Evaluation of a Rapid Homogeneous Method for Direct Measurement of High-Density Lipoprotein Cholesterol

Transcription

1 Clinical Chemistry / ORIGINAL ARTICLE Evaluation of a Rapid Homogeneous Method for Direct Measurement of High-Density Lipoprotein Cholesterol R. Scott Hubbard, MD,1 Shaina V. Hirany, MS,2 Sridevi Devaraj, PhD,1 Lani Martin, MT(ASCP),2 Joseph Parupia, MT(ASCP),3 andhhwarlaljialal, MD, PhD, FRCPath1'2'4 Abstract We evaluated the performance of a direct Liquid Ngeneous HDL-C assay (N-HDL; Genzyme Diagnostics, Cambridge, Mass) and compared it with a Centers for Disease Control and Prevention (CDC) modified reference procedure (M-REF) and phosphotungstic acid (PTA) precipitation method in patients with normotriglyceridemia (triglyceride level, <400 mg/dl) and hypertriglyceridemia (triglyceride level, >400 mg/dl). Excellent intra-assay and interassay coefficients of variation were obtained (<2.0%) using the N-HDL assay. The N-HDL and PTA assays correlated well with M-REF in normotriglyceridemic samples. In hypertriglyceridemic samples, however, the N-HDL method exhibited better correlation with MREF than the PTA assay. In addition, compared with M-REF, the mean absolute percentage bias of N-HDL was lower than the PTA assay in normotriglyceridemic (4.9% vs 5.8%) and hypertriglyceridemic (5.4% vs 12.9%) samples. Hemolysis, ascorbic acid, and bilirubin did not interfere with the N-HDL assay. On the basis of these findings, the N-HDL assay compares favorably with the modified CDC reference method and seems superior to the PTA assay. It also has the advantage of being suited for complete automation and, thus, would prove useful in large clinical laboratories. Epidemiologic and clinical studies have shown that high-density lipoprotein cholesterol (HDL-C) is an independent inverse risk factor for coronary artery disease.,j* A low HDL-C (<35 mg/dl) is believed to increase a person's risk of coronary artery disease, while a high HDL-C (>60 mg/dl) is considered protective.3-4 In addition, the ratio of total cholesterol to HDL-C is thought to be a better indicator for cardiovascular risk than total cholesterol alone.5 Based on these findings, the National Cholesterol Education Program Adult Treatment Panel II (NCEP ATP II) has revised its guidelines for the diagnosis and treatment of hypercholesterolemia in adults to include HDL-C measurements at the initial screening stage along with total cholesterol.3-4 It is clear then that because of the importance of HDL-C, the demand for HDL-C determination will increase in clinical laboratories accordingly. Therefore, there is a great need for rapid and accurate methods to allow for routine clinical testing of HDL-C.4 The NCEP has recommended that the Centers for Disease Control and Prevention (CDC) reference method be used as the basis of assessing the accuracy of routine HDLC assays.4 This reference method uses ultracentrifugation and heparin/mncl-, precipitation to remove very low-density lipoprotein (VLDL) and apo B-containing lipoproteins, followed by quantification of cholesterol in the supernatant using the Abell-Kendall assay.2 Besides the reference method, other methods available for HDL-C measurement include high-pressure liquid chromatography, sequential ultracentrifugation, electrophoresis, and precipitation methods.2'6-7 Currently, most clinical chemistry laboratories use precipitation methods because techniques such as ultracentrifugation and high-pressure liquid chromatography are laborious and require expensive instrumentation. Precipitation Am J Clin Pathol 1998;110: Key Words: High-density lipoprotein cholesterol; Coronary artery disease

2 Hubbard et al / EVALUATION OF A HOMOGENEOUS HDL-C ASSAY The Liquid N-geneous HDL-C assay (Genzyme Diagnostics, Cambridge, Mass) is a new homogeneous procedure that directly measures HDL-C on an automated clinical chemistry analyzer using 2 ready-to-use liquid reagents. Reagent 1 contains a chromogen and a polyanion that aggregates LDL, VLDL, and chylomicrons forming stable lipoprotein-polyanion complexes. Reagent 2 contains standard cholesterol enzymes and a unique detergent that not only selectively solubilizes HDL-C particles, but also adsorbs to the surface of LDL, VLDL, and chylomicrons, thereby inhibiting their reaction with the cholesterol enzymes. The HDL-C concentration is then determined enzymatically using standard Trinder chemistry. This method is rapid and well suited for automation