You are on page 1of 3




Measurement of Plasma LDL Cholesterol in Patients With Diabetes


OBJECTIVE To assess the accuracy of plasma LDL cholesterol concentrations estimated by the Friedewald formula and a direct immunoseparation method by comparison with a reference ultracentrifugation procedure in patients with diabetes. RESEARCH DESIGN A N D M E T H O D S Fasting plasma samples with triglyceride concentrations <4.5 mmol/1 were collected from 100 patients with diabetes (28 type 1 and 72 type 11) and LDL cholesterol concentrations were compared by the three methods. RESULTS LDL cholesterol values determined by the reference (3-quantitation procedure were highly correlated with both the Friedewald formula (r = 0.96) and a direct immunoseparation method (r = 0.92). Calculated (Friedewald) LDL cholesterol coincided with the reference method with <10% error in 74% of the total diabetic group (82% of type I and 68% of type 11 diabetic patients). However, agreement between the direct LDL cholesterol and reference methods was significantly less (P = 0.02), with only 44% of patients having an error of <10% (52% of type I and 41% of type II diabetic patients). The direct immunoseparation method for LDL cholesterol showed a positive bias with increasing triglyceride concentrations, particularly for patients with type II diabetes. CONCLUSIONS In the group of diabetic patients studied with plasma triglyceride concentrations <4.5 mmol/1, the Friedewald formula provided an accurate estimation of LDL cholesterol. The direct immunoseparation method significantly overestimated LDL cholesterol at triglyceride levels between 2 and 4.5 mmol/1.

in the management of patients with diabetes (5). It has long been recognized that new methods for the direct determination of LDL cholesterol should be developed (2), and a rapid immunoseparation assay for LDL has recently been evaluated (6). The aims of the present study were to re-examine the accuracy of the Friedewald formula in measuring plasma LDL cholesterol in diabetic patients and to investigate the accuracy of the new immunoseparation assay for direct LDL cholesterol determination in diabetic patients.

RESEARCH DESIGN A N D METHODS Fasting plasma samples

with triglyceride concentrations < 4 . 5 mmol/1 were collected using EDTA as anticoagulant from 100 patients with diabetes (49 men, 51 women; age range 22-83 years). Twenty-eight patients (mean age 48 years) had type I diabetes, while 72 patients (mean age 63 years) were classified as having type II diabetes. Of the type II patients, 28% were receiving treatment with diet alone, 33% with oral hypoglycemic agents, and 39% with insulin. As a measure of diabetic control, HbA lc was determined in all patients. Plasma total cholesterol and triglyceride were measured enzymatically on a Hitachi 717 analyzer (Boehringer Mannheim, Germany). HDL cholesterol was measured following precipitation of VLDL and LDL from plasma with phosphotungstic acid/magnesium chloride (7). Only samples with triglyceride concentrations <4.5 mmol/1 were included in the study to allow calculation of LDL cholesterol using the Friedewald formula (LDL cholesterol = total cholesterol HDL cholesterol triglyceride/2.2) (8). LDL cholesterol was also determined in unfrozen plasma using a recently developed method (D-LDL) involving immunoseparation of LDL particles from chylomicron, and VLDL and HDL particles (6). In this method, a small volume of serum or plasma (30 pi) is mixed with a latex bead suspension containing immobilized antibodies to apolipoproteins Al and E (Direct LDL, Sigma, St. Louis, MO). LDL particles do not bind to the beads and are recovered after filtration for cholesterol estimation.

iabetes is an important risk factor for coronary artery disease, and the Consensus Statement of the American Diabetes Association has set the desirable plasma LDL cholesterol concentration in adult diabetic patients at < 3 . 4 mmol/1 (1), the same value as proposed for patients with existing coronary disease. The Working Group on Lipoprotein Measurement for the National Cholesterol Education Program (NCEP) has recently published its recommendations for reliable LDL cholesterol determination (2). They have pointed out that the reference method for LDL cholesterol measurement should be combined ultracentrifugation-

polyanion precipitation, followed by cholesterol assay of fractions with methods that satisfy the accuracy and precision criteria of the NCEP Laboratory Standardization Panel. The commonly applied Friedewald equation should not be used as a reference method for LDL cholesterol measurement, and even as a routine method, the equation has well-described limitations. Some of these limitations are relevant to patients with diabetes, who often have hypertriglyceridemia or abnormal VLDL or LDL composition (3,4), which perturb the Friedewald equation. Indeed, it has been claimed that calculated LDL cholesterol values should not be used

