|Home | About | Journals | Submit | Contact Us | Français|
Elevated levels of oxidized phospholipids (OxPL) on apolipoprotein B-100 particles (OxPL/apoB) are associated with cardiovascular disease and predict new cardiovascular events. Elevated lipoprotein (a) [Lp(a)] levels are a risk factor for cardiovascular disease in Whites, and in Blacks if they also carry small apolipoprotein (a) [apo(a)] isoforms. The relationship of OxPL/apoB levels to race, cardiovascular risk factors and apolipoprotein(a) isoforms is not established.
OxPL/apoB levels were measured in 3481 subjects (1831 Black, 1047 White and 603 Hispanic) in the Dallas Heart Study and correlated with age, gender, cardiovascular risk factors, lipoprotein (a) [Lp(a)] and apolipoprotein(a) isoforms. Significant differences in OxPL/apoB levels were noted among racial subgroups, with Blacks having the highest levels, compared to Whites and Hispanics (p<0.001 for each comparison). OxPL/apoB levels generally did not correlate with age, gender or risk factors. In the overall cohort, OxPL/apoB levels strongly correlated with lipoprotein (a) [Lp(a)] (r=0.85, p<0.001), with the shape of the relationship demonstrating a “reverse L” shape for log-transformed values. The highest correlation was present in Blacks followed by Whites and Hispanics, was dependent on apo(a) isoform size, and became progressively weaker with larger isoforms. The size of the major apolipoprotein(a) isoform (number of kringle type-IV repeats) was negatively associated with OxPL/apoB (r=-0.49, p<0.001) and Lp(a) (r=-0.61, p<0.001), irrespective of racial group. After adjusting for apolipoprotein(a) isoform size, the relationship between OxPL/apoB and Lp(a) remained significant (r=0.67, p<0.001).
OxPL/apoB levels vary according to race, are largely independent of cardiovascular risk factors, and are inversely associated with apolipoprotein(a) isoform size. The association of OxPL with small apolipoprotein(a) isoforms, where a similar relationship is present among all racial subgroups despite differences in Lp(a) levels, may be a key determinant of cardiovascular risk.
Oxidized lipids play a central role in mediating a variety of immune, pro-inflammatory and plaque destabilizing processes that further amplify the inflammatory response.1 Plasma levels of specific OxPL on apolipoprotein B-100 (apoB) particles (OxPL/apoB) can be measured with the murine monoclonal antibody E06. OxPL/apoB levels are elevated in patients with coronary, carotid and femoral artery disease,2,3 acute coronary syndromes,4 and following percutaneous coronary intervention.5 Interestingly, in human plasma, E06-detectable OxPL are preferentially carried by Lp(a) lipoprotein (a) [Lp(a)], compared to other apoB-100 particles.2-8
We recently showed in the Bruneck population, which is entirely White, that OxPL/apoB and Lp(a) levels were strongly and significantly associated with the presence, extent and interim development of carotid and femoral atherosclerosis from 1995-2000.3 OxPL/apoB and Lp(a) also predicted new cardiovascular events over a 10 year period, independent of other risk factors, and provided additional prognostic information within each Framingham Risk Score tertile.9
The Dallas Heart Study is a unique epidemiological survey of middle-aged, asymptomatic subjects of different racial groups (approximately 50% Blacks, 30% Whites and 20% Hispanic with 56% female and 44% males) with the purpose of evaluating traditional cardiovascular risk factors, biomarkers and non-invasive measures of atherosclerosis.10 The relationship of Lp(a) and apolipoprotein isoform sizes to coronary calcium in this cohort were previously described.11 The purpose of the current study was to determine if racial differences exist in OxPL/apoB levels and evaluate their relationship to cardiovascular risk factors.
A description of the Dallas Heart Study (DHS) subjects was previously described in detail.10,11 The DHS is a multiethnic, probability-based sample of the Dallas county population in which Blacks were systematically over-sampled so the final sample was 50% black. In this study, 3,481 blood samples were available from 3 ethnic groups at the baseline timepoint.
