased plasma lipoprotein(a) [Lp(a)], whereas Lp(a) is increased in only 14% of those judged to be at low risk. Therefore, the importance of establishing a better understanding of the relative contribution of Lp(a) to the risk burden for CAD and other forms of vascular disease, as well as the underlying mechanisms, is clearly evident. However, the structural complexity and size heterogeneity of Lp(a) have hindered the development of immunoassays to accurately measure Lp(a) concentrations in plasma. The large intermethod variation in Lp(a) values has made it difficult to compare data from different clinical studies and to achieve a uniform interpretation of clinical data. A workshop was recently convened by the National Heart, Lung, and Blood Institute (NHLBI) to evaluate our current understanding of Lp(a) as a risk factor for atherosclerotic disorders; to determine how future studies could be designed to more clearly define the extent to which, and mechanisms by which, Lp(a) participates in these processes; and to present the results of the NHLBI-supported program for the evaluation and standardization of Lp(a) immunoassays. This report includes the most recent data presented by the workshop participants and the resulting practical and research recommendations. |
Introduction |
|---|
|
TopAbstractIntroductionDiscussion HighlightsRecommendationsReferences |
|---|
Analysis of an apo(a) cDNA from a human liver library in 1987 revealed that apo(a) bears striking homology with the fibrinolytic proenzyme plasminogen and is a member of the plasminogen gene superfamily (7). apo(a) contains domains that correspond to single copies of plasminogen kringle 5 and the serine protease domain, as well as multiple copies of a sequence that closely resembles plasminogen kringle 4. On the basis of amino acid sequencing, apo(a) contains 10 types of plasminogen kringle 4-like motifs (designated K4 types 1–10). Each K4 motif is present in a single copy, except for apo(a) K4 type 2, which is present in a variable number of identically repeated copies (3 to >40) (8)(9). This forms the molecular basis of the Lp(a) isoform size heterogeneity that is a hallmark of this unique lipoprotein. It is noteworthy that despite a high degree of sequence similarity, the serine protease-like domain of apo(a) is catalytically inactive (10), which has led to speculation that Lp(a) can interfere with the physiologic functions of plasminogen in the fibrinolytic cascade. This could provide a direct link between the processes of atherosclerosis and thrombosis.
Sites for N-linked glycosylation are present within the core of each apo(a) K4 motif (7)(11), whereas a minimum of six O-linked glycosylation sites are present within the linker sequences that join individual kringles (12). The hydrophilic O-linked glycans are primarily (80%) composed of monosialylated core structures (13), which may have implications for Lp(a) catabolism. apo(a), and thus Lp(a), is heterogeneous in its glycosylation, although the significance of the glycosylation modification with respect to the in vivo properties of Lp(a) is still unclear.
|
Discussion Highlights |
|---|
|
TopAbstractIntroductionDiscussion HighlightsRecommendationsReferences |
|---|
There is evidence to suggest that the rate of formation of Lp(a), rather than its catabolism, is the primary determinant of plasma Lp(a) concentrations (19). This is supported by data demonstrating that the fractional catabolic rates of 125I-labeled Lp(a) of differing isoform size composition are not significantly different (20), despite the general inverse correlation that has been reported between apo(a) isoform size and plasma Lp(a) concentrations (2). This inverse correlation has been attributed at least in part to the increased endoplasmic reticulum retention time that has been reported for larger apo(a) species, which leads to a greater extent of intracellular degradation in the hepatocytes (21).
In Lp(a) particles, apo(a) is covalently linked to apo B-100 by a single disulfide bond (22)(23); the stoichiometry of apo(a) and apo B-100 in the Lp(a) particle is 1:1 (24). Considerable evidence suggests that Lp(a) assembly occurs as a two-step process in which initial noncovalent interactions between apo B and apo(a) precede disulfide bond formation; this initial interaction can be effectively inhibited by lysine and lysine analogs (25)(26)(27). Experiments using site-directed mutagenesis of recombinant apo(a) suggest that the weak lysine binding sites present within apo(a) K4 types 7 and 8 play a major role in noncovalent association with apo B-100 [M.L. Koschinsky and L. Becker, National Heart, Lung, and Blood Institute (NHLBI) conference presentation]; other work from this group has identified key lysine residues within the NH2-terminal 18% of apo B, which bind to the weak lysine binding sites in apo(a) (28). A peptide spanning amino acid residues 680–704 in the NH2 terminus of apo B can reduce Lp(a) formation in vitro (28). Additional sites of noncovalent interaction within the COOH terminus of apo B have also been identified (29); Sharp et al. (30) have recently shown that a peptide spanning amino acid residues 4372–4392 in apo B-100 can completely inhibit the apo(a)–apo B binding interaction. The use of peptide inhibitors may represent an interesting strategy for inhibiting Lp(a) assembly. The ability of the lysine analog tranexamic acid to reduce plasma Lp(a) concentrations in humans has also been reported (31), again underscoring the potential utility of inhibiting Lp(a) assembly to lower plasma Lp(a) concentrations. Most recently, Becker et al. (32) have reported that a conformational change in apo(a), which can be elicited in vitro by lysine analogs, is accompanied by a significant increase in the rate of the covalent step of Lp(a) assembly; this may provide the rationale for the development of a new generation of compounds aimed at lowering plasma Lp(a) concentrations by inhibiting the process of covalent Lp(a) assembly.
