Lipid and Apolipoprotein Predictors of Atherosclerosis in Youth
Apolipoprotein Concentrations Do Not Materially Improve Prediction of Arterial Lesions in PDAY Subjects
Abstract—We compared serum lipid and apolipoprotein predictors of atherosclerosis in cases from the multicenter study, Pathobiological Determinants of Atherosclerosis in Youth (PDAY). The lipid measures included HDL cholesterol (HDL-C) and non-HDL-C, and the apolipoprotein measures included concentrations of apoA1, apoB, and Lp(a), and sizes of the apo(a) proteins. We tested whether the apolipoprotein measures predicted atherosclerotic lesions as well as the more traditional lipid measures. We estimated extent of lesions as fatty streaks or raised lesions (fibrous plaques, complicated or calcified lesions) in 3 sites: thoracic aorta, abdominal aorta, and right coronary artery. Neither apoA1 nor apoB measures were as strongly or consistently correlated with extent of lesions as the corresponding lipid measure (HDL-C and non-HDL-C, respectively). Beyond the basic model that included sex, age, race, smoking status, hypertension, and the lipid measures, apoA1 and apoB added only an average 1.3% increased explanatory ability to the model, whereas HDL-C plus non-HDL-C added an average 2.5%. The results suggest that the traditional lipid measures are more useful than apolipoprotein measures for detecting young persons at high risk of precocious atherosclerosis. Because of large racial differences, the two Lp(a)-related measures, Lp(a) concentrations and apo(a) size, were evaluated in blacks and whites separately. Under these circumstances, neither of the Lp(a)-related measures was strongly or consistently correlated with extent of lesions.
- Received September 15, 1998.
- Accepted October 14, 1998.
Early studies of the relationship of plasma lipids to atherosclerosis and coronary heart disease (CHD) emphasized the cholesterol content of plasma, although it was known that cholesterol and other lipids were transported as lipid-protein complexes. Ultracentrifugation separated these lipoproteins into classes by their hydrated density1 ; electrophoresis, by electrical charge2 ; and chemical methods, by selective precipitation.3 After the specific protein components were isolated and characterized, methods were developed to measure the proteins (apolipoproteins) directly by immunochemical methods,4 including a rapid and reliable electroimmunoassay method.5 Alaupovic6 first suggested that measuring the apolipoproteins might be more useful in predicting risk of CHD than the lipid components or intact lipoprotein particles.
During the 1970s and 1980s, a number of studies of the ability of plasma or serum apolipoprotein concentrations to predict risk of atherosclerotic disease produced conflicting and inconclusive results (reviewed by Sniderman and Silberberg7 and Vega and Grundy8 ). Some of these studies examined the association of apolipoprotein levels with atherosclerotic lesions measured by angiography or ultrasound in living persons, but none examined the relationship with atherosclerosis measured directly in the arteries of autopsied persons.
A multicenter cooperative study of atherosclerosis in young persons, Pathobiological Determinants of Atherosclerosis in Youth (PDAY),9 has provided an opportunity to compare the usefulness of serum concentrations of apoB and apoA1 with the usefulness of lipoprotein cholesterol concentrations to predict atherosclerotic lesions in 15- to 34-year-old men and women and to determine whether measures of Lp(a) add to prediction of atherosclerotic lesions. The results show that serum lipoprotein cholesterol concentrations are generally better predictors of atherosclerosis than the apolipoprotein concentrations.
Fifteen cooperating centers adopted a standard operating protocol and manual of procedures to collect specimens and information and to submit them to central laboratories for analysis. A statistical coordinating center received all data pertaining to each case from the collection centers and central laboratories.
Subjects were persons, 15 through 34 years of age, who died of external causes (accident, homicide, or suicide) within 72 hours of injury and who were autopsied within 48 hours of death in one of the cooperating medical examiners’ laboratories. Age and race were obtained from death certificates. Persons of race other than white or black were excluded from the study. The Institutional Review Board of each participating center approved the use of tissue, blood, and data from the human subjects in this study.
Additional details of criteria for inclusion and exclusion have been provided elsewhere.10 11 Because we previously have shown substantial effects of elevated glycohemoglobin10 on lesions, we also excluded individuals with high levels of glycohemoglobin (≥8%)10 or missing values. Altogether, 715 cases were available for this analysis.
