Lipoprotein Quantification: An Electrophoretic Method Compared with the Lipid Research Clinics Method

Transcription

1 CLIN. CHEM. 2810, (1982) Lipoprotein Quantification: An Electrophoretic Method Compared with the Lipid Research Clinics Method G. Russell Warnick, Thuy Nguyen,1 Robert 0. Bergelin,2 Patricia W. Wahl,2 and John J. Albers We compared a turbidimetric electrophoretic method (Lipidophor) for lipoprotein quantification with the standardized Lipid Research Clinics (LRC) method. In the Lipidophor procedure, major lipoproteins are separated by electrophoresis on agarose gels, precipitated on the gels with phosphotungstate-mg2 reagent, and the resulting turbidity is measured densitometrically. Measurements of relative turbidity were converted into lipoprotein cholesterol values by the use of numeric constants provided by the manufacturer. Among-day CVs (n = 46) for the Lipidophor method were 6.0%, 3.6%, and 9.9% for cholesterol in the alpha-, beta-, and pre-beta lipoproteins, respectively. The Lipidophor alpha-cholesterol was significantly lower (n = 171 specimens) than the LRC highdensity lipoprotein (HDL) cholesterol (514 vs 586 mgl), and beta-cholesterol was significantly higher than the corresponding LRC tow-density lipoprotein (LDL) cholesterol values (1505 vs 1409 mgl). The linear relation between the two methods for lipoprotein cholesterol quantification is as follows: Lipidophor alpha = 0.77 LRC HDL + 63 mgl with correlation coefficient (r) of 0.87; Lipidophor beta = 0.95 LRC LDL mgl (r = 0.96); Lipidophor pre-beta = 0.57 LRC very-low-density lipoprotein + 39 mgl (r = 0.82). We derived a revised algorithm for estimating lipoprotein cholesterol from turbidity measurements. Lipoprotein cholesterol values by the Lipidophor method agree well with those obtained by the LRC method when these constants are used. Addftlonal Keyphrases: turbidimetry cholesterol#{149} alpha-. beta-, and pro-beta lipoproteins high-density, low-density, and very-low density lipoprotein cholesterol Recognition of the inverse relationship between HDL3 cholesterol values and the risk of coronary artery disease (reviewed in 1) has emphasized the importance of quantitating the major lipoprotein classes. The lipoproteins can be separated by sequential density ultracentrifugation, but this approach is often impractical for routine quantification. The LRC method (2) involves preparative ultracentrifugation for separation of VLDL and a chemical precipitation method for separation of HDL. The lipoproteins are quantified in terms of their cholesterol content. This procedure has been standardized and used extensively in major population studies (3). HDL, LDL, and VLDL correspond approximately to the Northwest Lipid Research Clinic, Harborview Medical Center, Seattle WA 98104, the Department of Medicine, and 2 the Departments of Biostatistics and Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA Nonstandard abbreviations: HDL, LDL, VLDL: high-, low-, and very-low-density lipoproteins; apob, apolipoprotein B, the major protein of LDL and VLDL; EDTA, disodium ethylenediaminetetraacetate; CH, cholesterol; CDC, Centers for Disease Control; and LRC, Lipid Research Clinics. Received Feb. 16, 1982; accepted June 9, alpha-, beta-, and pre-beta lipoprotein fractions, respectively, as separated by electrophoresis. Electrophoretic methods offer potential advantages for routine quantification of lipoproteins. The major lipoprotein classes are separated in a single step, and the patterns provide a visual representation of the lipoproteins. The basic equipment and expertise are generally present in the clinical laboratory. Although previous electrophoretic methods have been useful for qualitative analysis of lipoproteins, better reproducibility is desirable for quantitative applications. Lipoproteins have been quantitated by densitometry after staining with a lipophilic dye or with enzymic cholesterol reagents after their electrophoretic separation by agarose gels (4, 5), cellulose acetate strips (6), or polyacrylamide gels (7). A turbidimetric electrophoretic method (Lipidophor