The pathway(s) of Lp(a) catabolism in vivo remain(s) unclear. A study using primed constant infusion of deuterated leucine in 23 individuals revealed that the residence time of apo(a) in plasma was 4.8 days, whereas the residence time of apo B-100 associated with Lp(a) particles was only 2.5 days (E. Schaefer, NHLBI conference presentation). These data were taken to suggest that apo(a) recirculates twice on an apo B-100 particle in plasma before catabolism. However, the mechanism that would allow for cellular recycling of apo(a) remains unclear at present. Additionally, this finding is not in agreement with other kinetic studies showing similar plasma residence times of apo(a) and apo B-100 in Lp(a) (18).
The major site of catabolism of Lp(a) particles in humans remains unidentified. Although fragments of apo(a) have been identified in urine, this pathway likely does not account for more than 1% of the fractional catabolic rate of Lp(a) (33). Several studies have identified receptors that can bind to and internalize Lp(a) [for a review, see Ref. (34)]. Kostner and colleagues have recently identified the mammalian hepatic asialoglycoprotein receptor, an endocytotic recycling receptor that mediates the binding and internalization of desialylated glycoproteins as a selective receptor and may be responsible for Lp(a) catabolism in vivo (G.M. Kostner, NHLBI conference presentation). Further studies will be required to assess the relative contributions of receptor- and non-receptor-mediated catabolism of Lp(a).
Studies by Gaubatz et al. (35) have demonstrated the existence of Lp(a) noncovalently associated with triglyceride-rich lipoproteins in patients with hyperlipidemia; they also reported that Lp(a) containing higher molecular weight apo(a) isoforms preferentially bound to the triglyceride-rich lipoprotein fraction. The catabolism of the Lp(a)–triglyceride-rich lipoprotein complex, which is more abundant in patients with hypertriglyceridemia, may be different from that of conventional Lp(a) particles and therefore warrants further study.
contribution of genetics to determination of plasma Lp(A) concentrations
Discovery of the genetic size polymorphism of apo(a) by Utermann et al. (2) has provided major insights into the genetic control of plasma Lp(a) concentrations. Results of early studies demonstrated that all apo(a) isoforms are codominantly expressed at a single locus by multiple autosomal alleles (differing in the number of sequences encoding K4 type 2 motifs) (8)(36). Additionally, it was demonstrated that the sizes of apo(a) isoforms are, in general, inversely correlated with plasma Lp(a) concentrations (2).
Examination of Lp(a) concentrations and apo(a) genotypes in 48 Caucasian families demonstrated that the apo(a) gene LPA accounts for >90% of the genetic variation in plasma Lp(a) concentrations and that the number of K4 repeats in LPA accounts for 70% of this variation (37). However, the extent to which variation in the size of LPA accounts for the variability in plasma Lp(a) concentrations is highly dependent on the ethnic composition of the study group. For example, although LPA itself is also the major determinant of plasma Lp(a) concentrations in the black American population, it has been reported to account for only one-half of the variance in the African population (38). In the latter population, other genetic factors (in addition to LPA) as well as nongenetic factors may contribute to the determination of Lp(a) concentrations.
Sequence variations in the coding and noncoding regions of LPA have been reported to contribute to variation in plasma Lp(a) concentrations. For example, a repeat polymorphism of the pentanucleotide sequence TTTTA present at -1371 upstream of the apo(a) translational start site in LPA may account for 10–14% of the interindividual variations in plasma Lp(a) concentrations in Caucasians (39). Although an inverse relationship between the number of pentanucleotide repeats and Lp(a) concentrations in Caucasians has been reported, no such correlation was observed in African individuals (39). Interestingly, a +93 C/T substitution in LPA has been reported to decrease translational efficiency in vitro and influences Lp(a) concentrations in African but not Caucasian populations (40).