Dissection and Preservation of Arteries
The pathologist removed the aorta from a point 2 cm proximal to the ligamentum arteriosum to a point 2 cm distal from the iliac bifurcation. The pathologist opened the aorta along a line on the dorsal surface that passed midway between the orifices of the intercostal and lumbar arteries, rinsed the intimal surface with Hanks’ modified balanced salt solution, flattened it with the adventitial surface down, split it longitudinally along a line that bisected the celiac, superior mesenteric, and inferior mesenteric ostia, and fixed the left half in 10% neutral buffered formalin. The pathologist opened the right coronary artery along the epicardial surface from its origin to the point at which it turned downward along the posterior interventricular sulcus and dissected the artery from the heart. After removing epicardial fat, the artery was rinsed and fixed as for the aorta. The collection centers placed each aorta and coronary artery in a plastic bag with 10% formalin and shipped them monthly to the central laboratory. The central laboratory stained the arteries with Sudan IV12 and packaged them in clear plastic bags with 10% formalin.
Grading Arterial Specimens
Three pathologists independently graded the preserved specimens. Using procedures developed in the International Atherosclerosis Project,12 they visually estimated the extent of intimal surface involved with fatty streaks and with raised lesions (fibrous plaques, complicated lesions, and calcified lesions). For each specimen, the average of the 3 gradings was used in the analyses; intraclass correlations among the pathologists have been reported elsewhere.9 13
Blood was collected at autopsy from the aorta, heart, or vena cava, and serum was isolated by low-speed centrifugation. Frozen serum was shipped to the central laboratory for storage at −80°C and later analysis.
The central laboratory measured serum cholesterol concentrations by the cholesterol oxidase method,14 precipitated apoB-containing lipoproteins in serum by heparin-Mn2+, and measured HDL cholesterol (HDL-C) in the supernatant.15 Non-HDL-C was calculated as the difference between total cholesterol and HDL-C concentrations. Coefficients of variation for total cholesterol and HDL-C measurements were 1.3% and 5.2%, respectively.11 Several studies have demonstrated that postmortem serum cholesterol levels are representative of premortem levels.16 17 18 However, because emergency medical teams often administer fluids intravenously to trauma patients, we excluded serum data from individuals with total serum cholesterol <2.59 mmol/L (100 mg/dL).
The central laboratory measured apoA1 and apoB concentrations by immunoprecipitin analysis using antibody reagents and calibrators from INCSTAR, Inc. and an Express Plus clinical chemistry analyzer (Ciba-Corning Diagnostics). After an initial incubation of the sample plus antibody diluent and measurement of a sample blank, monospecific antibodies were added and mixed, and the increase in light scattering was measured at 340 nm. Samples were assayed in duplicate and the average values were used. Coefficients of variation for these measurements were 5.6% for apoA1 at 129 mg/dL and 7.6% for apoB at 115 mg/dL.
The central laboratory measured Lp(a) concentrations by a “sandwich”-style ELISA as described previously.19 20 Capture was with polyclonal antibodies directed against human Lp(a) (Calbiochem), and detection was with monoclonal antibody LHLP-1, which is specific for Lp(a) or apoLp(a), but does not bind apo(a) alone.21 Assays were standardized with reference materials from INCSTAR and showed no cross-reactivity with plasminogen (data not shown). Each sample was run in duplicate on different plates and the values were averaged. Coefficients of variation for control products were 9.7% at 20 mg/dL and 5.8% at 46 mg/dL.
A laboratory at the Southwest Foundation for Biomedical Research characterized apo(a) size polymorphism by use of polyacrylamide gradient gel electrophoresis and immunoblotting as described.22 23 Samples were run at least in duplicate and size estimates were averaged. Repeatability of size estimates was 99.1%. Although most apo(a) isoform phenotypes appeared normal, some of the samples (5.5%) were obviously degraded (eg, ≥3 bands or immunoreactive material with no distinct bands) and no phenotype could be assigned. The prevalence of null phenotype samples (ie, no immunoreactive material in the lane) was 1.0%, lower than the 2.9% previously reported for Mexican Americans.24 This difference suggested that the PDAY apo(a) phenotypes were indicative of antemortem phenotypes (ie, the prevalence of null phenotype samples was not unusually high owing to degradation). A single predominant isoform band was identified for each sample. In the case of double-banded phenotype samples, samples were immunoblotted with 2D1, a monoclonal antibody believed to be directed against a nonrepeated epitope of human apo(a),25 and the predominant isoform band was identified by curve-fitting procedures.23 24 The smaller isoform was predominant for about 74% of the samples, compared with 79% observed in previous studies of Mexican Americans (data not shown). Null phenotype samples were excluded from analyses involving apo(a) isoform size.