The content of OxPL per apoB-100 particle (OxPL/apoB) was measured as previously described in detail by chemiluminescent ELISA using the murine monoclonal antibody E06, which binds to the phosphocholine (PC) headgroup of oxidized but not native phospholipids.2,3,6,12 As described, equal numbers of apoB-100 particles are captured from each plasma sample and thus the content of OxPL is normalized for apoB-100 in each subject. Thus, by design, the OxPL/apoB measurement is independent of apoB-100 (and LDL-cholesterol) levels. The OxPL/apoB values are expressed as relative lights units reflecting the amount of E06 bound to OxPL on apoB particles captured on microtiter well plates with antibody MB47. It is to be emphasized that the OxPL/apoB measure only represents those OxPL recognized by E06 (i.e. E06 immunoreactivity) and does not represent all OxPL present on apoB particles. In particular, E06 does not recognize lysoPC.12
Measurement of Lp(a) levels as nmol/L was performed with a well validated assay which is independent of apolipoprotein(a) isoform size.11 If expressed as mg/dl, the values would be ~2.5 fold lower than as in nmol/L, although at extremes of apo(a) size, larger differences would be present at either extreme. Apolipoprotein(a) isoforms were measured as previously described.11 The analyses in this study were based on size of the major apolipoprotein(a) isoform visualized on agarose gel electrophoresis which is directly proportional to the number of Kringle IV repeats.13 In this study, the major apolipoprotein(a) isoform was associated with the smaller of the 2 alleles in 87% of subjects.
Total cholesterol, LDL-cholesterol (LDL-C), HDL cholesterol (HDL-C), triglycerides, high sensitivity C-reactive protein (hsCRP), lipoprotein-associated phospholipase A2 (Lp-PLA2) mass and activity were measured as previously described.14,15 Campesterol, lathosterol, sitosterol, homocysteine, monocyte chemoattractant protein-1 (MCP-1), interleukin-18 (IL-18), peptidoglycan recognition protein-1 (PGLYRP-1), brain natriuretic peptide (BNP), troponin and liver fat by magnetic resonance imaging were measured as previously described.15-21
Six groups were formed from 3 ethnic groups and 2 genders. Omnibus tests were performed among these groups on baseline demographic and laboratory variables using chi-square tests for nominal variables. For continuous variables 1-way ANOVA or Kruskal-Wallis tests were used depending on whether or not the variable was normally distributed.
Significance tests for OxPL/apoB and Lp(a) values were computed on log-transformed variables using 3 (racial group) by 2 (gender) ANOVA models. Post-hoc Tukey tests among racial groups and between genders within racial groups were utilized for comparisons. Other tests of OxPL/apoB utilized nonparametric Kruskal-Wallis or Mann-Whitney U tests. Correlations were computed using Spearman's rank-order method to avoid distributional assumptions. Partial correlation was conducted using log-transformed values of OxPL/apoB and Lp(a). Some analyses were stratified based on kringle IV repeats grouped in tertiles.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Table 1 shows the baseline demographic and laboratory variables in male and female Black, White, and Hispanic subjects. Significant baseline differences among groups were noted in all variables.
The data for these analyses were evaluated according to sex and race in a 2X3 ANOVA design. There were significant differences in log-transformed OxPL/apoB levels among the racial subgroups, with Blacks having the highest levels, followed by Whites and then by Hispanics (p<0.001 for each comparison, Figure 1A). Differences between males and females in OxPL/apoB within each subgroup were not significant except that Black females had higher OxPL/apoB and Lp(a) values than Black males (p=0.002). Lp(a) levels were higher in Black (p=0.005) and Hispanic (p=0.015) females compared to their male counterparts, but levels were similar among White females and males (p=0.42, Figure 1B). The racial differences remained significant when these groups were stratified by decade age groups (data not shown).