because sample volume is small (3 ux) and all centrifugation and precipitation steps are eliminated. In the present study, we evaluated the analytical performance of the Liquid N-geneous HDL-C assay for HDL-C determination in normotriglyceridemic and hypertriglyceridemic samples obtained from a large clinical laboratory (Parkland Memorial Hospital, Dallas, Tex). We compared the HDL-C concentrations using this procedure with a modified CDC reference procedure (ultracentrifugation/magnetically enhanced dextran sulfate-mgcl2 method) and with the precipitation-based protocol currently used at our center (phosphotungstic acid [PTA]). Materials and Methods Reagents Liquid N-geneous HDL assay (N-HDL) kits were donated by Genzyme Diagnostics. Paramax cholesterol and triglyceride reagent tablets, as well as HDL Cholesterol Precipitating reagent, were from Dade International (Miami, Fla), Magnetic HDL Cholesterol reagent was 496 Am J Clin Pathol 1998;110: from Polymedco (Cortlandt Manor, NY), and ascorbic acid was from Sigma Diagnostics (St Louis, Mo). Patients Specimens were collected in accordance with the ethical standards established at Parkland Memorial Hospital and the University of Texas Southwestern Medical Center, Dallas. Comparison studies for HDL-C measurement were performed on 181 fasting serum samples received at our clinical chemistry laboratory at Parkland Memorial Hospital. Three methods were used to determine HDL-C: the N-HDL assay, the PTA method, and a modified ultracentrifugation/magnetic HDL reference method (M-REF). Samples were stored at 4 C and analyzed by all 3 assays within 3 days. Of the 181 patient serum samples, 131 had TG levels less than 400 mg/dl (range, mg/dl), and the remaining 50 had TG levels more than 400 mg/dl (range, 402-1,665 mg/dl). The overall cholesterol levels ranged from 86 to 577 mg/dl, and HDL cholesterol levels ranged from mg/dl. To determine how the postprandial state affects HDL-C concentrations measured by the N-HDL and PTA assays, paired fasting and nonfasting samples from 32 subjects were analyzed for HDL-C by the 2 methods. (To calculate Systeme International [SI] units for triglycerides in mmol/l, multiply the value in mg/dl by ; for the SI units for cholesterol in mmol/l, multiply the value in mg/dl by ) Procedures Lipid Quantitation Total cholesterol and TG levels were measured enzymatically on the Paramax Rx using the appropriate Paramax cholesterol and TG reagent (Dade International, Miami, Fla) tablets as described previously.15 This laboratory is accredited by the College of American Pathologists and participates in the ALERT Proficiency Testing program (Pacific Biometrics) for lipoprotein analysis. Phosphotungstic Acid Method HDL-C precipitating reagent (100 LIL) (Dade International) was used to precipitate VLDL and LDL cholesterol in 500 LIL of serum. After centrifugation (l,500g for 20 minutes), HDL-C was measured in the supernatant on the Paramax using normal saline and the Paramax Comprehensive Calibrator I (Dade International) as the low and high calibrators, respectively. Modified Reference Method Ultracentrifugation was performed following the Lipid Research Clinic's protocol as described previously.16 Briefly, the plasma sample was spun in a fixed angle rotor for 18 methods use a 2-step procedure to determine HDL-C concentrations. The first step uses a combination of a polyanion (such as heparin, phosphotungstate, or dextran sulfate) and a divalent cation (such as magnesium or manganese chloride) to precipitate apolipoprotein-b (apob)-containing lipoproteins. In the next step, the precipitated non-hdl lipoproteins are removed by centrifugation, and the HDL-C is determined by measuring the remaining cholesterol in the supernatant.z6~7 However, this method is problematic because it is fairly time consuming, affected by high triglyceride (TG) concentrations, not suitable for complete automation, and requires a large sample size. To circumvent these problems, investigators have developed "homogeneous" assays that measure HDL-C directly.8-14