In a recent study, the size of apo(a) alleles together with the pentanucleotide repeat polymorphism and single-nucleotide polymorphisms in the 5' and 3' regions of LPA were analyzed in >2000 unrelated individuals from 17 populations of African, Arctic, Asian, European, and Pacific origins (G. Utermann, NHLBI conference presentation). Core haplotype frequencies for the single-nucleotide polymorphisms were found to be different among populations of different geographic origin. Significant linkage disequilibria exist between the polymorphisms separated by the K4 type 2 repeats, suggesting that the size polymorphism in the gene encoding apo(a) has arisen from gene conversion rather than by recombination. Additionally, this study confirmed that the relationship between LPA variation and Lp(a) concentrations is heterogeneous across ethnic groups. Although these studies suggest a causal relationship between polymorphisms in the gene encoding apo(a) and variation in plasma concentrations of Lp(a), the exact identity of the causal polymorphisms and the mechanisms underlying their effects on Lp(a) concentrations remain to be pinpointed.
On the basis of the studies described above, the strong ethnic dependence of the relationship between apo(a) size and Lp(a) concentration needs to be taken into account in the design of prospective studies aimed at evaluating the contribution of Lp(a) to future cardiovascular events. This is because there are indications that small apo(a) isoforms, in addition to being associated with higher plasma Lp(a) concentrations, may in and of themselves be more atherogenic (41). As such, the relationship between apo(a) isoform size and risk for cardiovascular disease (CVD) needs to be examined in various ethnic groups.
use of genomics to study mechanisms of apo(A) gene regulation and evolution
The orthologous distribution of the apo(a) gene is limited to Old World monkeys and humans (42), which prevents the use of a standard comparative genomics approach to identify key regulatory sequences in this gene. Recently, the technique of phylogenetic shadowing has been applied by Boffelli et al. (43) to uncover primate-specific regulatory elements in the apo(a) gene. Sequence analysis of the apo(a) gene locus in humans, chimpanzees, and baboons revealed a region (1.6 kb) of extreme conservation adjacent to the transcription start site; this region was analyzed in 18 Old World monkeys and hominoids. Phylogenetic shadowing revealed that regions with the lowest variation corresponded to sequences containing the first exon, the TATA box, and a hepatocyte nuclear transcription factor-1
binding site; the last two have been functionally identified in a previous study (44). Eight additional regions (40–70 bp in length) showed a high degree of conservation, which suggests that they are also potentially important in regulation of the apo(a) gene. The 10 most conserved regions of sequence were found to interact with DNA binding proteins and had a larger functional impact on expression based on their ability to drive expression of a reporter gene in vitro, compared with nonconserved regions.
Extensive genomic sequence comparisons of species both with (e.g., hedgehog, baboon, and human) and without (e.g., mouse and lemur) the apo(a) gene indicate that twice during evolution, duplication of the plasminogen gene has occurred, giving rise to the apo(a) gene. In both cases, this has led to the generation of a molecule that can bind to both LDL and fibrin, thereby providing a compelling example of convergent evolution. Taken together, the studies presented by Boffelli et al. (43) underscore the usefulness of genome analysis in understanding the regulation of the apo(a) gene, as well as its evolutionary history.
mechanism of Lp(A) action: insights from animal models
Studies of Lp(a) biology using animal models have been complicated by the unusual species distribution of this lipoprotein. Lp(a) is present only in humans, Old World monkeys, and the hedgehog (45). The principal animal models for Lp(a) have been transgenic mice and rabbits. Both species have their advantages and disadvantages, and both models have provided fundamental insights into biochemical and pathophysiologic aspects of Lp(a). The most powerful application for transgenic apo(a) animals is assessment of the role of apo(a) or Lp(a) in the development of atherosclerosis and analysis of the mechanism(s) by which this may occur.
The earliest data suggesting that apo(a) is atherogenic came from transgenic mouse experiments. It was reported that apo(a) transgenic mice fed an atherogenic diet had significantly larger aortic lesion areas than their nontransgenic littermates fed the same diet, and apo(a) was found colocalized with (mouse) apo B in the lesions (46). Subsequent studies in mice transgenic for both human apo(a) and human apo B-100 confirmed that animals with the apo(a) transgene had increased aortic fatty streaks, whereas animals with both transgenes had lesions 2.5 times larger than those expressing the apo(a) transgene alone (47). In contrast, Mancini et al. (48) found that expression of apo(a) alone did not promote atherosclerosis in the proximal aorta of fat-fed mice. A subsequent study comparing human apo B-100 with apo(a)/human apo B-100 transgenic mice in LDL-receptor knockout mice also found that expression of apo(a) did not potentiate aortic lesion development (49). However, Berg et al. (50), studying relatively aged apo(a) transgenic mice that possessed the LDL receptor and had been fed normal chow, found that these mice had more extensive atherosclerotic lesions than nontransgenic control mice.