Frequency distribution graphs of log-transformed OxPL/apoB (Figure 2A) and Lp(a) (Figure 2B) levels among the racial groups showed a positive skewness for Hispanics and Whites, whereas less positive skewness was noted in Blacks.
Figure 3A displays the distribution of apo(a) isoforms in the 3 racial groups. In Blacks, a Gaussian distribution is present, whereas a bimodal distribution is more evident in Whites and Hispanics. In the entire cohort with all the 3 racial groups combined, when plotted against the number of kringle IV-repeats, OxPL/apoB (Figure 3B) and Lp(a) (Figures 3C) levels were highest in the subjects with small apo(a) isoforms and lowest in the subjects with large apo(a) isoforms. There was a larger variability in OxPL/apoB values in small apo(a) isoforms, whereas a larger variability in Lp(a) values was present in large apo(a) isoforms.
In the overall cohort, OxPL/apoB and Lp(a) levels were highly correlated (Spearman's r=0.85, p<0.001). When plotted on a linear scale, the positive relationship of OxPL/apoB to Lp(a) is noted in the entire cohort (Figure 4A). When plotted on a logarithmic scale (Figure 4B), there was evidence of a “reverse L” shape, with a weaker correlation between OxPL/apoB and Lp(a) (r=0.14, p<0.001) at Lp(a) levels 0<30 nmol/L, and then a stronger correlation (r=0.83, p<0.001) for Lp(a) levels ≥30nmol/L. Furthermore, when evaluating OxPL/apoB versus Lp(a) according to tertiles of apolipoprotein(a) isoform size (number of kringle IV repeats of the major apo(a) isoform: 12-20 repeats, 21-26 repeats and 27-41 repeats), a reverse L relationship was noted in apo(a) isoforms with 12-20 and 21-26 kringle-IV repeats, but less evident in those with 27-41 repeats (Figure 4C).
Figure 5 displays the relationship between OxPL/apoB and Lp(a) with apolipoprotein(a) isoform number in different racial groups. It can be appreciated that an inverse relationship is present between OxPL/apoB (Figure 5A) and Lp(a) (Figure 5B) and the number of Kringle IV repeats in Blacks (r= -0.614, p<0.001), Whites (r= -0.567, p<0.001), and Hispanics (r= -0.570, p<0.001). However, within each isoform class, Blacks have higher OxPL/apoB and Lp(a) levels than Whites and Hispanics.
Table 2 displays the correlations between OxPL/apoB, Lp(a) and apo(a) isoforms in the overall cohort and in the 3 racial groups. The correlation between OxPL/apoB and Lp(a) was strongest in Blacks, followed by Whites and then Hispanics. When the data were analyzed in groups based on the size of the major apolipoprotein(a) isoform, the racial differences were less marked, with a negative correlation between the major apolipoprotein(a) isoform and OxPL/apoB in all racial subgroups, ranging from r=-0.50 in Black females to r=-0.34 in Hispanic males.
Further analyzing data according to apo(a) tertiles, it is evident that a strong correlation (range r= 0.85-0.79) is noted between OxPL/apoB and Lp(a) in the group with 12-20 kringle IV repeats, irrespective of race. A more modest correlation is noted in all racial subgroups in OxPL/apoB and 21-26 kringle IV repeats (range r-values 0.84-0.43), with the strongest correlations in Blacks and Hispanics. However, weak to no correlations were noted in all racial subgroups with OxPL/apoB and 27-41 kringle IV repeats (Table 2).
In the overall cohort, after adjusting for apolipoprotein(a) isoform size, the relationship between OxPL/apoB and Lp(a) remained significant (r=0.67, p<0.001). Also, the partial correlation adjusting for the sum of the 2 kringle isoforms was 0.686 (p<0.001). For Black, White, Hispanic, the correlations are: 0.729, 0.612, 0.540, respectively, controlling for the sum of the two isoforms and 0.714, 0.582, 0.510, controlling for both major and minor isoforms (all p<0.001).