We analyzed the associations of sex, race, 5-year age group, apo(a) size, and concentrations of non-HDL-C, HDL-C, apoB, apoA1, and Lp(a) with extent (percent arterial intimal surface area involved) of atherosclerotic lesions by multiple linear regression analysis.26 The linear model included main effects and selected 2-factor interactions. We applied a logit transformation to the proportion of surface area involved with lesions after adding a small constant to avoid the logarithm of zero.27 We applied a logarithmic transformation to the serum lipoprotein and apolipoprotein concentrations. Previously, we have shown strong associations of lesions and hypertension (obtained from measurements of the small renal arteries)28 29 and smoking (obtained from the serum thiocyanate concentration measured postmortem)11 with lesions. We adjusted for the effects of hypertension and smoking rather than exclude cases because the numbers of these cases were appreciable (104 hypertensives and 286 smokers).
Because the number of cases available for these analyses was small, we limited our model to second-order terms.26 Previous analyses10 11 28 and preliminary analyses conducted as part of this analysis indicated no interactions of sex or race and the lipid or apolipoprotein concentrations or apolipoprotein size. Therefore, such interactions were not included in the model. The effect of sex was included in the model, allowing men and women to have different extents of lesions, and, similarly, the effect of race was included in the model, allowing blacks and whites to have different extents of lesions. Because Lp(a) levels are different in blacks and whites, we conducted separate analyses for the two races. Interactions of age with the risk factor variables were included in the model. Preliminary analyses did not indicate interactions among the risk factors and these were not included in the model. We did, however, investigate the interaction of non-HDL-C and Lp(a) levels.
Lipid and Apolipoprotein Values
Table 1⇓ gives the mean lipid and apolipoprotein values according to sex, race, and age group for 715 cases. Non-HDL-C levels increased with age in both sexes (P=0.0020; no interaction of sex with age), but this increase was more pronounced in whites (race and age interaction, P=0.0398). Blacks had higher Lp(a) levels than did whites (P=0.0001), and white women had higher Lp(a) levels than white men, but black women did not differ from black men (sex and race interaction, P=0.0172). Table 2⇓ presents the correlations among the lipid and apolipoprotein measures. Most of the correlations were statistically significant, but no variable explained >25% of variation in another variable. The strongest correlations were between the 2 measures of β-lipoproteins (non-HDL-C and apoB) and the 2 measures of HDL (HDL-C and apoA1). Lp(a) concentrations were associated with the other measures, although not strongly.
Relationships of Lipid and Apolipoprotein Measures With Lesions
Lipid and apolipoprotein measures were categorized into thirds. For non-HDL-C and HDL-C, cut points were based on published values11 for a larger group from the PDAY study and are only approximate for this subset. For apoA1 and apoB, cut points were based on values from the cases used in this analysis. Because of the large differences among races, Lp(a) cut points were determined for each race separately.
Table 3⇓ gives mean lesion involvement by thirds of non-HDL-C and apoB. Non-HDL-C was positively associated with extent of fatty streaks in all 3 arteries (P=0.0001) and with extent of raised lesions in the abdominal aorta (P=0.0465) and the right coronary artery (P=0.0103). ApoB was significantly, although not as strongly, associated with fatty streaks in all 3 arteries, but not with raised lesions.
Table 4⇓ gives mean lesion involvement by thirds for HDL-C and apoA1. HDL-C was inversely associated with fatty streaks in all 3 arteries (thoracic aorta, P=0.0019; abdominal aorta, P=0.0013; coronary artery, P=0.0308) and with raised lesions in the thoracic aorta (P=0.0189) and the right coronary artery (P=0.0186). ApoA1 was inversely associated with fatty streaks in the thoracic (P=0.0464) and abdominal (P=0.0384) aortas and with raised lesions only in the thoracic aorta (P=0.0011).