In sharp contrast to the inconsistent atherogenic effects of apo(a) in transgenic mouse models is the striking atherogenicity of apo(a) when expressed from a transgene in rabbits. Studies by Fan et al. (51) evaluating diet-induced atherosclerosis in transgenic rabbits revealed that expression of apo(a) led to an approximately two- to fivefold increase in lesion area. Unlike transgenic apo(a) mice, in which atherosclerosis was limited to the proximal aorta, in the transgenic rabbits lesions were observed throughout the aorta and in the carotid, iliac, and coronary arteries. As reported recently, considerably more dramatic results were seen when apo(a) was expressed from a transgene in Watanabe Heritable Hyperlipidemia (WHHL) rabbits (52). The WHHL rabbit is an excellent model for atherosclerosis because the animals spontaneously develop severe atherosclerosis that is characterized by complex lesions closely resembling those found in advanced human disease (53). The morphologic features of the lesions differed markedly between nontransgenic and transgenic animals. The lesions in the nontransgenic WHHL rabbits resembled fatty streaks, whereas the apo(a)-expressing animals had much more advanced lesions that contained atheroma, fibroatheroma, and calcification. These lesions were covered by a fibrous cap containing smooth muscle cells and had a necrotic core containing macrophage-derived foam cells (52). These findings are particularly noteworthy because it is this type of lesion morphology that has been implicated in the "vulnerable plaque" in human disease that is prone to rupture, from which the catastrophic thrombotic sequelae of atherosclerosis ensue.
The animal model studies conducted to date provide prima facia evidence that apo(a) is an atherogenic molecule. It is clear that both the mouse and rabbit models have their advantages and disadvantages. The models should be viewed as complementary and should, coupled with continued in vitro studies of the biochemical properties of apo(a) and Lp(a), provide critical insights into the pathogenic mechanisms of Lp(a). In addition, the animal models can be used to investigate whether smaller apo(a) isoforms are more atherogenic and whether there are gender differences in risk attributable to Lp(a), as well as to discover and apply therapeutic strategies aimed at lowering plasma Lp(a) concentrations.
measurement of Lp(A) concentrations in plasma
The high degree of size heterogeneity of apo(a), derived from the variable number of K4 type 2 repeats, provides a variable number of identical epitopes to interact with the antibodies used in immunoassays to measure the concentration of Lp(a) in plasma. Considering that the main prerequisite for the development of an accurate immunoassay is that the antibodies possess the same immunoreactivity per particle for the assay calibrator and for the samples being analyzed, it is evident that the size variation of apo(a) affects the development of immunoassays on two fronts: (a) the choice of apo(a) size in the calibrator is arbitrary, and independent of the choice, the calibrator would not be representative of the size of apo(a) in most of the samples; and (b) the immunoreactivity of the antibodies directed to the identically repeated epitopes in K4 type 2 will vary depending on the size of apo(a). As a consequence, the immunoassays will tend to underestimate Lp(a) in samples with apo(a) sizes smaller than the apo(a) size of the assay calibrator and, conversely, overestimate Lp(a) in samples with larger apo(a) isoforms.
To document the impact of apo(a) size on the accuracy of Lp(a) measurements, Marcovina et al. (54) generated and characterized a variety of monoclonal antibodies (MAbs) directed to different apo(a) epitopes. A MAb (a5) directed to an epitope present in K4 type 2, whose immunoreactivity per particle would therefore vary depending on apo(a) size, and another MAb (a40) specific for a unique epitope located in K4 type 9 were used to develop two enzyme immunoassay (ELISA) methods to measure Lp(a) in a large number of samples. Both assays were calibrated with the same serum containing an apo(a) with 21 K4 motifs. The Lp(a) protein value was assigned to the assay calibrator by use of a primary preparation, and the values were expressed in nmol/L to reflect the number of Lp(a) particles in plasma. As expected, practically identical values were obtained by the two ELISA methods in the samples with the same apo(a) size as in the assay calibrator, whereas the Lp(a) values measured by MAb a5 were under- or overestimated in samples with apo(a) containing <21 or >21 K4 repeats, respectively, compared with the values obtained by MAb a40. This MAb a40-based ELISA has been carefully optimized, extensively validated in a large number of individuals (14), and used as a reference method in the NHLBI-sponsored contract for the standardization of Lp(a) measurement.
The availability of a primary reference material is the first main step for the accuracy of immunoassays. As part of the NHLBI contract, Celina Edelstein, in collaboration with Dr. Angelo Scanu, developed a suitable method to prevent the marked physical, chemical, and immunologic changes that Lp(a) undergoes when isolated from human plasma (55). When isolated Lp(a) was lyophilized in the presence of suitable cryopreservatives and then reconstituted, it was found to be indistinguishable from the starting material in terms of composition and physical, chemical, and biological properties. Moreover, the reconstituted products exhibited unchanged immunochemical properties and appeared to have all of the requisites to be used as a primary reference material (C. Edelstein, NHLBI conference presentation).