Adjusting for the socioeconomic status variables of income and education level did not affect any of the above relationships (data not shown).
In the overall cohort, correlations of OxPL/apoB with blood pressure were very weak (r=0.078 for systolic pressure and r=0.089 diastolic pressure, p<0.001 for both) and not significant with BMI (r=0.019, p=0.27). OxPL/apoB levels were higher in subjects with versus without hypertension (medians of 4910 vs. 3755 RLU, p<0.001) but not in those with versus without diabetes (medians of 4198 vs. 4008 RLU, p=0.135). There were very weak relationships between OxPL/apoB blood glucose levels (Spearman's r = -0.051, p=0.003).
The relationship of OxPL/apoB and Lp(a) to a variety of laboratory variables is shown in Table 3. Although many of these correlations were statistically significant, most were quite weak. Of interest, there was a negative association of both OxPL/apoB and Lp(a) with triglyceride levels and liver fat content measured with MRI. Both OxPL/apoB and Lp(a) correlated very weakly with hsCRP. There were also weak correlations between BNP and OxPL/apoB and Lp(a). There was no relationship between OxPL/apoB or Lp(a) with MCP-1, PGLYRP-S, or troponin levels. Lp-PLA2 mass correlated inversely but weakly with OxPL/apoB and Lp(a) as did Lp-PLA2 activity. No major differences were noted among the racial groups in these associations.
This large epidemiological study of apparently healthy, young to middle aged individuals demonstrates that significant racial differences exist in OxPL/apoB levels, with Black subjects displaying the highest levels, followed by Whites and then by Hispanics. OxPL/apoB levels were largely independent of most clinical and laboratory variables. However, OxPL/apoB levels were distributed in a manner consistent with genetically determined Lp(a) levels,22,23 and furthermore, were positively correlated with small apolipoprotein(a) isoforms. Interestingly, when the data were analyzed according to the size of the major apolipoprotein(a) isoform, the racial differences in OxPL/apoB were less marked. In fact, in subjects of all racial subgroups with isoform sizes ≥30 kringle IV repeats, no significant relationship was noted between OxPL/apoB and Lp(a). These data suggest that OxPL/apoB levels may reflect a key component of the atherogenicity of elevated Lp(a) levels, particularly in the presence of small apo(a) isoforms.
Previous studies documenting the relationship between OxPL/apoB and Lp(a) were performed primarily in White populations (MAYO,2 Bruneck3,9 and MIRACL6 Studies). The Dallas Heart Study documents that this relationship also exists in Blacks and Hispanics, but is quantitatively different in these groups compared to Whites. For example, the highest correlation of OxPL/apoB with Lp(a) was present in Blacks and the weakest was present in Hispanics, compared to Whites. High Lp(a) levels are inherited as a dominant quantitative trait and most subjects (~80%) express 2 distinct apolipoprotein(a) alleles, although subjects with one allele or null alleles have been described.24 Elevated Lp(a) levels can be found in a large proportion of individuals in most racial groups, with the prevalence being lowest in Whites and Asians.25,26 The median Lp(a) levels in Black subjects, and in Asian Indians from southern locations,27 are 2-4 fold higher compared to Whites and up to 68% of Blacks have Lp(a) levels >30 mg/dl whereas levels above this threshold are present in ~25% in Whites.23,26,28,29
In Whites, elevated Lp(a) levels are generally associated with small apo(a) isoforms in over 80% of subjects. In Blacks, elevated Lp(a) levels are distributed over a broader range of apo(a) isoform sizes, and plasma Lp(a) levels are higher in Blacks within the same apo(a) isoform sizes as also documented in this study.23,28,30 In particular, differences in Lp(a) levels between Blacks and Whites are prominent in the range of 20-25 kringle IV repeats, but the underlying reasons why Blacks have higher Lp(a) levels than Whites for similar isoform size is not understood.28 The current data are consistent with the interpretation that OxPL/apoB levels are also genetically determined in most subjects in a manner parallel to Lp(a), and also highly reflect apo(a) isoform size.