All in 12; Immuno Diagnostika GmbH, Vienna, Austria) involves lipoprotein separation on agarose gels followed by precipitation of the alpha-, beta-, and pre-beta bands with sodium phosphotungstate-mgcl2--nacl solution (8, 9). This is a modification of an earlier method involving precipitation of lipoproteins on gels with dextran sulfate and Mg2 (10, 11). The lipoproteins are quantified in terms of their turbidity after densitometric scanning. Relative turbidity measurements of the lipoproteins obtained by densitometry are converted to lipoprotein cholesterol values by using numeric constants that relate turbidity to lipoprotein cholesterol content. In this study, the reproducibility of the Lipidophor electrophoresis method was determined by repeated analysis of control materials. Quantification of lipoprotein cholesterol by the electrophoresis method was compared with results obtained by the LRC method for fresh plasma specimens. Revised constants for estimating lipoprotein cholesterol were derived from the observed relationship between the two methods. Materials and Methods Equipment and reagents for evaluation of the Lipidophor All in 12 kit were supplied by the manufacturer. The following reagents were included: agarose electrophoresis gel with two sample wells per gel, agarose solution, premixed buffer, developer solutions, and a lyophilized serum control material. Equipment included an electrophoresis chamber, power supply, thermostated heater for melting agarose, densitometer, and viewing chamber, Developer Solution 1 contained, per liter, 18 g of sodium phosphotungstate, 0.18 mol of MgCl2, and 0.7 mol of NaCI at ph 6.8. Developer Solution 2 was the same except the NaCI concentration was 1.6 moll and the ph was 6.6. Lipidophor Method Treat EDTA-anticoagulated plasma specimens with a few crystals of bovine thrombin (Parke, Davis and Co., Detroit, Ml 48232). Allow to stand for 10 mm at room temperature and remove the fibrin by low-speed centrifugation. (The Lipidophor protocol recommended analysis of serum because of in - terference by fibrinogen in plasma. Because this laboratory 2116 CLINICAL CHEMISTRY, Vol. 28, No. 10, 1982

2 routinely performs lipoprotein separations on EDTA plasma, we used thrombin to remove fibrinogen from EDTA plasma. In paired comparisons, results wer nearly identical for serum and for thrombin-treated EDTA plasma.) Add 100 tl of melted agarose solution (61-66 #{176}C) to 100 L of specimen in a disposable tube, with rapid, thorough mixing. Quickly, before the mixture cools, apply 10 L to the sample well in the agarose gel. Seal the sample in the well by covering with a drop of melted agarose solution. With each set of subject specimens, similarly analyze an aliquot of freshly reconstituted lyophilized serum control material (Lipidophor Control Serum, Lot 3). Subject up to six gels (12 specimens) to electrophoresis for 80 mm, as described in the package insert. After electrophoresis, fully immerse the gel trays for 1 h at ambient temperature in 30 ml of developer solution 1. Transfer the gel trays to 30 ml of Developer Solution 2 for at least 1 h (but not more than 15 h) at ambient temperature. Before densitometric scanning, rinse the gels with 0.15 moll NaC1 solution and thoroughly dry the underside of the plastic trays. By densitometry determine the relative turbidity (area per 100 arbitrary units of total area) associated with the alpha-, beta-, and pre-beta lipoprotein bands. Calculate the cholesterol content of each lipoprotein fraction from the relative turbidity measurements, using numeric constants for lipoprotein turbidity and composition (8, 9) provided in the package insert. Note: The constants 4.28,4.70, and 8.29 for alpha, beta, and pre-beta, respectively, relate turbidity to lipoprotein mass in mgdl, whereas the proportions 0.18, 0.45, and 0.15 represent the relative amount of cholesterol in the same lipoproteins. Multiplication of the turbidity constants with the cholesterol proportions gives factors 0.77, 2.12, and 1.24 for alpha-, beta-, and pre-beta lipoproteins, respectively. The relative