turbidimetric apolipoprotein B assays for the clinical laboratory

ELSEVIER

Clinica Chimica

Acta 223 (1993) 3 I-42

Immunonephelometrickurbidimetric apolipoprotein B assays for the clinical laboratory Stanley S. Levinson*““, “Department

of Pathology,

‘LaboratoryService.

hDepartment ‘Medical

800 Torn

Stephen G. Wagnerb3”

ofMedicitte.

Universrty

Service. Department Avenue, Lorrrsvilie,

of Loutsvrlle.

of Veterans Affairs

KY 40206-1466.

LoutsviNe.

KY.

USA

Medico1 Center.

USA

(Received 7 May 1993: revision received IS July 1993: accepted

18 July 1993)

Abstract Because apolipoproteins are a part of complex macromolecular particles, modifications to the assay system may substantially alter results of immunological measurement. Accuracy as analytical recovery cannot be effectively determined by adding exogenous apolipoproteins because antibody access differs from access to endogenous apolipoproteins. Clinical studies are essential for determining accuracy in terms of clinical effectiveness. Since different kit methods use different reagent systems, the purpose of the present study was to compare total cholesterol and LDL cholesterol as markers for coronary artery disease with apo B by automated rate nephelometric, end-point nephelometric and turbidimetric kit methods. The subjects were age matched, male patients with and without angiographically documented coronary artery disease. High correlation coefficients (0.95-0.96) between the assays for both the normal and disease groups indicate that the methods are providing similar information: apo B was a better marker for coronary artery disease (CAD) than total or LDL cholesterol on the basis of univariate, multivariate and Bayesian statistics and correlated best with nonHDL cholesterol. Apo B along with HDLC could explain the variability between the CAD and normal groups without LDLC, total C, or triglycerides. Key words:. Apo B; LDL cholesterol; terol; Triglycerides

VLDL cholesterol;

Non-HDL

cholesterol;

HDL choles-

1. Introduction Apolipoprotein B (apo B) has been shown to be a better marker artery disease than LDL cholesterol (LDLC) [l-4]. Yet, measurement * Corresponding author, Laboratory Louisville, KY 40206.1466, USA.

Service,

Veterans

0 1993 Elsevier Science Publishers 0009-8981/93/%06.00 SSDI 0009-8981(93)05659-3

Affairs

Medical

Center,

B.V. All rights reserved.

for coronary of apolipo-

800 Zorn

Avenue.

proteins is fundamentally different from measurement of most other serum proteins because they are a part of macromolecular complexes in which antigenic sites are hidden by lipids which may alter the immunoreactivity of apo B [5-lo]. Denaturating agents have been used to expose antigenic sites on apolipoproteins [8-lo]. Valid experiments for analytical recovery cannot be performed because exogenously added apolipoproteins are not a part of the endogenous complexes. Because of these characteristics, apolipoprotein assays may produce peculiar results depending on the nature of the antiserum, reagent system, detergent and measuring technique [5,6,8,9]. As a result, although a number of immunonephelometric/immunoturbidimetric kits have been evaluated analytically and have been shown to perform adequately in terms of linearity and reproducibility, [ 1 I- 141 analytical studies of linearity, precision and apparent recovery are not sufficient for demonstrating dependability of these assays for clinical laboratory usage. Clinical studies are necessary as an adjunct because, regardless of what appears to be analytical validation, apolipoprotein assays have minimal applicability for the clinical laboratory unless they are better indicators of CAD than conventional lipoprotein lipid assays. For example, a recent comparison of persons who suffered a myocardial infarction with those who did not showed little difference between LDLC and Apo B assayed by an RID kit [15]. Previously, we showed that an automated immunonephelometric kit for assaying apo B, which contains a detergent for denaturing lipoproteins, performed well on an analytical basis [ 1 l] and that this kit method performed better as a marker for CAD than LDLC [16]. We also presented evidence that the kit was measuring apo B in very low density/intermediate density lipoprotein (VLDL/IDL) as well as LDL and suggested that measurement of VLDLiIDL-apo B might be one reason for the increased diagnostic sensitivity. Here, we compared the ability of three widely used apo B kits, each with a different feature in technique (rate nephelometric, end-point nephelometric and turbidimetric), as markers for CAD as compared with apo B containing lipoprotein lipids. We also further examined the question as to whether or not these kits are measuring apo B in VLDL/IDL. 2. Materials and methods