Table 5⇓ gives mean lesion involvement by thirds for Lp(a). Because of large differences in Lp(a) concentrations, blacks and whites were analyzed separately. Except for fatty streaks in the right coronary artery of whites (P=0.0006), Lp(a) concentrations were not associated with either fatty streaks or raised lesions. Lp(a) was positively associated with raised lesions in the thoracic (Lp(a) and age interaction, P=0.0075) and abdominal aorta (Lp(a) and age interaction, P=0.0033) in 25- to 34-year-old blacks and was negatively associated with raised lesions in the abdominal aorta (Lp(a) and age interaction, P=0.0487) in 25- to 34-year-old whites. To investigate whether the association of Lp(a) with lesions depends on the level of non-HDL-C (that is, an interaction of Lp(a) and non-HDL-C), we analyzed the data after including non-HDL-C levels (dichotomized at the median value of 3.35 mmol/L). There was no significant interaction of Lp(a) and non-HDL-C in their effects on lesions, and including non-HDL-C levels did not abolish the significant association of Lp(a) levels with extent of lesions in the right coronary artery of whites (results not shown). Similarly, neither apoB concentrations nor the ratio non-HDL-C/HDL-C interacted with Lp(a) concentrations in their effects on lesions (results not shown).
Table 6⇓ shows that apo(a) size was associated with extent of lesions for several categories (fatty streaks of the abdominal aorta [P=0.0261] and right coronary artery [P=0.0058] of blacks and raised lesions of the abdominal aorta of whites [P=0.0098]). The association of apo(a) size with raised lesions of the abdominal aorta in whites remained after adjusting for Lp(a) concentration (P=0.0184), but the two associations in blacks were not significant after adjustment for Lp(a) concentrations (results not shown).
Fraction of Lesion Involvement Explained by Lipoprotein Measures
Table 7⇓ shows the fraction of lesion involvement explained by the basic model (sex, age, race, smoking, and hypertension), by the basic model plus lipid measurements, by the basic model plus apolipoprotein measurements, and by the basic model plus lipids and apolipoproteins. The basic model explained between 6.2% and 12.1% of the variation in fatty streaks and between 7.8% and 23.5% of the variation in raised lesions. Including the lipid data (model 2), the apolipoprotein data (model 3), or both (model 4) increased the fraction of variation explained. However, most of the improvement in prediction by the combined model was caused by including the lipid data. Including the apolipoprotein data increased the average R2 for 6 lesion and artery categories by 1.3%, whereas including the lipid data increased the average R2 by 2.5%. Only for raised lesions of the thoracic aorta did apolipoprotein data significantly improve the models beyond the basic covariates plus lipids information.
Table 8⇓ shows results of analyses that included lipid and Lp(a) data for blacks and whites considered separately. The basic model (sex, age, smoking, and hypertension) explained between 2.6% and 16.5% of the variation in fatty streaks and between 3.2% and 25.6% of the variation in raised lesions. Including the lipid data yielded an average 4.7% increase in R2 and a significant increase in 7 of the race/lesion/artery categories. However, with 3 exceptions (fatty streaks in the right coronary arteries of blacks and whites and raised lesions in the abdominal aorta of whites), the Lp(a) data did not provide a significant improvement over the model with the basic covariates plus lipids.
Summary of the Results
Compared with apoB levels, non-HDL-C concentrations were more strongly associated with extent of fatty streaks and raised lesions in the aorta and right coronary artery of young people aged 15 to 34 years. Similarly, HDL-C was more strongly associated with lesions than was apoA1. When both non-HDL-C and HDL-C were included in the predictive model, the 2 apolipoprotein measures generally did not provide a substantial increase in explanatory capability. Compared with Lp(a) concentration, apo(a) size was a slightly better predictor of lesions, but neither concentration of Lp(a) nor apo(a) size was associated with lesions as consistently as were the other lipoprotein measures.