cholesterol values are obtained by multiplying the area percent for each lipoprotein band by the appropriate factor. A correction based on the measured total cholesterol is made by multiplying each relative cholesterol value by the ratio of total cholesterol (quantified by independent assay) to the total sum of the relative cholesterol values. Example: Specimen with total cholesterol of 171 mgdl as quantified by LRC assay. Specimens EDTA plasma specimens were obtained from apparently healthy volunteers who had fasted overnight. After collection, the blood was cooled immediately to 4 #{176}C and cells were removed within 3 h by centrifugation. Lipoprotein separations were generally made on the day of collection but no later than the following day. Specimens that were definitely turbid (indicating hypertriglyceridemia) or that demonstrated a beta band on agarose electrophoresis of the isolated VLDL fraction (indicating possible Type 3 hyperlipoproteinemia) were excluded. The presence of sinking pre-beta lipoprotein was noted. This variant lipoprotein, associated with the Lp(a) antigen (12), which is quantified together with LDL by the LRC method, has pre-beta mobility on electrophoresis. Therefore, its presence might be expected to adversely affect the relationship between the LRC and Lipidophor values. However, because in fact the relationship between the methods was similar in specimens with and without sinking pre-beta lipoproteins, the 17 sinking pre-beta-positive subjects were not excluded. Of the 171 subjects remaining, 82 (48%) were men with average age of 46.7 and ranging from age 20 to 73 years; 89 were women ranging in age from 20 to 71 years with a mean age of 48.0 years. The distribution for total cholesterol was (mean ± SD, and range, in mgl) 2140 ± 382, and for triglycerides was 1113 ± 609, Results and Discussion The reproducibility of the Lipidophor Method in lipoprotein quantification is shown in Table 1. The within-run variation for alpha- and beta-lipoproteins was excellent, with CVs of 2.3% and 1.1%, respectively; the CV for pre-beta was 23.7%. For among-day variation, CVs of 6.0% (SD = 29.5 mgfl), 3.6%, and 9.9% were obtained for alpha-, beta-, and pre-beta lipoproteins, respectively. By CDC criteria an acceptable method for quantification of HDL cholesterol should give results averaging within 30 [Area % X factor = relative CH] x total CH lipoprotein total relative CH = CH, mgdl Alpha [58.3 X 0.77 = 44.9J X 171 = Beta [35.9 X 2.12 = 76.1] X 171 = Pre-beta f 3.9 X 1.24 = X 171 = 6 Total relative cholesterol = mgdl Lipid Research Clinic Method mgl of the true value and a standard deviation of less than Separate VLDL from plasma by ultracentrifugation and 30 mgl (13). Precision reported for commercial lipoprotein HDL by heparin-mn2 precipitation as described previously electrophoretic systems is shown in Table 2. An agarose (2). Quantify cholesterol in plasma and in the lipoprotein method with a lipophilic dye required the use of an internal fractions by the LRC Liebermann-Burchard reagent method standard and computerized pattern analysis (4) to attain the for the AutoAnalyzer 11(2). Standards and control materials necessary precision. An agarose method with enzymic chowere prepared by the Clinical Chemistry Standardization lesterol reagents (5) met the precision criterion in the case of Section of the CDC, Atlanta, GA pools with above-average HDL values and was marginal for Table 1. Reproducibility in Lipoprotein Quantification by the Lipidophor Method Alpha Beta Pre-beta mgl a cv, % mgl a CV, % mgi a cv, % Within-run (n = 12) Fresh serum 700 ± ± ± Among-dayt Lipidophor pool ± ± ± a Mean ± b Forty-six runs over a six-month period. CLINICAL CHEMISTRY, Vol. 28, No. 10,