2.1. Subjects and angiography The samples were drawn from consecutively examined patients, except for the exclusion criteria listed below. Patients were males, 44-70 years old, entering the hospitals for clinically indicated angiographic studies. The mean age of the normal group was 57.6 years old and that of the CAD group was 60.6 years old. Patients on known lipid altering medications (gemlibrozil, HMG CoA reductase inhibitors, bile acid resins, niacin and heparin), type I diabetics (juvenile) and persons experiencing a myocardial infarction within 3 months of the procedure were excluded from this study. Both groups of patients exhibited risk factors for CAD including hypertension and some patients from each group were being treated with antihypertension medications and anti-angina1 medications. We found no difference in the frequency of use of these drugs.

33

Blood nor lipoprotein studios was drawn without prsservatives( fro~th~ femoral artery priortotheinfusion ofthecontrast radium. Thepr~dure was approved by the VAMC review board. The patients were fasting overnight. Sera was obtained by centrifugation at room temperature. Cholesterol, triglycerides, and high density lipoprotein cholesterol (HDLC) were determined on fresh sera within 48 h. Aliquots were frozen at -20°C for apolipoprotein measurement. We found no effect of freezing on the measured concentrations of apo B, as compared with fresh specimens. Cardiac catheterization and cineangiography were performed by standard methods used at the VAMC, as previously described 1161. As was the case in other studies, CAD was defined as at least one major artery having more than 70% stenosis and normal arteries were defined as those having less than 20% stenosis [ 17,181. These criteria were selected in order to ensure a comparison between normal persons and persons with well-defined CAD For determining diagnostic sensitivity and specificity in these initial studies. During this study, 40 of the consecutively examined patients who conformed to the inclusion criteria exhibited stenosis greater than 20% and less than 70%. Lipoproteins from these patients were not examined. 2.2. Assays and statistics Triglycerides and cholesterol were assayed by routine enzymatic methods with the Ektachem Analyzer (Eastman Kodak Company, Rochester, NY). HDLC was extracted using dextran sulfate M, 50,000 in the presence of magnesium ions [ 19,201. These methods have consistently proven to be accurate according to the College of American Pathologists Survey Program. The cholesterol assay is routinely compared with the VAMC-CDC (Center for Disease Control) lipid standardization program and has consistently been shown to exhibit accuracy and precision of less than 3”/# as compared with the CDC reference material. LDLC and VLDLC were calculated by the Friedewald equation 1211. LDLC was not calculated from samples with triglycerides 2 4.4 mmol/l. Apo B was measured using kits, according to the manufacturers instructions, by rate immunonephelometry with the Array (Beckman Instruments, Brea, CA), by end-point immunonephelometry using the Behring nephelometer (Behring Diagnostics inc., Somerville, NJ) and by immunoturbidimetry with the Fara centrifugal analyzer (Roche diagnostic systems, Montclair, NJ). The analytical performance of each of these kit methods has been evaluated and shown to be suitable for routine clinical laboratory use with good precision [l l-141; as a result we did not collectively assess the precision of the apo B assays using an independent control, but followed the reproducibility for each assay using the controls supplied with each kit. The precision for each control and for the lipoprotein lipid assays during the period of the study is indicated below. Statistics were calculated on a Macintosh. The Mann-Whitney U-test was used for comparing the ability of each test to differentiate between the medians of the CAD and normal samples on a univariate basis. Pearson’s product moment coeflicient was used for correlation studies and forward stepwise regression with an F value of 4 to enter and an F value of 3.976 to remove was used for multivariate analysis. These were calculated using the software Statview IITM or version 4 (Abacus Concepts, Berkeley, CA). Receiver operating characteristic (ROC) curves were calculated using