Limitations of the Study
Hemodilution results from fluids administered in some cases to trauma patients. After excluding samples with total cholesterol levels <2.59 mmol/L, we have found that HDL-C and non-HDL-C levels matched well the values reported in other studies for similar sex, age, and race categories.11 As with non-HDL-C and HDL-C, the serum concentrations of apoB and apoA1 are also similar to those reported for young men and women.30 Furthermore, the average Lp(a) concentration for blacks in this study (50.7 mg/dL) was similar to 48.1 mg/dL measured with the same assay in a study of living blacks (n=45; W.D. Scheer, unpublished observations, 1998). Although the data suggest the values from these postmortem samples are similar to those taken from living individuals, any additional variation in the measurements of the lipoproteins and apolipoproteins because of hemodilution is expected to reduce or degrade the associations of the lipoproteins and apolipoproteins with atherosclerosis.
Postmortem degradation of proteins also may affect the results. In earlier work with dogs and nonhuman primates, we found no appreciable differences in apoA1 and apoB concentration values measured in postmortem and antemortem serum samples.18 Kronenberg and colleagues31 have shown that concentrations of Lp(a) particles bearing small apo(a) isoforms, the ones most often reported to be associated with CHD, decrease more rapidly in frozen serum or plasma than larger isoforms. Serum from PDAY cases was stored at −80°C for 1 to 8 years before analysis, so it is possible that Lp(a) concentrations were higher in life than we measured postmortem, thus blunting possible associations with lesions. A related issue is whether apo(a) isoform phenotypes might be altered by postmortem degradation. However, we have found that postmortem apo(a) isoform phenotypes were identical to those in antemortem samples from baboons.32 Our studies of 57 postmortem samples showed that in no case did postmortem degradation create a false apo(a) phenotype.32a Furthermore, after excluding obviously degraded samples from PDAY cases, we found that the frequencies of null and single-banded phenotypes were similar to those in other populations.33
Lipoprotein subpopulation phenotypes are emerging as potentially important predictors of atherosclerosis risk. Subpopulations of HDLs and apoB-containing lipoproteins (eg, Lp(a) and small dense LDLs) have been reported to be associated with risk of CHD and with severity of atherosclerosis.34 35 36 37 38 Except for Lp(a), however, we have not examined in detail associations of lipoprotein subpopulations with extent of lesions owing to the potential for postmortem enzymatic modifications of lipoprotein particle phenotypes.
Relationship of ApoB and ApoA1 Concentrations to Atherosclerosis
Early reports of the capability of apolipoprotein concentrations to predict atherosclerosis and CHD suggested that they might be better predictors than the lipid components of lipoproteins.39 40 41 42 43 However, later reports showed the predictive ability of the apolipoproteins to be similar to, but weaker than, those for the lipid components of lipoproteins, and that apolipoproteins did not add predictive or discriminatory ability over that of the lipid components.44 45 46 47 The controversial nature of the issue was reflected in the titles of reviews and editorials concerned with the topic.7 8 48 Recent results from the Québec Cardiovascular Study suggested that apoB concentrations, in conjunction with insulin levels and LDL particle size, provide additional information on the risk of ischemic heart disease compared with conventional lipid measures.49
Most of these studies of apolipoproteins related serum or plasma levels to clinical manifestations of CHD. A few used angiographic measures of coronary artery lesions as end points. No other studies have examined the association of apolipoprotein concentrations with atherosclerotic lesions measured directly in autopsied persons. Although apolipoprotein measures did increase R2 values for raised lesions of the thoracic aorta (Table 7⇑), the present results generally suggest that the lipid measures are better predictors of atherosclerosis and are consistent with the results of most recent studies.
Relationship of Lp(a) With Lesions
With few exceptions,50 51 epidemiological studies have found a positive association of Lp(a) concentrations and cardiovascular disease. The basis for this association is not known, but probably depends in part on the similarity of apo(a), the diagnostic protein component of Lp(a), to plasminogen, the zymogen for clot lysis.52 53 The potential for apo(a) to interfere with fibrinolysis and to bind to fibrin surfaces has been discussed elsewhere.54 55 Of course, Lp(a) particles also may contribute to intimal lipid deposition as do LDL particles. Although not as consistently as the lipid measures, Lp(a) concentrations were significantly associated with lesions in blacks and whites. We tested the hypothesis, suggested by other investigators,56 that Lp(a) might be more atherogenic in the presence of high levels of non-HDL-C, or in the presence of high total cholesterol/HDL-C ratios.57 However, we found no evidence for interaction between Lp(a) and non-HDL-C concentrations, nor between Lp(a) concentrations and total cholesterol/HDL-C ratios.