3 Agarose Method gel Fat Red 7B5 Enzymic cholesterol b Enzymic Enzymic Cellulose cholesterol cholesterol acetate (No.tlme) (241 yr) (723 days) (99 days) (not given) Enzymic cholesterol (726 plates) Enzymic cholesterol (not given) Polyacrylamide Table 2. Precision in Lipoprotein Cholesterol Quantification Reported for Other Commercial Electrophoretic Systems gel Alpha Mean SD, CV, Mean, mgl mgi % mgi Beta SD, CV, mgi % Reference Sudan Black (20) #{149} Quantification of total lipoprotein mass. b Data reported for beta-ilpoprotein represents beta plus pre-beta lipoproteins. pools with low values. Results for the cellulose acetate (6) and polyacrylamide (7) methods did not meet the precision criterion. By comparison, the reproducibility of the Lipidophor method (SD, 29.5 mgl) was acceptable for quantifying alpha-cholesterol. Furthermore, if one considers only the initial nine runs, the SD for alpha-cholesterol quantified by the Lipidophor method was 18.9 mgl (CV 4.1%) compared with SDs of 22.4 to 47.7 mgfl in nine runs (Table 2) for the method with enzymic cholesterol reagent and agarose electrophoresis. The LRC Liebermann-Burchard cholesterol method demonstrated among-day CVs of about 0.5% or less for pools with cholesterol values in the 1800 to 3000 mgl range and 1.25% for a pool with a cholesterol value of 517 mgl. In quantifying HDL cholesterol a pool that was assayed along with specimens through both the precipitation and analysis steps had an SD of 15,2 mgl (CV 3.6%) with a mean value of 423 mgl. Cholesterol values on control pools averaged within 15 mgl of CDC target values (16). Analytical recovery of cholesterol in the lipoprotein fractions from the ultracentrifugation step averaged 99.2%. In general, results for alpha-cholesterol by the Lipidophor method were lower than those for HDL cholesterol by the LRC method (Figure la). The relationship between the two methods estimated by linear regression was: Lipidophor alpha = LRC HDL mgfl, correlation coefficient (r) In these specimens, the mean alpha-cholesterol was 514 mgl compared with 586 mgl for HDL cholesterol with a paired SD of 75.3 mgl (difference significant at p <0.001), by Student s paired t-test) (Table 3). It is clear from the regression statistics and the data in Figure la, that, although the methods were reasonably well correlated, differences were observed, especially at high concentrations of HDL. The Lipidophor calculation factors could account for these differences. The Lipidophor constants that relate relative turbidity measurements to lipoprotein mass are a function of the average particle size within a lipoprotein class; the larger the particle, the larger the constant. Similarly, the cholesterol compositions represent an average for each class, Heterogeneity in particle size or composition within a lipoprotein class is not considered. Because specimens with increased HDL cholesterol generally have relatively more of the HDL2 sub- 0 HDL b LDL i leoo LOC HDL rogl) LRC LOL (ogl).rc VLOL 000IL) Fig. 1. Linear regression statistics for alpha-, beta-, and pre-beta cholesterol, by the Lipidophor method (y) vs results for HDL, LDL, and VLDL cholesterol by the LRC method (x), respectively The dptted lines represent y equals x, or the line of perfect agreement; the solid lines indicate the estimated relationships by linear regression between the two methods for each lipoprotein class. The statistics for the relationships are given in the text and Table CLINICAL CHEMISTRY, Vol. 28, No. 10, 1982

4 Table 3. Agreement between LRC (x) and Lipidophor (y) Lipoprotein Cholesterol Values by Two Calculation Algorithms a Difference statisticallysignificant by Student s paired f-test (p <0.001). Mean Mean Pairod r Slope y-lntercept x y Difference SD I Lipidophor HOL LDL VLDL Revised HOL LDL VLDL factors factors class, which is both larger and richer in cholesterol than is the HDL2 subclass, underestimation of total alpha-cholesterol would be likely. These conclusions are supported by the fact that specimens from men only, which could be expected to have less HDL2, showed better agreement for HDL cholesterol, as shown by- the following relationship estimated by linear regression: Lipidophor alpha = LRC HDL (r = 0.817). In those subjects with LRC HDL cholesterol below the median value of 574 mgl, the agreement was even better: Lipidophor alpha = LRC HDL (r = 0.700). Of course the observed difference might be due in part to the fact that