The second step in the standardization process is the selection of a secondary reference material to be used to achieve comparability of values among different immunoassays. On the basis of extensive evaluations performed by the IFCC Working Group on Lp(a), a lyophilized serum pool preparation was selected as secondary reference material for Lp(a). As a collaboration between the recipients of the NHLBI contract and the IFCC Working Group on Lp(a), the MAb a40-based ELISA calibrated with a purified Lp(a) primary preparation was used to assign a target value to the IFCC secondary reference material. A target value of 107 nmol/L of Lp(a) protein was assigned to the IFCC preparation based on 144 replicate analyses. To evaluate the performance of commercially available methods, the IFCC reference preparation was then used to transfer an accuracy-based value to the various immunoassay calibrators. After uniformity of calibration was demonstrated in the 22 evaluated systems, Lp(a) was measured in 30 fresh-frozen samples covering a wide range of Lp(a) concentrations and apo(a) sizes with values assigned by the reference method. The CV among the methods on the IFCC preparation was 2.8%, clearly indicating the lack of matrix effect of this material. However, the CV among laboratories on the 30 samples ranged from 6% to 31%. Additionally, significant apo(a) size-dependent biases were observed among the values obtained by the majority of the evaluated systems. Only a turbidimetric assay was found to be insensitive to apo(a) isoform size. With this method, there was a nearly perfect correlation and a negligible bias between the obtained and the expected values, indicating the suitability of the IFCC reference material (56).
Despite the use of a common reference preparation, no harmonization in Lp(a) values among the different methods was achieved. This indicates that the impact of apo(a) isoform size on Lp(a) concentrations varies among the different methods as a function of the apo(a) size of the assay calibrators. In addition, other factors, such as differences in antibody properties; assay precision and robustness; and sensitivity of the assays to sample handling, storage conditions, and length of storage, play a role in and contribute to the lack of comparability of data obtained by different methods. All of these factors need to be taken into consideration in the standardization and validation of Lp(a) immunoassays. From these observations, it is obvious that no reference material, either primary or secondary, would be able to eliminate the biases in Lp(a) values obtained by methods that are affected by apo(a) size heterogeneity and/or are not properly optimized. However, the IFCC preparation has been shown to have excellent stability and commutability properties, and it has been proposed to the WHO to be accepted as the WHO-IFCC Reference Material for Lp(a) (G.M. Kostner, NHLBI conference presentation). Its availability could play an important role by providing an accuracy-based calibration of validated assays.
To reduce the apo(a) size-dependent differences in results obtained with the immunoassays, a different approach to assay calibration was investigated. Instead of using serial dilutions of a single calibrator, five fresh-frozen samples with different apo(a) sizes from small to large and suitable Lp(a) concentrations were used to calibrate a turbidimetric assay affected by apo(a) size variation. This assay was selected because it had good robustness and precision characteristics, and no significant differences were observed in Lp(a) values between fresh samples and samples stored at -70 °C over a period of 15 months. Analyses were performed in parallel using the original assay calibrator and the five-sample calibrator on a large number of samples. Consistent apo(a) size-dependent biases were observed with the original assay calibrator. In contrast, comparability between the observed values and the values obtained by the reference method was good when the five independent samples were used to calibrate the assay (J.J. Albers, NHLBI conference presentation). Although this approach appears promising for reducing assay inaccuracy, it may not be equally effective in all of the methods or in all of the samples. Therefore, its potential use for assay standardization will require further evaluation using multiple methods and a large number of samples with a good representation of single and double apo(a) isoforms.
impact of Lp(A) method inaccuracy on the interpretation of Lp(A) values
In the majority of clinical studies, Lp(a) concentrations have been determined by methods affected by apo(a) size heterogeneity. Therefore, for the conclusions of these studies to be valid, we must rely on the assumption that the distribution of apo(a) isoforms is similar between cases and controls, thus minimizing the potential that method-dependent over- or underestimation of Lp(a) values may contribute to the observed difference or lack thereof between cases and controls. Despite the importance of this topic, no studies have been performed to evaluate the impact of method inaccuracy on the interpretation of clinical data. To evaluate to what extent method inaccuracy affects the correct stratification of individuals for coronary artery disease risk, based on their Lp(a) values, Lp(a) concentrations were determined by the reference method in 2940 samples from the participants in the Framingham Study collected during the fifth cycle and apo(a) isoforms were determined as described previously (57). During the same cycle, Lp(a) values were also determined in other laboratories for 2556 of the samples by a turbidimetric method and for 2662 of the samples by a commercially available ELISA. A Lp(a) value of 75 nmol/L, which approximates the 75th percentile of the Framingham population as determined by the reference method, was arbitrarily selected as the clinical decision point.