Ap03 (Beckman) ApoR (Behring) Ap&3 (Roche)

Triglycerides

Non-HDLC

Total C

LDLC

Test

Table 1 Comparison

VWXi blockage

between average

64 168 65 t75 65 175 65 175 65 175 65 175 65 175

values of lipoprotein

3.07 3.46 4.84 5.25 3.85 4.39 1.63 2.00 (0.99) (1.01) (1.12) (1.07) (1.12) (1.07) (1.01) (1.06)

Mean (SD.) (mmolfij

lipids and apa B Mean (S.D.)

(0.43) (0.41) (0.43) (0.41) 1.48(0.92) 1.82 (0.96) 1.27 (0.38) 1.46(0.36) I.1 I (0.33) I .25 (0.32) 1.25 (0.33) 1.38 (0.31)

I .86 2.02 1.48 1.69

I,033(0.39)

1.18 (0.38)

(g/if mmol mmol mmd mmol mmol mmol mmol i .76 mmol 1.25 g 1.45 g 1.13 g 1.26 g 1.25 g 1.38 g

2.91 3.43 4.86 5.23 3.74 4.37 I .24

Median (units/l) 100.5 122.5 103.2 126.9 96.9 129.3 98.5 128.7 95 129.9 98.6 128.6 100.9 127.8

Mean Rank

0.0076

0.0029

0.0005

O.alZ&

0.0013

0.0191

0.0252

P Rank

35

S.S. Levinson, S.G. WagnerlClin. Chim. Acta 223 (1993) 31-42

a program available from Dr. Charles E. Metz, Department of Radiology, University of Chicago Medical Center, 5841 South Maryland Avenue Chicago, IL 60637- 1470. 3. Results During the period in which this study was conducted we observed the following results for precision: mean, coefficient of variation (C.V.) and number of samples, respectively; normal control: total C 159 (2%) n = 45; triglycerides 116 (2.4%)) n = 45; apo B-Beckman 97 (3.5%), n = 10; apo B-Behring 112 (5.1”%),n = 8; apo BRoche 121 (l%), n = 5; and for the abnormal control, respectively: total C 245

,

AREA UNDER FITTED CURVE

TEST

/-

_ w

apo B (Beckman) NON-HDLC Total Cholesterol

= = =

0.645 0.635 0.599

M

LDL Cholesterol

=

0.595

o-0

I

0

-

I

0.1

-

I

0.2

=

I

.

I

0.4

-

I

-

I

0.6

-

I

-

I

-

I

0.8

-

I

1.0

FALSE POSITIVE FREQUENCY (1 - SPECIFICITY) Fig. I. Sensitivity and specificity for lipoprotein fractions depicted by receiver operating characteristic curves. The area for each fraction is indicated. The areas for the Behring and Roche kits were 0.63 and 0.61, respectively, also greater than for LDLC or total C.

(2.2%), n = 45; triglycerides 236 (2.90/u), n = 45; apo B-Beckman 175 (2.2%) n = 10; apo B-Behring 155 (5.7”/0), n = 8; apo B-Roche 170 (1.9%)) n = 5. Comparisons of the assays by univariate statistics with the Mann-Whitney test are shown in Table 1. On the basis of the significance value (P), apo B was more powerful than LDLC or total C as a marker for CAD. Figure 1 which shows diagnostic sensitivity and specificity in the form of ROC curves also indicates that apo B was a better marker than LDLC or total C. The multivariate analysis results shown in Table 2 indicate that apo B could explain the variability between the normal and CAD groups without the need for LDLC cholesterol. When HDLC was added to the stepwise regression equation, HDLC and apo B proved to be independent predictors for CAD, and LDLC, total C and triglycerides were redundant. The relationship between the three apo B methods is shown in Fig. 2. For each test, the regression lines for the CAD and normal groups were nearly superimposable. The correlation was excellent (r varying between 0.94-0.96). The Behring method showed about a 14- 19% proportional bias as compared with the Beckman