From the Department of Biochemistry and Chemical Pathology, Flinders Medical Centre, Bedford Park, South Australia. Address correspondence and reprint requests to M.J. Whiting, PhD, Dept. Biochemistry and Chemical Pathology, Flinders Medical Centre, Bedford Park, South Australia 5042. E-mail: malcolm.whiting Received for publication 25 April 1996 and accepted in revised form 22 August 1996. B-LDL, LDL determined by fj-quantitation; D-LDL, LDL determined by direct immunoseparation; FLDL, LDL determined by the Friedewald formula; IDL, intermediate-density lipoprotein; NCEP, National Cholesterol Education Program.




Whiting, Shephard, and Tallis

As the reference procedure, LDL cholesterol was determined in all samples using fj-quantitation (B-LDL) (2). Plasma (1.0 ml) was overlaid with NaCl solution of density 1.006 g/ml and centrifuged at 513,000g for 2.5 h at 20C in a Beckman Optima TLX tabletop ultracentrifuge, using a TLA 120.2 rotor. After slicing the centrifuge tube to separate VLDL (upper layer) from LDL and HDL (lower layer), LDL were precipitated with phosphotungstate and magnesium chloride and the remaining HDL cholesterol determined on the supernatant. The mean recovery of HDL cholesterol by this technique was 90 8% for the 100 plasma samples. LDL cholesterol was calculated by subtracting the HDL cholesterol concentration from the total cholesterol recovered in the lower fraction. Data were analyzed using a Statview version 4.5 statistics package, with differences between group means tested by unpaired Student's t test. Simple linear regression was carried out with B-LDL as the independent variable, and correlation coefficients between continuous variables were calculated by Pearson's correlation test. RESULTS LDL cholesterol concentrations calculated by the Friedewald formula were highly correlated with values obtained by fJ-quantitation (r = 0.96), as were LDL cholesterol values determined by the direct immunoseparation method (r = 0.92). To investigate the accuracy of the Friedewald formula and the immunoseparation method relative to (3-quantitation, the ratios F-LDL/B-LDL and D-LDL/BLDL were calculated and expressed as a percentage. The Friedewald LDL cholesterol coincided with B-LDL with <5% error (ratio between 95 and 105) in 41 samples and with <10% error (ratio between 90 and 110) in 74 out of 100 samples. Agreement was better in plasma samples from type I compared with type II diabetic subjects (82 and 68% of samples with <10% error, respectively), but this difference was not statistically significant. Agreement between the D-LDL method and B-LDL was not as good, with only 20 samples with <5% error and 44 out of 100 samples with <10% error. Again, agreement was better in plasma samples from type I compared with type II diabetic subjects (52 and 41% of samples with <10% error, respectively).



TRIG (mmol/L)

TRIG (mmol/L)

Figure 1Regression plots of the plasma triglyceride concentration in mmol/l (x-axis) against the ratio of LDL cholesterol concentrations determined by different methods (y-axis), according to diabetes type. A: ratio of LDL cholesterol determined by the Friedewald formula and fi-quantitation (l:LDL/B-LDL); B: ratio of LDL cholesterol determined by direct immunoseparation and $-quantitation (D-LDUB-LDL). The overall correlation coefficients were 0.23 and 0.69 for A and B, respectively.