the lipoproteins with alpha mobility are not exactly identical to the heparin-mn2-soluble lipoproteins. Also, the heparin_mn2+ procedure may slightly overestimate HDL because of incomplete precipitation of the apo B-associated lipoproteins (17, 18). Results for LDL and beta-lipoproteins were highly correlated and in good agreement (r = 0.959) (Figure lb). The relationship estimated by linear regression was as follows: Lipidophor beta = LRC LDL mgl. LDL cholesterol by the LRC method averaged 1409 mgl in this group, compared with 1505 mgl for beta-cholesterol by the Lipidophor method. The paired SD was 96.9 mgl and the difference was significant at p < These results are consistent with a relatively constant overestimation by the Lipidophor method across the LDL range. Figure ic illustrates the relationship between the two methods in quantification of VLDL cholesterol. The correlation coefficient was and the relationship estimated by linear regression was as follows: Lipidophor VLDL = LRC VLDL mgl. The mean pre-beta cholesterol by Lipidophor was 122 mgl vs 147 mgl for LRC VLDL cholesterol. The Lipidophor pre-beta cholesterol values were consistently underestimated in specimens with moderately increased VLDL cholesterol. This could be a result of trailing or reduced mobility of the larger pre- beta particles on agarose, enhancing the beta at the expense of the pre-beta region, a potential problem that is acknowledged in the Lipidophor protocol. In addition, the turbidity constant for pre-beta might not be appropriate for large VLDL particles. For some applications, these apparent systematic differences in quantifying lipoprotein cholesterol may not be important. In cross-sectional studies (19, 20), the ratio by the Lipidophor method of beta- to-alpha-cholesterol was reportedly a better indicator of coronary disease risk than were absolute concentrations of the lipoproteins. In fact, for some screening applications the ratio of beta to alpha by relative turbidity measurements might be useful without estimation of the lipoprotein cholesterol content. However, for other applications it may be important to obtain lipoprotein cholesterol values that are comparable with those obtained by the established quantification methods. Therefore, on the basis of the observed relationship with the LRC method in the 171 specimens, we derived the following augmented constants for estimating Lipidophor lipoprotein cholesterol values: alpha-cholesterol = [(area % X 0.77) beta-cholesterol (area % X 0.77)2] total C Htotal relative CH = [area % X 2.12] total CHtotal relative CH pre-beta cholesterol = [(area % X 1.24) (area % X 1.24)2] total CHtotal relative CH When we used these modified algorithms to estimate lipoprotein cholesterol values from Lipidophor relative turbidity measurements, the relationships estimated by linear regression for the three lipoproteins were improved as shown in Table 3. For HDL the slope became with a y-intercept of 41.4, compared with and 63.4 obtained by using the manufacturer s constants. The difference between the two methods decreased from 72 mgl to 3 mgl, no longer statistically significant. For LDL the slope did not change appreciably but they-intercept decreased from to 86.6 and the average difference between the methods was no longer statistically significant. For VLDL the slope increased from to 0.750, they-intercept decreased from 38.8 to 28.8, and the average difference decreased to 8 mgl, which was not statistically significant. In summary, the turbidimetric electrophoretic method demonstrated good precision in lipoprotein quantification. The method was in reasonable agreement with the LRC method for lipoprotein cholesterol, given the numeric constants provided by the manufacturer. Use of the revised equations reported here improved the agreement, however. Estimation of the beta-to-alpha lipoprotein ratio either in terms of relative turbidity or calculated cholesterol might prove to be a useful predictor of coronary artery disease risk. We thank the laboratory staff of the Northwest Lipid Research Clinic for technical support, and Mary Miller and Siobhan Thomas for assistance in preparation of the manuscript. This work was supported in part by Contract NO1-HV-12157L from the Lipid Metabolism Branch of the National Heart, Lung and Blood Institute and by Immuno Diagnostika GmbH, Vienna, Austria. J.J.A. is an established investigator of the American Heart Association. References 1. Heiss, G., Johnson, N., Reiland, S., et al., The epidemiology of CLINICAL CHEMISTRY, Vol. 28, No. 10,