Table 2 Stepwise regression Number steps Step 1

Step I Step 2

Step I

of

summary

comparing

apo B (Beckman)

Variables entered

Variables in model”

Apo Bb Total C LDLC

Apo B

Non-HDLC Total C

Ln TriglyceridesC LDLC Total C

Variables in model

and lipoprotein not

P values

0.002

@Coefficient

0.002

Total C LDLC b

Non-HDLC Total C LDLC

0.002 <0.0001

0.002 -0.009

LDLC

Ln Triglycerides

0.0005

0.19

0.0002

0.002

LDLC Total C LDLC

Step 2

lipids

Total C Total C

Step 3 Step I Step 2

Ln HDLCC ApoB Total C Ln Triglycerides LDLC

Ln HDLC ApoB

0.0002

-0.008

0.0005 < 0.0001

-0.38 0.002

Total C Ln Triglycerides LDLC

aVariables that were in the model in the final step. bApo B was assayed using the Beckman method. Because of the large amount of collinearity apo B and non-HDLC were not evaluated at the same time. ‘The natural logarithm (Ln) was used for HLDC and triglycerides because of their non-parametric distributions.

S.S. Levinson. S.G. Wagner/

Clin. Chim. Actu 223 (1993) 31-42

37

and Roche methods. This bias appears to influence the assay minimally as compared with the others over the physiological range of 0.4-2.6 g/l and could easily be eliminated by a mathematical adjustment. The Roche method showed a constant bias as compared with the other two. Linear correlation coefficients between all of the tests are shown in Table 3. It can be seen that the highest correlation coefficients between lipoprotein lipids and apo

APOLIPOPROTEIN 0.5

1.0

1.5

B (g/L) 2.0

2.5

1. I .8

1.6

1.6

3.8

3.

0.4

&

m

1.8

1.8

i

1.6

1.6

8 s

1.2

z ,,2

l ~20. Normal b .70- CAD

0.8

1.8

1.6

1.6

1.2

0.8

-BEHRlNG 0.5

1.0

0.4

1.5

APOLIPOPROTEIN

2.0

2.5

B (g/L)

Fig. 2. Correlations between apo B assays. 0, represents points for normal group (~20% blockage); A. represents points for CAD group (> 70%). Regression lines are given by: (I) apo B Beckman vs. apo B Behring: (Normal): y = 0.82 (0.76-0.88)x + 0.07, r = 0.96, SEE = 0.09, n = 65, P = 0.0001. (CAD): y = 0.86 (0.82-0.89)x + 0.0, r = 0.96, SEE = 0.09. n = 175, P = 0.0001; (2) apo B Beckman vs. apo B Roche: (Normal): y = 0.84 (0.78-0.9)x + 0.18, r = 0.96, SEE = 0.09, n = 65, P = 0.0001, (CAD): y = 0.81 (0.77-0.85)x+ 0.2, r = 0.96, SEE = 0.09, n = 175, P= 0.0001; (c) apo B Behring vs. apo B Roche: (Normal): y=O.97 (0.89-1.05)x + 0.17, r= 0.95, SEE = 0.11, n = 65, P= 0.0001, (CAD): y= 0.89 (0.85-0.94)x + 0.26, r = 0.94. SEE = 0.1, n = 175, P = 0.0001.