3 Clinical Chemistry / ORIGINAL ARTICLE Liquid N-geneous HDL-C Assay HDL-C levels were determined by the N-HDL assay using the Hitachi 717 according to manufacturer's instructions. Briefly, 3 ul of sample was added to 300 ul of reagent 1 and incubated at 37 C for 5 minutes. A second reagent (100 LLL) was added at 5 minutes and incubated for an additional 5 minutes. HDL-C was then measured using dual wavelengths of 600 and 700 nm. HDL-C controls from Sigma Diagnostics (high, medium, and low) were analyzed with each run. Precision Studies The precision profile for N-HDL was performed on 2 levels of pooled serum samples with HDL-C less than 42 mg/dl and 42 mg/dl or more. For the intra-assay (withinrun) precision test, 2 serum sample pools with HDL-C concentrations of 36 mg/dl and 56 mg/dl were assayed 20 times using the N-HDL method. All assays were conducted with a single reagent lot. For the interassay (between-run) precision test, aliquots of 2 serum sample pools having HDL-C levels of 31 mg/dl and 61 mg/dl were assayed 10 times in duplicate over 5 days. Fasting and Nonfasting Comparison Thirty-two subjects fasted overnight for 12 hours and their blood was drawn in the morning by venipuncture into EDTA tubes. They were then asked to consume a high-fat breakfast containing 37 g of fat and 0.26 g of cholesterol. A second blood sample was drawn 3 to 3.5 hours later. Fasting and postprandial samples were then assayed for TG and cholesterol using the Paramax, and for HDL-C using the NHDL and PTA methods. Interference Studies The N-HDL assay was evaluated for possible interference by ascorbic acid, bilirubin, and hemoglobin. For evaluation of ascorbic acid interference, HDL-C was determined in pooled sample aliquots spiked with ascorbic acid to give a final concentration range of 1.63 mg/dl to 30 mg/dl. HDL-C also was measured in pooled serum aliquots with total bilirubin levels ranging from 1.1 mg/dl to 36.6 mg/dl by the N-HDL assay. An endogenous hyperbilirubinemic sample (36.6 mg/dl) was serially diluted with normal serum (0.1 mg/dl) to obtain the different concentrations. Aliquots from a whole blood sample hemolyzed through several freeze-thaw cycles (hemoglobin level of 1.2 g/dl) and serially diluted with saline were analyzed for HDL-C (hemoglobin range, g/dl). (To convert to SI units for ascorbic acid in Limol/L, multiply the value in mg/dl by 56.78; for bilirubin in umol/l, multiply the value in mg/dl by 17.1; and for hemoglobin in g/l, multiply the value in g/dl by 10.) Total Error Total error (TE) was calculated by adding systematic error and random error as described by the NCEP, ie, TE = %B (CVT).17 Systematic error is determined by using the linear regression equation y = bx + a, in which b is the slope of the regression equation and a is the y-axis intercept.1819 At an HDL-C concentration of yf, systematic error is equal to -f ~ *"% where y = bx + a. Random error is defined as 1.96 x SD from the between-run precision study.18,19 Statistical Analysis Linear regression analysis was used to compare HDL-C values obtained by the N-HDL and PTA methods with the MREF method in normotriglyceridemic and hypertriglyceridemic samples. Paired t tests were used to compare differences between the mean HDL-C levels determined by the 3 methods in the 2 subgroups. Similarly, differences between fasting and nonfasting samples were analyzed by paired t test. The mean absolute percentage bias Z{(1 test - ref.l/ref) x 100}/n and the mean percentage bias Z{(test - ref. /ref.) x 100}/n were calculated for the N-HDL and PTA methods compared with M-REF. Intra-assay and interassay precision were evaluated by determining means, SDs, and coefficients of variation (CVs). Results Precision Studies The precision profile for the N-HDL cholesterol assay was performed on the Hitachi 717 using low (<42 mg/dl) and normal (>42 mg/dl) concentrations of HDL-C (range, mg/dl). Intra-assay and interassay CVs were less than 2% for both concentrations of HDL-C liable II. The total imprecision (CVT) for the N-HDL assay was 2.7% (HDL-C, <42 mg/dl) and 2.3% (HDL-C, >42 mg/dl). Validity of N-HDL Assay in the Presence of Interfering Substances The N-HDL assay was evaluated for possible interference by ascorbic acid, bilirubin, and hemoglobin. No Am J Clin Pathol 1998; 110: hours at 109,000g. The top (d < g/ml) and bottom (d > g/ml) fractions were collected into separate tubes, and HDL-C was measured in 500 ul of the bottom fraction after precipitation of low-density lipoprotein (LDL) cholesterol using 100 ul of Magnetic HDL Precipitating reagent from Polymedco. The HDL-C in the supernatant was measured on the Paramax. The M-REF method was validated against the CDC Reference method using low, medium, and high reference samples provided by M. Kimberly, PhD, from the CDC, Atlanta, Ga. The M-REF method was within ± 5% of the CDC target values.