On the basis of the Lp(a) values obtained by the turbidimetric assay, 136 individuals were misclassified as being at increased risk (false positive) and 23 were misclassified as being not at risk (false negative). On the basis of the Lp(a) values obtained by the commercial ELISA, 329 individuals were misclassified as being at increased risk (false positive) and 25 were false negative. The vast majority of the misclassifications observed with the turbidimetric method were explained by the over- or underestimation of Lp(a) values based on the apo(a) size in the samples. The larger number of false positives compared with the false negatives obtained by this assay is explained by the small apo(a) size in the assay calibrator and by the high frequency of samples in the general population with apo(a) sizes larger than that in the calibrator. This observation is consistent with previously reported data (56). However, the large number of false-positive values generated by the ELISA method was not explained by the apo(a) sizes in these samples. In fact, a high degree of variability in Lp(a) values was observed with this ELISA method. In addition, the values obtained from frozen samples were generally higher than those from fresh samples, but the magnitude of the increase was sample dependent. These findings clearly indicate that assay standardization can be achieved only if each assay is properly optimized in addition to being evaluated for its sensitivity to apo(a) size polymorphism (S.M. Marcovina, NHLBI conference presentation).
To evaluate method differences in the interpretation of clinical data, a study was performed to directly compare the ability of the ELISA reference method (54) and a commercially available latex-based nephelometric method to predict future angina pectoris in men participating in the Physicians’ Health Study. apo(a) isoform size was determined and plasma Lp(a) concentration was simultaneously measured by the ELISA reference method and in a different laboratory by the nephelometric method. Analyses were performed in samples from 195 study participants who subsequently developed angina and from paired controls, matched for age and smoking, who remained free of reported vascular disease. The median baseline Lp(a) value in cases, as determined by ELISA, was 35% higher (P = 0.02) than that in controls. Additionally, Lp(a) was associated with increased relative risk for angina, and this association was strengthened after controlling for lipid risk factors. However, the median Lp(a) value determined by the commercial method was not statistically different between cases and controls, and the association with angina was also not significant. On the other hand, very high Lp(a) values obtained by the nephelometric assay (>95th percentile of controls) significantly predicted angina, although the relative risks were not as strong than those with the reference method.
The results of these two studies clearly indicate the importance of using suitable and standardized methods for risk assessment and for the interpretation of clinical outcomes (S.M. Marcovina, NHLBI conference presentation).
Lp(A) as an emerging risk factor for vascular disease: clinical perspectives
On the basis of the Adult Treatment Panel (ATP) III guidelines, Lp(a) is currently classified as an "emerging" lipid risk factor for CVD (58). For increased Lp(a) to be considered a "major" risk factor (i.e., in the same category as cigarette smoking and increased LDL-cholesterol) it has to meet the following criteria: (a) robust predictive power for CVD; (b) high prevalence of Lp(a) concentrations above an arbitrary risk threshold in the general population; (c) ready availability of clinical samples and a widely available, standardized, inexpensive means to measure Lp(a); and (d) evidence for a benefit of lowering Lp(a) concentrations (S. Grundy, NHLBI conference presentation).
For these reasons, it is essential to define stringent criteria for assay standardization. It is evident that a robust predictive power of Lp(a) can be fully explored only if measurements are made with methods that have been evaluated for their suitability to produce accurate Lp(a) values. On the basis of his long-standing experience with the Framingham Study, Dr. Kannel (NHLBI conference presentation) provided strong rationales for the use of highly standardized laboratory procedures for the measurement of lipids and lipoproteins. Standardized methods are essential for (a) promulgating guideline recommendations for establishing and implementing goals for therapy; (b) defining with confidence the existence, strength, and nature of the relationship of lipid and lipoproteins to disease; (c) estimating the prevalence and determinants of the abnormality within and among population subgroups; (d) comparing trends by age, sex, and populations; and (e) combining results from different studies to define critical values in relation to the disease outcome. It is clear that the above-mentioned goals need to be reached for defining the role of Lp(a) in clinical practice and that these can be accomplished only with the use of accuracy-based standardized methods.
Metaanalysis of 27 prospective studies provided information on 5436 coronary heart disease (CHD) cases observed during a mean follow-up time of 10 years (59). Despite differences in methods to measure Lp(a), this analysis indicates that individuals in the general population with Lp(a) concentrations in the top one-third of baseline measurements are at 70% increased risk of CHD compared with individuals in the bottom one-third.
The predictive strength of Lp(a) for CHD was evaluated in the Atherosclerosis Risk in Communities (ARIC) Study, with 10-year follow-up and 725 CHD events (60). Lp(a) was found associated with modest risk ratios. In terms of population differences, the ARIC results suggest that Lp(a) confers less risk in blacks than in whites. A recent prospective study (61) examined a cohort of 1216 patients with a mean follow-up time of 6.7 years with total mortality and mortality attributable to CVD used as outcome variables. In this study, Lp(a) concentrations in excess of 300 mg/L were present in 30% of the study population and were found to be an independent predictor of death. On this basis, the authors suggested that Lp(a) values in excess of 300 mg/L are suggestive of a poor prognosis and may serve to identify patients who would benefit from aggressive secondary prevention programs.