38

S.S. Luinson,

S. G. Wugnrr / Clin. Chh.

Ac,ro 223 i IYY3) JI-42

Table 3 Correlations between cholesterols and apo B Total C

LDLC

1 0.93 0.98 0.25 (0.0007)

0.92 0.11 (0.16)

0.32

1

0.91

0.85

0.93

0.29

1

0.90

0.83

0.92

0.33

0.96

1

0.88

0.81

0.91

0.3

0.96

0.94

NonHDLC

VLDLC

Apo B Apo B (Beckman) (Behring)

Apo B (Roche)

CAD group

Total C LDLC Non-HDLC VLDLC* Apo B (Beckman) Apo B (Behring) Apo B (Roche)

1 1

1

Normal group

Total C LDLC Non-HDLC VLDLC Apo B (Beckman) Apo B (Behring) Apo B (Roche)

I I

0.93 0.95 0.36 (0.004) 0.89

0.92 0.11 (0.37) 0.86

0.46

f

0.96

0.87

0.84

0.93

0.88

0.87

0.96

0.45 (0.0002) 0.44 (0.0002) 0.41 (0.0007)

1

1 0.96

1

0.96

0.95

I

*Unless indicated in parentheses the significance levels for the correlation coefficients are: P 5 0.0001.

B are with non-HDLC and that this is true for both the normal and CAD groups. It can also be seen that, although the correlations between calculated VLDLC and apo B are weak, they are nevertheless significant. In order to determine whether or not these assays were measuring apo B in VLDLiIDL as well as in LDL, we assayed a sample containing 11.99 mmolil of triglycerides (high VLDL/IDL) and 0.52 mmol/l of LDLC (low LDLC) using the Beckman apo B assay. We found that the concentration of apo B decreased linearly with dilution over a range of 1.4 g/l to 0.4 g/l. According to our data (Table I), approximately 3.2 mmolil of LDLC corresponds to about 1.3 g/l of apo B. Calculation indicates that there was about 0.2 g/l of apo B in the LDL of this sample, which is below the lower limit of detection for the assay. The observation that the concentration of apo B exhibited linear behavior with dilution, indicates that the dilution line was parallel to the calibration line. These data suggest that the assay was measuring apo B in VLDL/IDL as well as LDL and that the assay cannot distinguish between the two.

S.S. Levinson,

S.G. Wagner/Clin.

Chim. Acta 223 (1993) 31-42

39

4. Discussion A number of studies have compared apo B kits on the basis of analytical properties [ 1l- 141. Yet, because of the peculiar immunological properties of apo B in lipoprotein particles, it has been unclear whether commercial kits, using routine automated methods, would perform as effectively as research studies indicated was possible. In a prior study, we showed that one of the kits (Beckman) examined here was a better marker for CAD than total C or LDLC [ 161. This encouraged us to perform the present study in which we reexamined this kit along with two others. The data presented here indicate that the automated apo B commercial kits were better markers for CAD than total or LDLC in our population of high risk patients. On the basis of the Mann-Whitney test, the kits were a better marker for CAD than LDL or total C (Table 1) and on the basis of stepwise regression apo B could explain the variability between the CAD and normal groups without the need for total C or LDLC (Table 2). Bayesian statistics in the form of ROC curves (Fig. 1) indicate that apo B was a more effective marker for CAD than LDL, or total C. The accuracy of the test is given by the area under the curve. An area of 1 indicates perfect accuracy and an area of 0.5 no discrimination. It has been suggested that 0.5-0.7 indictes low accuracy, 0.7-0.9 accuracy useful for some purposes and > 0.9 high accuracy [22]. On this basis, although it appears that apo B was a better marker than total C or LDLC, neither showed good accuracy for identifying CAD. This is not surprising since CAD is well known to be multifactorial. After the present study was completed, Zweig and associates described very similar relationships between ROC curves when they compared LDLC with apo B assayed by the Beckman kit, also used here [22]. Our results confirm these findings. If these results are verified in future studies, measurement of apo B should substantially increase over measurement of total C or LDLC the absolute number of high risk persons that can be identified when testing is applied to large numbers of people in the general population. The correlation data in Fig. 2 indicate that the kits do not yet correspond in absolute values. The Behring method showed a proportional bias as compared with the Beckman and Roche methods, and the Roche method showed a constant bias as compared with the other two. Nevertheless, the very good correlation coefficients indicate that it should be possible to manipulate each assay so that all can produce equivalent results. The high correlation coefficients (0.95-0.96) between apo B kits obtained here are similar to those reported by others [ 141. Because of the peculiarities associated with measuring apolipoproteins, when comparing kits, it is important to compare samples from persons with known CAD as well as normal samples. Most studies comparing kits have examined samples from patient populations where a majority were probably disease free [ 1 1- 141. The importance of comparing apolipoproteins in persons with and without disease is emphasized by data obtained for apo A-I, using kits from the same three manufacturers, where correlations between the CAD group were much poorer than between the normal group and the regression lines showed lack of parallelism indicating differences in immunoreactivity between kits [23]. It seemed important to investigate