Several studies have reported isoform-specific variation in lysine and fibrin binding properties of apo(a).58 59 60 61 Such variation could affect the ability of Lp(a) particles to interact with the vessel wall and has led to the hypothesis of significant variation in the relative atherogenicity of Lp(a) particles bearing different isoforms. Several groups have reported significant associations of apo(a) protein size with measures of cardiovascular disease,62 63 64 65 but for the most part, these associations were not independent of Lp(a) concentrations. That is, because of their well-known inverse relationship, apo(a) size appeared to behave in the models as a surrogate measure of Lp(a). End points for these previous studies included measures of CHD, but not atherosclerosis.
We found that apo(a) size was a significant predictor of fatty streaks in blacks and abdominal aortic raised lesions in whites. The association with fatty streaks became nonsignificant after adjusting for Lp(a) concentration, but the association with raised lesions in the abdominal aorta of whites remained significant (P=0.0184). Thus, the contribution of both Lp(a) concentration and apo(a) size to atherosclerosis in adolescence and young adulthood is, at best, small in comparison with the effects of the 2 major lipoprotein classes and smoking. Taken together with the abundant evidence that Lp(a) concentrations or apo(a) size or both are associated with increased risk of CHD, these results suggest that Lp(a) probably acts by interfering with thrombolysis in the advanced and terminal occlusive stages of atherosclerosis rather than augmenting the early and intermediate stages.
Implications for Primary Prevention
These results suggest that detection of high-risk individuals and control of risk factors in teenagers and young adults should be focused on total serum cholesterol and lipoprotein cholesterol concentrations, and not on apolipoprotein or Lp(a) concentrations.
The PDAY Research Group
The investigators cooperating in the multicenter study, “The Pathobiological Determinants of Atherosclerosis in Youth,” and the grants supporting their activities, are listed below.
Program Director: Jack P. Strong, MD, 1996-date; Robert W. Wissler, PhD, MD, 1985–1996.
Steering Committee: J. Fredrick Cornhill, DPhil; Henry C. McGill, Jr., MD; C. Alex McMahan, PhD; Gray T. Malcom, PhD; Margaret C. Oalmann, DrPH; Jack P. Strong, MD; Robert W. Wissler, PhD, MD.
Participating Centers, Principal and Co-investigators, and Supporting Grants from the National Heart, Lung, and Blood Institute: University of Alabama, Birmingham, AL: Department of Medicine, Steffen Gay, MD (grant HL-33733), and Department of Biochemistry, Edward J. Miller, PhD (HL-33728); Albany Medical College, Albany, NY: Assad Daoud, MD (HL-33765); Baylor College of Medicine, Houston, TX: Louis C. Smith, PhD (HL-33750); University of Chicago, Chicago, IL: Robert W. Wissler, PhD, MD, Dragoslava Vesselinovitch, DVM, MS, Robert J. Stein, MD, and Robert H. Kirschner (HL-33740; HL-45715); The University of Illinois, Chicago, IL.: Abel L. Robertson, Jr., MD, PhD, Edmund R. Donoghue, MD, and Robert J. Buschmann, MD (HL-33758); Louisiana State University Medical Center, New Orleans, LA.: Jack P. Strong, MD, Gray T. Malcom, PhD, William P. Newman III, MD, and Margaret C. Oalmann, DrPH (HL-33746, HL-45720); University of Maryland, Baltimore, MD: Wolfgang Mergner, MD (HL-33752, HL-45693); Medical College of Georgia, Augusta, GA: A. Bleakley Chandler, MD (HL-33772); University of Nebraska Medical Center, Omaha, NE: Bruce M. McManus, MD, PhD (HL-33778); The Ohio State University, Columbus, OH: J. Fredrick Cornhill, DPhil (HL-33760, HL-45694); Southwest Foundation for Biomedical Research, San Antonio, TX: James E. Hixson, PhD (HL-39913) and David L. Rainwater, PhD (HL-50521); The University of Texas Health Science Center at San Antonio, San Antonio, TX: C. Alex McMahan, PhD, and Henry C. McGill, Jr., MD (HL-33749, HL-45719); Vanderbilt University, Nashville, TN: Renu Virmani, MD (HL-33770, HL-45718); and West Virginia University Health Sciences Center, Morgantown, WV: Singanallur N. Jagannathan, PhD (HL-33748).