The OxPL/apoB assay was initially designed as a method to quantitate minimally oxidized phospholipids on apoB particles, as a measure of “oxidized LDL”. Unexpectedly, in all clinical studies done to date, a significant correlation has been found between OxPL/apoB and Lp(a), in the range of R=0.80-0.90.2-7 This finding was also confirmed in the current study, but a more extended analysis of this study provides further, novel insights into this relationship. For example, on a linear scale the relationship of OxPL/apoB to Lp(a) appears positive and roughly linear, but on a logarithmic scale, a “reverse L” shape was noted with a flat relationship up to Lp(a) levels of approximately ~30 nmol/L and then a log-linear relationship at Lp(a) levels >30 nmol/L. Part of this association may be related to the different distribution of Lp(a) levels in this study, where Black patients had highest levels compared to Whites and Hispanics, thus potentially creating an artificial relationship. However, this is not likely the explanation, as a reverse L shape can be seen in all racial groups. This is further substantiated when the data are evaluated by apolipoprotein(a) isoform size demonstrating that this reverse L relationship was evident in all racial groups with isoforms having <26 but not >27 kringle IV repeats. Furthermore, adding both kringle sizes together from each isoform did not substantially change the findings. This also suggests that a threshold effect may occur in this relationship and that a critical value of Lp(a) or a critical number of apolipoprotein(a) particles may be needed to mediate significant binding of OxPL. Further work is required to quantitatively ascertain the extent to which Lp(a) may bind and release OxPL and to identify factors, such as lipoprotein-associated phospholipase A2, that may affect this relationship.31
In support of the ability of Lp(a) as opposed to other lipoproteins in binding OxPL recognized by E06, recent data from our laboratory demonstrated that more than 85% of E06 reactivity (i.e. OxPL) co-immunoprecipitates with Lp(a).8 In lipoprotein ultracentrifugation experiments, nearly all OxPL associated with lipoproteins were found in fractions containing apolipoprotein(a), as opposed to other apolipoproteins. Furthermore, in vitro transfer studies showed that oxidized LDL preferentially donates OxPL to Lp(a), as opposed to LDL, in a time and temperature dependent manner, even in aqueous buffer. Additionally, ~50% of E06 immunoreactivity could be extracted from isolated Lp(a) following exposure of plasma to various lipid solvents. These in vitro studies lend further proof that OxPLs are strongly associated physically with Lp(a) and corroborate the insights derived from clinical populations into its physiological function and mechanisms of atherogenicity.32 Consistent with the above observations, studies from independent laboratories have demonstrated that small dense LDL contains electronegative LDL which is enriched in lipoperoxides and has enhanced susceptibility for generating OxPL.33 Furthermore, electronegative LDL is enriched in Lp-PLA2, which may mediate some of its atherogenic properties.34
It is well known that elevated Lp(a) levels are associated with increased risk of cardiovascular disease in White subjects, particularly when elevated LDL-C levels are also present,35 but this association has not consistently been demonstrated in Blacks.36,37 In females, Lp(a) appears to be a stronger risk factor when very high Lp(a) levels (>65 mg/dl) are present38 However, apolipoprotein(a) phenotypes, particularly those where apolipoprotein(a) contains <22 kringle IV repeats, appear to be associated with increased cardiovascular risk in both Whites and Blacks and are more predictive of cardiovascular disease than elevated Lp(a) levels.23,28,39,40 In the current study OxPL/apoB levels correlated most strongly with both elevated Lp(a) levels and with smaller apolipoprotein(a) isoforms, irrespective of race, age and gender, whereas they did not correlate in the larger apolipoprotein(a) isoforms, suggesting that this relationship is related to underlying genetic differences in apolipoprotein(a) size and/or number rather than race per se. However, since Blacks had higher Lp(a) levels, they also had correspondingly higher OxPL/apoB levels than Whites and Hispanics. Identifying the potential sites on apolipoprotein(a) and Lp(a) that mediate binding of OxPL and identifying the specific OxPL on these lipoproteins merits further exploration.