40

S.S. Levinson.

S.C.

Wagner / C‘lin. C’him. Actu 223

ilYY3J

31-42

this question with the apo B kits because more dense LDL particles which contain less carbohydrate and more triglycerides have been shown to be associated with CAD [24,25], which may alter the immunoreactivity of the particles causing discrepant results with assays using different alterations of technique or antibodies from different sources [5]. The very similar correlation coefficients and overlaping lines for both groups seen here when the kits were compared (Fig. 2) suggest that all can provide similar clinical information. This along with the similar areas under the ROC curves (Fig. 1) suggest that there is no important difference between ability of the kits to function as a marker for CAD and that the three kit methods are measuring identical phenomena in the both the normal and CAD groups. In the present study, the strong statistical associations between apo B and nonHDLC are of interest. Non-HDLC correlated better on the basis of the MannWhitney test (Table l), ROC analysis (Fig. 1) and linear correlation coefficients (Table III) with apo B than with LDLC or total C. It has long been known that VLDL/IDL lipids are associated with increased risk of CAD [29-311. Grundy suggests that apo B levels may be measuring the atherogenic potential of both VLDL and LDL [29]. This higher correlation of apo B with non-HDLC may be due to the additive effect from the strong correlation between apo B and LDLC plus the weak, but significant, correlation (Table 3) with VLDLC (derived from triglycerides). The highly linear relationship obtained between apo B and dilution of a sample containing essentially no LDL indicates that besides measuring apo B in LDL, the assay is measuring apo B in VLDLIIDL. Thus, the greater efficiency of apo B measurement over LDL and total C may be due to measurement of the atherogenic ability of VLDLIIDL. A major effort is now under way to ensure international uniform standardization for apo B assays [26-281. Pure apo B is insoluble in aqueous solution. LDL contains only apo B and can thus be standardized on the basis of its protein content, while VLDL contains several other apolipoproteins, including large amounts of apo E. As a result, the calibration material is prepared from a narrow density LDL fraction. This may cause the problem of standardization to be especially difficult for assays that measure VLDL/IDL-apo B as well as LDL-apo B, since VLDL/IDL may have different properties than the calibrator. Nevertheless, our data suggest that when the standardization process is completed and the kit manufacturers achieve correspondence in units, these apo B kits have the potential to be better devices for routine screening of CAD risk than total or LDLC. Furthermore, our data indicate that of the testing presently available in most clinical laboratories non-HDLC is a better marker than the more traditional total C and LDLC. This suggests that evaluation of this marker may improve the efficiency for identifying high risk persons. Additional studies will be needed to determine whether it is equivalent to apo B in all circumstances. Certainly, apo B holds the methodological advantage that it is directly measured, while non-HDLC is a calculated value. Besides measuring apo B in LDL and VLDL/IDL, it is possible that these assays are measuring apo B in lipoprotein(a) (Lp(a)). Lp(a) is a lipoprotein particle consisting of apo B attached to ape(a) [33] and has been shown to be atherogenic

S.S. Levinson, S.G. Wagner/Clin.