Grants from the NIH supported this research and they are listed in the Appendix.
Gofman JW, Glazier F, Tamplin A, Strisower B, De Lalla O. Lipoproteins, coronary heart disease, and atherosclerosis. Physiol Rev. 1954;34:589–607.
Kunkel HG, Slater RJ. Lipoprotein patterns of serum obtained by zone electrophoresis. J Clin Invest. 1952;31:677–684.
Cohn EJ, Gurd FRN, Surgenor DM, Barnes BA, Brown RK, Derouaux G, Gillespie JM, Kahnt FW, Lever WF, Liu CH, Mittleman D, Mouton RF, Schmid K, Uroma E. A system for the separation of the components of human blood: quantitative procedures for the separation of the protein components of human plasma. J Am Chem Soc. 1950;72:465–474.
Sniderman AD, Silberberg J. Is it time to measure apolipoprotein B? Arteriosclerosis. 1990;10:665–667.
Vega GL, Grundy SM. Does measurement of apolipoprotein B have a place in cholesterol management? Arteriosclerosis. 1990;10:668–671.
McGill HC Jr, McMahan CA, Malcom GT, Oalmann MC, Strong JP, PDAY Research Group. Relation of glycohemoglobin and adiposity to atherosclerosis in youth. Arterioscler Thromb Vasc Biol. 1995;15:431–440.
McGill HC Jr, McMahan CA, Malcom GT, Oalmann MC, Strong JP. Effects of serum lipoproteins and smoking on atherosclerosis in young men and women. Arterioscler Thromb Vasc Biol. 1997;17:95–106.
PDAY Research Group. Natural history of aortic and coronary atherosclerotic lesions in youth: findings from the PDAY study. Arterioscler Thromb. 1993;13:1291–1298.
Allain CC, Poon LS, Chan CSG, Richmond W, Fu PC. Enzymatic determination of total serum cholesterol. Clin Chem. 1974;20:470–475.
Glanville JN. Post-mortem serum cholesterol levels. BMJ. 1960;2:1852–1853.
Enticknap JB. Lipids in cadaver sera after fatal heart attacks. J Clin Pathol. 1961;14:496–499.
Doetsch K, Roheim PS, Thompson JJ. Human lipoprotein(a) quantified by ‘capture’ ELISA. Ann Clin Lab Sci. 1991;21:216–224.
Doetsch KM, Roheim PS, Thompson JJ. Optimization and characterization of capture ELISA methodology for Lp(a) lipoprotein quantification. Ann Clin Biochem. 1992;29:275–282.
Duvic CR, Smith G, Sledge WE, Lee LT, Murray MD, Roheim PS, Gallaher WR, Thompson JJ. Identification of a mouse monoclonal antibody, LHLP-1, specific for human Lp(a). J Lipid Res. 1985;26:540–548.
Rainwater DL, Manis GS, VandeBerg JL. Hereditary and dietary effects on apolipoprotein[a] isoforms and Lp[a] in baboons. J Lipid Res. 1989;30:549–558.
Rainwater DL, Haffner SM. Insulin and 2-hour glucose levels are inversely related to Lp(a) concentrations controlled for LPA genotype. Arterioscler Thromb Vasc Biol. 1998;18:1335–1341.
Taddei-Peters WC, Butman BT, Jones GR, Venetta TM, Macomber PF, Ransom JH. Quantification of lipoprotein(a) particles containing various apolipoprotein(a) isoforms by a monoclonal anti-apo(a) capture antibody and a polyclonal anti-apolipoprotein B detection antibody sandwich enzyme immunoassay. Clin Chem. 1993;39:1382–1389.
Draper NR, Smith H. Applied Regression Analysis. New York: John Wiley & Sons; 1966:58–134.