Of the 100 patients, 31 were assessed as having good long-term glycemic control (HbAlc between 6.1 and 8.0%), 37 had moderate glycemic control (HbAlc between 8.1 and 10.0%), and 25 had poor control (HbAk >10.0%). There was no association between the accuracy of F-LDL and D-LDL relative to B-LDL and the degree of diabetic control. However, when samples were divided into low triglyceride (<2.0 mmol/l) and high triglyceride (2.1-4.5 mmol/l) groups, a significant difference was observed between the percent ratios of F-LDL/B-LDL (100.9 8.2 vs. 105.4 12.0; P = 0.03) and D-LDL/B-LDL (101.4 12.5 vs. 119.1 17.6; P < 0.001) for low and high triglyceride groups, respectively. This difference was further investigated by examining regression plots of the ratios against plasma triglyceride concentrations. There was a weak association (P = 0.02) between the ratio F-LDL/B-LDL and plasma triglyceride, with a correlation coefficient of 0.23. As shown in Fig. 1A, the correlation coefficient was higher for plasma samples from type II diabetic subjects compared with type I diabetic subjects. However, the correlation between the ratio D-LDL/B-LDL and triglyceride was much stronger, with the correlation coefficient equal to 0.69 (P < 0.0001). Again, the correlation coefficient was higher for type II diabetic patients than for type I (Fig. IB). Thus, as the triglyceride concentration in plasma increased toward 4.5 mmol/l, the direct immunoseparation method overestimated LDL cholesterol relative to fj-quantitation, particularly in

type II diabetic patients. In contrast, the Friedewald formula remained reasonably accurate, even in samples with triglyceride concentrations up to 4.5 mmol/l. C O N C L U S I O N S The determination of risk status for coronary artery disease depends on plasma LDL cholesterol levels and other risk factors, such as diabetes (1). Even though the reference method for LDL cholesterol measurement has been confirmed as combined ultracentrifugation-polyanion precipitation (2), for routine patient evaluation, LDL cholesterol is estimated indirectly by using the Friedewald formula (8) and measurements of total cholesterol, triglyceride, and HDL cholesterol. The Friedewald formula assumes that a constant cholesterol/triglyceride ratio is present in VLDL particles, but this assumption is known to be invalid as plasma triglyceride levels increase (9) and if type III hyperlipoproteinemia (8) or raised intermediate-density lipoprotein (IDL) (10) is present. To identify patients who require treatment for hypercholesterolemia, the Friedewald equation is generally regarded as adequate. In a large study of 4,736 specimens, >90% of estimated LDL cholesterol values were within 10% of values measured by the reference method when the triglyceride concentrations were <2.3 mmol/l (9). Another study by McNamara et al. (11) analyzed 4,797 plasma samples, which were collected mainly during the third cycle of the Framingham Heart Study. They found that for samples with triglycerides <2.3 mmol/l,


LDL cholesterol in diabetes

the Friedewald LDL cholesterol concentrations agreed with the fi-quantitation value (10%) in 86% of subjects (11). Because of possible abnormalities in plasma lipoprotein lipid composition in diabetic patients, the validity of the Friedewald formula has been examined in groups of type I and type II patients. Winocour et al. (12) found from a study of 48 type I diabetic subjects that LDL cholesterol estimated by the Friedewald formula correlated closely with results based on ultracentrifugation and was valid for use in epidemiological studies of diabetes. Rubies-Prat et al. (5) suggested that calculated LDL cholesterol should not be used for the management of lipoprotein abnormalities in patients with type II diabetes. In their study, the Friedewald formula overestimated by >10% the actual LDL cholesterol concentration in 39% of diabetic patients and underestimated the true value in 13% of patients, i.e., the formula was only 48% accurate, compared with a result of 68% accuracy for type II diabetic subjects in the present study. Our results are not significantly different from the results of McNamara et al. (11) for the Framingham population when values are stratified by triglyceride level. In our opinion, the Friedewald formula is adequate for the measurement of LDL cholesterol in diabetes of both type I and II. To allow the direct measurement of LDL cholesterol, a new immunoseparation method has been developed recently and evaluated by several laboratories (6,13,14). In comparing this assay to the reference (3quantitation procedure, McNamara et al. (6) found that the direct LDL assay correctly classified LDL cholesterol concentrations according to NCEP guidelines on 85% of samples tested. Jialal et al. (13) reported a good correlation between the two methods, with a correlation coefficient of 0.88 for 249 serum samples with a range of triglyceride concentrations from 0.15 to 25.2 mmol/1. Nevertheless, in one of these studies (6), a positive bias for direct LDL was noted in moderately hypertriglyceridemic samples. This finding agrees with the results reported in the present study for