Chim. Acra 223 (1993) 31-42

41

[34,35]. Thefrequencyto whichLp(a)contributesto elevatedapoB levelsin populations is unknown, Difficulties associated with measurement of true recovery of Lp(a) is similar to that of other lipoproteins. Standardization of Lp(a) assays may also be difftcult because ape(a) and the Lp(a) particles themselves have been shown to be very heterogeneous in size [35,36]. For this reason the sensitivity of light scatter type assays for identifying apo B in Lp(a) particles may be compromised. 5. Acknowledgments

These studies have been supported by the Department of Veteran Affairs, a Barkley funded grant from the University of Louisville and gifts of reagents from Beckman Instruments, Corp. Roche and Behring. We thank Dale Pike, R.N. and Jessica Ralston M.T. (ASCP) for technical assistance. 6. References

I 2

3

4

5 6

7 8 9 10 11 12

13 14 15

Avogaro P, Cazzolato G, Bittolo Bon G, Quinci GB. Are apolipoproteins better discriminators than lipids for atherosclerosis. Lancet 19?9;i:901-903. Sniderman A, Shapiro S, Marpole D, Skinner B, Teng B, Kwiterovich Jr PO. Association of coronary arteriosclerosis with hyperapobetalipoproteinemia (increased protein but normal cholesterol levels in human plasma low density lipoproteins), Proc Nat1 Acad Sci USA 1980;77:604-608. Wayne TF, Alaupovic P. Curry MD et al. Plasma apolipoprotein B and VLDL-, LDL-. and HDLcholesterol as risk factors in the development of coronary artery disease in male patients examined by angiography. Atherosclerosis 1981;39:41 I-424. Durrington PN, Hunt L, Ishola M, Kane J, Stephens WP. Serum apolipoproteins Al and B and lipoproteins in middle aged men with and without previous myocardial infarction. Br Heart J 1986;56:206-212. Schonfeld G, Patsch 8, Pfleger B, Witztum JL, Weidman SW. Lipolysis produces changes in the immunoreactivity and cell reactivity of very low density lipoproteins. J Clin Invest 1979~64: 1288- 1297. Aviram M, Lund-Katz S, Phillips MC, Chait A. The influence of triglyceride content of low density lipoprotein on the interaction of apolipoprotein B-100 with cells. J Biol Chem 1988;263: 16842-16848. Kinoshita M, Krul ES, Schonfeld G. Modification of the core lipids of low density lipoproteins produces selective alterations in the expression of apo B-100 epitopes. J Lipid Res 1990:31:701-708. Maciejko JJ, Mao STJ. Radioimmunoassay of apolipoprotein A-I. Application of a non-ionic detergent (Tween-20) and solid-phase staphylococcus. Clin Chem 1982;28:199-204. Levinson SS. Problems with measurement of apoiipoproteins AI and All. Ann Clin Lab Sci 1990;20:307-3 17. Walmsley WA, Grant S, George PM. Effect of plasma triglyceride concentrations on the accuracy of immunoturbidimetric assays of apolipoprotein B. Clin Chem 1991;37:748-753. Maciejko JJ, Levinson SS, Markyvech L, Smith M. New assay of apolipoproteins Al and B by rate nephelometry evaluated. Clin Chem 1987; l1:2065-2069. Adolphson JL, Albers JJ. Comparison of two commercial nephelometric methods for apoprotein Al and apoprotein B with standardized apoprotein Al and B radioimmunoassay. J Lipid Res 1989;30:59?-606. Glueck CJ, MaCray C, Speirs J. Measurement of serum apo Al and apo B: comparison of immunoturbidimetric and rate nephelometric techniques. Clin Chim Acta 1991;197: 123-132. Brustolin D, Maierna M, Aguzzi F et al. Immunoturbidimetric method for routine determinations of apolipoproteins A-I and B. Clin Chem 1991;37:742-747. Stampfer MJ, Sacks FM, Safvini WC, Willett WC, Hennekens CH. A prospective study of cholesterol, apolipoproteins, and risk of myocardial infarction, N Engl J Med 1991;325:373-381.