This is the largest study to examine the relationship of OxPL/apoB levels to cardiovascular risk factors and laboratory variables in an epidemiological cohort. Due to the large number of subjects, there are some statistically significant associations. However, most of these are quite weak or borderline and these correlations are unlikely to exert a significant pathophysiological influence on cardiovascular risk. Notably, OxPL/apoB levels were either not correlated or minimally associated with several inflammatory variables (hsCRP, IL-18, MCP-1, PGLYP-1), markers of myocardial damage (troponin) or increased left ventricular wall stress (BNP). Interestingly, a negative association was noted between both OxPL/apoB and plasma triglyceride level and liver fat content by MRI. This observation has been made previously for Lp(a), although the reasons are not well understood and require further mechanistic insights.41,42 It is also conceivable that anti-PC antibodies may modulate plasma levels of OxPL/apoB. Although such levels were not measured in this study, previous studies have not documented any significant relationship between plasma levels of OxPL/apoB and IgG or IgM autoantibodies to copper oxidized LDL (CuOxLDL), a fraction of which may be anti-PC antibodies, or with autoantibodies to MDA-LDL or to apoB-immune complexes.43 It has been reported in several studies that very low levels of IgM autoantibodies to PC-BSA,44,45 as well as IgM CuOxLDL and MDA-LDL,43 are associated with higher incidence of various manifestations of atherosclerosis and clinical events. It is also possible that such antibodies may also have therapeutic potential through a variety of mechanisms.46 Additional determinants of OxPL/apoB plasma levels besides genetically determined Lp(a) will require future studies.
In a previous publication from the Dallas Heart Study, Lp-PLA2 mass and activity were lowest in Blacks compared to Whites and Hispanics.14 In the current analysis, weak inverse correlations were noted between OxPL/apoB and Lp(a) and both Lp-PLA2 mass and activity. These data are consistent with the physiological action of Lp-PLA2 in cleaving the oxidized fatty acid at the sn-2 position of OxPL and suggest that there may also be an influence of plasma levels of these measures. Furthermore, since Lp(a) is preferentially enriched in Lp-PLA2 activity on an equimolar basis compared to LDL, Lp-PLA2 activity on Lp(a) particles may serve as a mechanism to clear OxPL bound to Lp(a).31
This study did not measure clinical outcomes and therefore it cannot be determined if racial differences in OxPL/apoB predict clinical outcomes.
This study documents that elevated levels of pro-inflammatory oxidized phospholipids carried primarily by Lp(a) represent a genetic predisposition to increased oxidative stress. Furthermore, the study suggests that the differences in apolipoprotein(a) isoforms explain some, but not all, of the racial differences in OxPL/apoB and Lp(a) seen in this population. These findings help to understand the mechanistic underpinnings of the potential atherogenicity, and potentially the inhibition of fibrinolysis by Lp(a), which has previously been documented in vitro. Future studies should focus on apolipoprotein(a) isoforms and their relationship to cardiovascular events mediated by OxPL.
The authors thank Helen Hobbs for her support and assistance in establishing this investigation.
Funding Sources These studies were supported by grants from the Donald W. Reynolds Foundations to the University of Texas Southwestern, to the Brigham and Women's Hospital and to the University of California San Diego, by the Fondation Leducq and by the General Clinical Research Center, University California, San Diego with funding provided by the National Center for Research Resources, M01RR00827, USPHS.
Disclosures Drs Witztum and Tsimikas are named as inventors in patents and patent applications for the potential commercial use of antibodies to oxidized LDL.