Although there is little or no correlation between plasma Lp(a) concentrations and other vascular risk factors, evidence has been provided from several studies to suggest that the risk attributable to Lp(a) is dependent on the concomitant presence of other such risk factors. In the Familial Atherosclerosis Treatment Study (FATS), Lp(a) was a strong predictor of events at baseline, but lost its predictive value when LDL-cholesterol was reduced to <1000 mg/L in the treatment group (62). More recently, in the Prospective Epidemiological Study of Myocardial Infarction (PRIME) (63), Lp(a) was investigated as a CHD risk factor in a prospective cohort of 9133 French and Northern Irish men, 50–59 years of age, without a history of CHD. Increased Lp(a) increased the risk for myocardial infarction and angina pectoris, and the effect was most pronounced in men with high LDL-cholesterol. The results of the Quebec Cardiovascular Study also suggest that Lp(a) is not an independent risk factor for ischemic heart disease in men, but increases the risk associated with increased apo B and total cholesterol and appears to attenuate the beneficial effects of increased HDL (64). The same interactions of increased Lp(a) with other risk factors were found in the PROCAM study. As a consequence, a high Lp(a) concentration further increased the risk of myocardial infarction in men with high or moderately increased estimated global risk (i.e., risk of a coronary event >10% in 10 years) but not in men with a low estimated global risk (65).
In addition to the potential synergy of Lp(a) risk with other markers of dyslipidemia, the interaction of Lp(a) with other thrombotic risk factors for stroke in children has been reported by Nowak-Gottl et al. (66). Increased plasma Lp(a) concentrations in combination with resistance to activated protein C, attributable to the presence of the factor V Leiden mutation, was associated with an odds ratio of 30 for stroke in children. In other studies, this group has also reported that increased Lp(a) concentration, familial protein C deficiency, and ischemic stroke of vascular origin had significant and independent associations with recurrent ischemic stroke in children (67). Clinical evidence suggests that Lp(a) may also be an independent risk factor for venous thromboembolism (68)(69). Recently, the role of Lp(a) as a risk factor for secondary thromboembolic events was investigated prospectively in a cohort of 301 children. The results suggest that the combination of increased Lp(a) with any thrombophilic risk factor increased the risk of a thromboembolic event by a factor of 2.6; this risk ratio increased to 6.2 when increased Lp(a) occurred together with the factor V Leiden defect (70). Taken together, these studies clearly indicate that the risk contributed by Lp(a) to venous thrombosis is exacerbated by the presence of other vascular risk factors.
An area of clinical investigation that is receiving increased attention is the predictive value of apo(a) isoform size. The results of the Bruneck study demonstrated that small (<22 K4 repeats) apo(a) isoforms predicted risk for advanced atherosclerosis independently of Lp(a) concentration (41). Prospective data from the Physicians’ Health Study examining apo(a) isoform size and Lp(a) concentrations [measured using an apo(a) size-independent method] and the risk of angina pectoris in men demonstrated that both Lp(a) concentrations and apo(a) isoform size were predictive for risk of future angina (F.M. Sacks, NHLBI conference presentation). On the basis of the independent contributions of apo(a) size and Lp(a) concentrations in a multivariate model, the authors of this study found that in men with LDL concentrations >1600 mg/L, smaller apo(a) isoform size was associated with a greater magnitude of risk and remained an independent risk factor after adjustment for the contribution of other variables.
modulation of plasma Lp(A) concentrations
Compared with plasma LDL, Lp(a) concentrations are relatively resistant to alteration by traditional pharmacologic and nonpharmacologic approaches. However, there are reports of several interventions as well as disease conditions, such as renal disease and poorly controlled diabetes mellitus, that may significantly increase plasma Lp(a) concentrations [for a review, see Ref. (71)]. The most effective method to decrease Lp(a) concentrations by 50% or more is by LDL or Lp(a) apheresis procedures (72)(73). However, these procedures are costly and generally reserved for patients with extreme forms of familial hypercholesterolemia. Several agents have been reported to reduce Lp(a) concentrations, including niacin in high doses (74), L-carnitine at a dose of 2 g/day (75), and ascorbic acid (3 g/day) together with L-lysine monohydrochloride (3 g/day) (76). Various hormones have also been shown to modulate plasma Lp(a) concentrations, and both tamoxifen and estrogen have been shown to lower Lp(a) significantly in postmenopausal women (77).
Contradictory findings have been reported concerning the effect of statins on Lp(a) concentrations. A modest increase in Lp(a) concentrations was found in patients with increased cholesterol receiving simvastatin (78), and a 36% increase in plasma Lp(a) concentrations was reported in patients receiving atorvastatin (79). Use of pravastatin did not influence Lp(a) values (80), whereas fluvastatin was found to significantly reduce Lp(a) concentrations (81). In a recent study of a large patient cohort, both atorvastatin and simvastatin therapy for 6 weeks produced a modest but significant reduction in Lp(a) values (82). Clearly the effect of statins on Lp(a) concentrations requires further analysis in large cohorts to define the roles of baseline Lp(a) concentration as well as apo(a) isoform size on the magnitude of the statin effect. Additionally, mechanistic studies need to be performed to identify the molecular basis for the potential modulation of Lp(a) by statin therapy.