16 17 18 19 20 21 22

23

24 25 26

27

28

29 30 31 32 33 34

35 36

Levinson SS, Wagner S. Assay of apolipoprotein B containing lipoproteins for routine use. Arch Pathol Lab Med 1992;1161350-1161354. Maciejko JJ, Homes DR, Kottke BA, Zinsmeister AB, Dinh DM, Mao SJT. Apolipoprotein AI as a marker of angiographically assessed coronary artery disease. N Engi J Med 1983;309:385-389. Kottke BA, Zinsmeister AR, Holmes Jr DR et ai. Apolipoproteins and coronary artery disease. Mayo Clin Proc t986;61:313-320. Finley PR, Schifman RB, Williams RJ. Lichti DA. Cholesterol in high density lipoprotein: use of Mg’+i dextran sulfate in its enzymatic measurement. Clin Chem 1978;24:931-933. Warnick GR, Nguyen T, Albers JJ. Comparison of improved precipitation methods for quantitation of high-density lipoprotein cholesterol. Clin Chem 1985;31:217-222. Friedewaid WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol without the use of the preparative ultracentrifuge. Chin Chem 1972; 18:499-502. Zweig MH, Broste SK, Reinhart RA. ROC curve analysis: an example showing the relationships among serum lipid and apolipoprotein concentrations in identifying patients with coronary artery disease. Clin Chem 1992;38:1425-1428. Levinson SS, Wagner SC. The relative usefulness of automated apolipoprotein A-I and high density lipoprotein cholesterol assays as markers for coronary artery disease. Am J Clin Pathoi 1993; in press. Austin MA, Breslow JL, Hennekens CH et al. Low-density hopprotein subclass patterns and risk of myocardial infarction. J Am Med Assoc 1988;260:1917- 1921. La Belle M, Krauss RM. Differences in carbohydrate content of low density lipoproteins associated with low density lipoprotein subclass patterns. J Lipid Res 1990;31: 1577-1588. Marcovia SM, Albers JJ, Dati F, Ledue TB, Ritchie RF. International federation of clinical chemistry standardization project for measurements of apohpoproteins A-I and B. Clin Chem 1991;37:1676-1682. Cooper GR, Henderson LO, Smith SJ, Hannon WH. Clinical applications and standardization of apolipoprotein measurements in the diagnostic workup of lipid disorders. Clin Chem 1991;37:619-620. Albers JJ, Marcovina SM, Kennedy H. International Federation of Clinical Chemistry standardization project for measurement of apolipoproteins A-I and B. II. Evaluation and selection of candidate reference materials. Clin Chem 1992;38:658-662. Grundy SM, Vega GL. Two different views on the relationship of hy~rtrigIy~eridemia to coronary heart disease. Arch Intern Med 1992;152:28-34. Vergani C, Trovato G, Dioguardi N. Serum total lipids, lipoproteins cholesterol, apoproteins A and B in cardiovascular disease. Clin Chim Acta 197887: 127-l 33. Castelli WP. The triglyceride issue: a view from Framingham. Am Heart J 1986;112:432-438. Kwiterovich PO Jr. HyperapoB: a pleiotropic phenotype characterized by dense low-density lipoproteins and associated coronary artery disease. Clin Chem 1988;34:B71-77. Scanu AM, Fless GM. Lipoprotein(a). Heterogeneity and biological relevance. J Clin Invest 1990;85:1709-1715. Dahlen GH, Guyton JR, Attar M, Farmer JA, Kautz JA, Gotto Jr AM. Association of levels of lipoprotein Lp(a), plasma lipids and other lipoproteins with coronary artery disease documented by angiography. Circulation 1986;74:758-765. Rader DJ, Brewer HB. Lipoprotein(a). Clinical approach to a unique artherogenic lipoprotein. J Am Med Assoc 1992;267:1109-1112. Grinstead GF, Ellefson RD. Heterogeneity of lipoprotein Lp(a) and apolipoprotein(a). Clin Chem 1988;34:1036-1040.