5 plasma high-density lipoprotein cholesterol levels. Circulation 62, (1980). 2. Manual of Laboratory Operations, Lipid Research Clinics Program, Vol. 1, Lipid and Lipoprotein Analysis, DHEW Publication No. (NIH) , Bethesda, MD, (Obtainable from the U.S. Govt. Printing Office, Washington, DC). 3. Heiss, G., Tamir, I., Davis, C. E., et al., Lipoprotein-cholesterol distributions in selected North American populations. Circulation 61, (1980). 4. Lindgren, F. T., Silvers, J., Jutagir, R., et al., A comparison of simplified methods for lipoprotein quantitation using the analytic ultracentrifuge as a standard. Lipids 12, 278 (1977). 5. Conlon, D., Blankstein, L. A., Pasakarnis, P. A., et al., Quantitative determination of high-density lipoprotein cholesterol by agarose gel electrophoresis. Clin. Chem. 25, (1979). 6. Cobb, S., and Sanders, J. L., Enzymic determination of cholesterol in serum lipoproteins separated by electrophoresis. Clin. Chem. 24, (1978). 7. Muniz, N., Measurement of plasma lipoproteins by electrophoresis on polyacrylamide gel. Clin. Chem. 23, (1977). 8. Wieland, H., and Seidel, D., Advances in the analysis of plasma lipoproteins. Inn. Med. 5, (1978). 9. Seidel, D., Electrophoresis of plasma lipoproteins. Chap. 6 in Report of the High Density Lipoprotein Methodology Workshop, K. Lippel, Ed., NIH Publication No , Bethesda, MD, 1979, pp U.S. Govt. Printing Office, Washington, DC, No Seidel, D., Wieland, H., and Ruppert, C., Improved techniques for assessment of plasma lipoprotein patterns. I. Precipitation in gels after electrophoresis with polyanionic compounds. Clin. Chem. 19, (1973). 11. Neubeck, W., Wieland, H., Habenicht, A., et al., Improved assessment of plasma lipoprotein patterns. III. Direct measurement of lipoproteins after gel-electrophoresis. Clin. Chem. 23, (1977). 12. Albers, J. J., Cabana, V. G., Warnick, G. R., and Hazzard, W. R., Lp(a) lipoprotein: Relationship to sinking pre-(3 lipoprotein, hyperlipoproteinemia, and apolipoprotein B. Metabolism 24, (1975). 13. Hainline, A., Jr., Cooper, G. R., Olansky, A. S., et al., C.D.C. survey of HDL cholesterol measurement-a report. Centers for Disease Control, Atlanta, GA, Stein, E. A., McNeely, S., and Steiner, P., Electrophoretic separation of high density lipoprotein cholesterol evaluated and compared with the modified Lipid Research Clinic procedure. Clin. Chem. 25, (1979). 15. Baillie, E. E., Welborn, P., Hill,J., and Sexton, B., Cholesterol in high-density lipoproteins quantified by agarose gel electrophoresis, cellulose acetate electrophoresis, and magnesium-sodium phosphotungstate precipitation. Clin. Chem. 26, (1980). 16. Eavenson, E., Grier, 0. T., Cisson, J. G., and Witter, R. F., A semiautomated procedure for the determination of serum cholesterol using the Abell-Kendall method. J. Am. Oil Chem. Soc. 43, (1966). 17. Warnick, G. R., and Albers, J. J., A comprehensive evaluation of the heparin-manganese precipitation procedure for estimating high density lipoprotein cholesterol. J. Lipid Res. 19, (1978). 18. Albers, J. J., Warnick, G. R., Weibe, D., et al., Multi-laboratory comparison of three heparin_mn2+ precipitation procedures for estimating cholesterol in high-density lipoprotein. Clin. Chem. 24, (1978). 19. Zimmer, F., Riebeling, V., Benke, B., et al., The LDL-HDL ratio in patients with coronary arteriosclerosis. Z. Kardiol. 69, (1980). 20. Wieland, H., Seidel, D., Wiegand, V., and Kreuzer, H., Serum Iipoproteins and coronary artery disease (CAD). Comparison of the lipoprotein profile and the results of coronary angiography. Atherosclerosis 36, (1980). I 2120 CLINICAL CHEMISTRY, Vol. 28, No. 10, 1982