Aspirin therapy (81 mg/day) was reported to lower serum Lp(a) concentrations in a recent study of 70 patients with atherosclerotic disease (83). Interestingly, the magnitude of the decrease in Lp(a) was larger in patients with high concentrations, irrespective of apo(a) isoform size. This has been speculated to result from a greater reduction by aspirin of apo(a) gene transcription in patients with high baseline transcriptional activity of the gene. Clearly these intriguing results need to be confirmed with a larger patient population, and the mechanism underlying the effect of aspirin on apo(a) gene transcription requires further analysis.
In contrast to plasma LDL concentrations, Lp(a) concentrations are thought to be relatively resistant to diet and exercise. However, in a study of obese women with high baseline Lp(a), a low-calorie diet with concomitant weight loss produced a significant reduction in Lp(a) values (84). In terms of specific dietary effects, the authors of a recent randomized cross-over study reported that almonds, rich in monounsaturated fat, significantly reduced Lp(a) concentrations (85). Ginsberg et al. (86) reported a significant increase in Lp(a) values in individuals who reduced their intake in saturated fat. Therefore, the results of these two studies support the notion that fat intake may lower Lp(a).
Systematic lowering of plasma Lp(a) concentrations can provide us with a study design in which the consequences of Lp(a) reduction can be evaluated prospectively. This could allow a direct assessment of the contribution of Lp(a) concentration to atherosclerotic risk.
|
Recommendations |
|---|
|
TopAbstractIntroductionDiscussion HighlightsRecommendationsReferences |
|---|
75 nmol/L. More studies performed with standardized methods are required to confirm the clinical validity of this cut-point value and to establish cut-points for other populations, particularly those of African origin. To gain a full understanding of the mechanisms by which Lp(a) may contribute to the risk for CVD, which may provide the rationale for the development of therapeutic strategies aimed at inhibiting the harmful effects of this lipoprotein, we make the following recommendations:
We thank the participants in the workshop, who all contributed to the concepts explored in this report. The following is the roster of participants.
Santica Marcovina, PhD, ScD, University of Washington (Seattle, WA) and Sonia Skarlatos, PhD, Director, Vascular Biology Research Program, NIH/NHLBI/DHVD (Bethesda, MD; Co-Chairs); John Albers, PhD, University of Washington (Seattle, WA); Kåre Berg, Ullevaal University Hospital (Oslo, Norway); James I. Cleeman, MD, NIH/NHLBI (Bethesda MD); Arnold von Eckardstein, MD, University and University Hospital Zuerich (Zurich, Switzerland); Celina Edelstein, BA, The University of Chicago (Chicago, IL); Jianglin Fan, MD, PhD, University of Tsukuba (Tsukuba, Japan); Scott M. Grundy, MD, PhD, University of Texas Southwestern Medical Center (Dallas, TX); Helen Hobbs, MD, University of Texas Southwestern Medical Center (Dallas, TX); William B. Kannel, MD, MPH, Boston University School of Medicine (Boston, MA); Marlys Koschinsky, PhD, Queen’s University (Kingston, Ontario, Canada); Gerhard Kostner, MD, University of Graz (Graz, Austria); Peter Kwiterovich, MD, The Johns Hopkins Medical Institutions (Baltimore, MD); Sally McCormick, PhD, University of Otago (Dunedin, New Zealand); Joel D. Morrisett, PhD, Baylor College of Medicine (Houston, TX); Daniel J. Rader, MD, University of Pennsylvania School of Medicine (Philadelphia, PA); Edward Rubin, MD, Lawrence Berkeley National Laboratory (Berkeley, CA); Frank M. Sacks, MD, Harvard School of Public Health (Boston, MA); Ernst Schaefer, MD, Tufts University School of Medicine (Boston, MA); Angelo M. Scanu, MD, The University of Chicago (Chicago, IL); Gerd Utermann, MD, University of Innsbruck (Innsbruck, Austria).
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: Lp(a), lipoprotein(a); apo(a), apolipoprotein(a); apo B, apolipoprotein B; NHLBI, National Heart, Lung, and Blood Institute; CVD, cardiovascular disease; WHHL, Watanabe Heritable Hyperlipidemia; MAb, monoclonal antibody; and CHD, coronary heart disease. 
|
References |
|---|
|
TopAbstractIntroductionDiscussion HighlightsRecommendationsReferences |
|---|
. J Biol Chem 1994;269:19757-19765.