diabetic patients. The reason for this bias is uncertain, although it has been suggested that some triglyceride-rich lipoprotein particles may not bind to the latex beads during the assay, thus falsely elevating the measured D-LDL cholesterol concentration (6). For normotriglyceridemic samples, there is excellent agreement between the direct LDL cholesterol assay and the Friedewald formula (6,14). Because plasma triglycerides are often mildly elevated in diabetic patients, the direct LDL cholesterol assay may not perform as well as the Friedewald formula in estimating the true LDL cholesterol concentration in plasma from diabetic patients. Further studies are required to elucidate the source of positive bias with the direct immunoseparation assay for LDL cholesterol.




9. Acknowledgments We would like to thank the Flinders Medical Centre Research Foundation for financial assistance.

References 1. American Diabetes Association: Detection and management of lipid disorders in diabetes (Consensus Statement). Diabetes Care 16:828-834, 1993 2. Bachorik PS, Ross JW for the NCEP Working Group on Lipoprotein Measurement: National cholesterol education program recommendations for measurement of low-density lipoprotein cholesterol: executive summary. Clin Chem 41:1414-1420, 1995 3. Kasama T, Yoshino G, Iwatani I, Iwai M, Hatanaka H, Kazumi T, Oimomi M, Baba S: Increased cholesterol concentration in intermediate density lipoprotein fraction of normolipidaemic non-insulin-dependent diabetics. Atherosclerosis 63:263-266, 1987 4. Joven J, Vilella E, Costa B, Turner PR, Richart C, Masana L: Concentrations of lipids and apoproteins in patients with clinically well-controlled insulin-dependent and non-insulin-dependent diabetes. Clin Chem 35:813-816, 1989 5. Rubies-Prat J, Reverter JL, Senti M, PedroBotet J, Salinas I, Lucas A, Nogues X, Sanmarti A: Calculated low-density lipoprotein cholesterol should not be used for






management of lipoprotein abnormalities in patients with diabetes mellitus. Diabetes Care 16:1081-1086,1993 McNamara JR, Cole TG, Contois JH, Ferguson CA, Ordovas JM, Schaefer EJ: Immunoseparation method for measuring low-density lipoprotein cholesterol directly from serum evaluated. Clin Chem 41:232240,1995 Calvert GD, Yeates RA, Mannik T: A comparison of the phosphotungstate magnesium precipitation and heparin-manganese precipitation procedures for estimating high-density lipoprotein cholesterol. Ann Clin Biochem 17:105-108, 1980 Friedewald WT, Levy RI, Fredrickson DS: Estimation of the concentration of lowdensity lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 18:499-502, 1972 Warnick GR, Knopp RH, Fitzpatrick Y Branson L: Estimating low-density lipoprotein cholesterol by the Friedewald equation is adequate for classifying patients on the basis of nationally recommended cutpoints. Clin Chem 36:15-19, 1990 Senti M, Pedro-Botet J, Nogues X, RubiesPrat J: Influence of intermediate-density lipoproteins on the accuracy of the Friedewald formula. Clin Chem 37:1394-1397, 1991 McNamara JR, Cohn JS, Wilson PWF, Schaefer EJ: Calculated values for lowdensity lipoprotein cholesterol in the assessment of lipid abnormalities and coronary disease risk. Clin Chem 36:36-42, 1990 Winocour PH, Ishola M, Durrington PN: Validation of the Friedewald formula for the measurement of low density lipoprotein cholesterol in insulin-dependent diabetes mellitus. Clin ChimActa 179:79-84,1989 Jialal I, Hirany SV, Devaraj S, Sherwood TA: Comparison of an immunoprecipitation method for direct measurement of LDL-cholesterol with beta-quantification (ultracentrifugation). Am J Clin Pathol 104:76-81, 1995 Cobbaert C, Broodman I, Swart GR, Hoogerbrugge N: Performance of a direct, immunoseparation based LDL-cholesterol method compared to Friedewald calculation and a polyvinyl sulphate precipitation method. Ear J Clin Biochem 33:417-424, 1995