Carroll RJ, Ruppert D. Transformation and Weighting in Regression. New York: Chapman&Hall; 1988.
McGill HC Jr, Strong JP, Tracy RE, McMahan CA, Oalmann MC, PDAY Research Group. Relation of a postmortem renal index of hypertension to atherosclerosis in youth. Arterioscler Thromb Vasc Biol. 1995;15:2222–2228.
McGill HC Jr, McMahan CA, Tracy RE, Oalmann MC, Cornhill JF, Herderick EE, Strong JP, PDAY Research Group. Relation of a postmortem renal index of hypertension to atherosclerosis and coronary artery size in young men and women. Arterioscler Thromb Vasc Biol. 1998;18:1108–1118.
Donahue RP, Jacobs DR Jr, Sidney S, Wagenknecht LE, Albers JJ, Hulley SB. Distribution of lipoproteins and apolipoproteins in young adults: the CARDIA study. Arteriosclerosis. 1989;9:656–664.
Kronenberg F, Trenkwalder E, Dieplinger H, Utermann G. Lipoprotein(a) in stored plasma samples and the ravages of time: why epidemiological studies might fail. Arterioscler Thromb Vasc Biol. 1996;16:1568–1572.
Rainwater DL. Stability of apolipoprotein(a) isoform phenotype to postmortem conditions. Clin Chim Acta. In press.
Rainwater DL, Kammerer CM, VandeBerg JL, Hixson JE. Characterization of the genetic elements controlling lipoprotein(a) concentrations in Mexican Americans: evidence for at least three controlling elements linked to LPA, the locus encoding apolipoprotein(a). Atherosclerosis. 1997;128:223–233.
Grundy SM. Small LDL, atherogenic dyslipidemia, and the metabolic syndrome. Circulation. 1997;95:1–4.
Sniderman AD, Cianflone K. Measurement of apoproteins: time to improve the diagnosis and treatment of the atherogenic dyslipoproteinemias. Clin Chem. 1996;42:489–491.
Lyu L-C, Shieh M-J, Ordovas JM, Lichtenstein AH, Wilson PWF, Schaefer EJ. Plasma lipoprotein and apolipoprotein levels in Taipei and Framingham. Arterioscler Thromb. 1993;13:1429–1440.
Sharrett AR, Patsch W, Sorlie PD, Heiss G, Bond MG, Davis CE. Associations of lipoprotein cholesterols, apolipoproteins A-I and B, and triglycerides with carotid atherosclerosis and coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) Study. Arterioscler Thromb. 1994;14:1098–1104.
Brunzell JD, Sniderman AD, Albers JJ, Kwiterovich PO Jr. Apoproteins B and A-I and coronary artery disease in humans. Arteriosclerosis. 1984;4:79–83.
Eaton DL, Fless GM, Kohr WJ, McLean JW, Xu Q-T, Miller CG, Lawn RM, Scanu AM. Partial amino acid sequence of apolipoprotein(a) shows that it is homologous to plasminogen. Proc Natl Acad Sci U S A. 1987;84:3224–3228.
Hopkins PN, Wu LL, Hunt SC, James BC, Vincent GM, Williams RR. Lipoprotein(a) interactions with lipid and nonlipid risk factors in early familial coronary artery disease. Arterioscler Thromb Vasc Biol. 1997;17:2783–2792.
Kang C, Durlach V, Soulat T, Fournier C, Anglés-Cano E. Lipoprotein(a) isoforms display differences in affinity for plasminogen-like binding to human mononuclear cells. Arterioscler Thromb Vasc Biol. 1997;17:2036–2043.
Sandholzer C, Saha N, Kark JD, Rees A, Jaross W, Dieplinger H, Hoppichler F, Boerwinkle E, Utermann G. Apo(a) isoforms predict risk for coronary heart disease: a study in six populations. Arterioscler Thromb. 1992;12:1214–1226.
Kraft HG, Lingenhel A, Köchl S, Hoppichler F, Kronenberg F, Abe A, Mühlberger V, Schönitzer D, Utermann G. Apolipoprotein(a) kringle IV repeat number predicts risk for coronary heart disease. Arterioscler Thromb Vasc Biol. 1996;16:713–719.