4 Hubbard et al / EVALUATION OF A HOMOGENEOUS HDL-C ASSAY Table II Precision Profile for the N-HDL Cholesterol Assay Intraassay (n = 20) Interassay (n = 20) N-HDL Mean ± SD (mg/dl) C V (%) Mean ± SD (mg/dl) CV (%) Level 1 Level ± ± ± ± o o -a _i Q n = 131 r=0.98 N-HDL = REF Sy/x = I i HDL E (mg/dl) Triglyceride <400 mg/dl e Triglyceride <400 mg/dl ^v^v 40 - n = 131 r=0.98 PTA =REF Sy/x = 3.21 #***' r 100 i 20 M-REF HDL-C (mg/dl) 1 40 i 60 M-REF HDL-C (mg/dl) Figure I I Comparison of the Liquid N-geneous high-density lipoprotein cholesterol assay (N-HDL; Genzyme Diagnostics, Cambridge, Mass) with M-REF (A) and the phosphotungstic acid (PTA) method with modified CDC reference procedure for high-density lipoprotein cholesterol (HDL-C) measurement (M-REF) (B) in 131 samples with triglycerides levels < 400 mg/dl. The equation for the regression line and the correlation coefficient is shown. significant interference was observed for ascorbic acid (P=. 1), hemoglobin (P =.7), and bilirubin (P -.2) at levels up to 30 mg/dl, 1.2 g/l, and 36.6 mg/dl, respectively. Also, the mean percentage bias was less than ± 2% for each interfering substance. Comparison of HDL-C Values Obtained From PTA, N-HDL, and M-REF For the comparison studies, HDL-C concentrations were measured in 181 fasting samples by the 3 assays using normotriglyceridemic (TG, <400 mg/dl, range, mg/dl; n = 131) and hypertriglyceridemic (TG, >400 mg/dl, range, 400-1,665 mg/dl; n = 50) subgroups. Linear regression for the N-HDL and PTA assays IFigure II show good correlation with M-REF for TG concentrations less than 400 mg/dl (r = 0.98 and 0.98, respectively). However, as shown in IFigure 21, there is only fair correlation of PTA with M-REF for TG levels of 400 mg/dl or more (r = 0.84), while excellent correlation is seen for N-HDL with M-REF (r = 0.96). In addition, the mean HDL-C value of PTA was significantly higher when compared with M-REF for the overall TG range and in normotriglyceridemic and hypertriglyceridemic subgroups (P <.001), whereas the mean HDL-C value of N-HDL did not show any statistical difference with M-REF for the overall TG range and in samples 498 Am J Clin Pathol 1998;110: with TG levels less than 400 mg/dl (P >.7) ITable 21; however, with hypertriglyceridemic samples, the N-HDL assay gave higher values than the M-REF (P <.002). The mean absolute percentage bias of PTA vs M-REF was more pronounced than that obtained with N-HDL vs MREF in each sample group ITable 31. Also, as shown in IFigure 31, a small mean percentage bias (0.9%) is seen with N-HDL that increases only slightly in hypertriglyceridemic samples (3.3%), while a larger mean percentage bias (2.6%) is seen with PTA that becomes more pronounced in hypertriglyceridemic samples (9%). The TE of the N-HDL assay was less than 9.0% in normotriglyceridemic and hypertriglyceridemic samples at both levels of HDL-C ITable 41, thus meeting the 1998 NCEP TE guidelines for HDL-C (<13%). The PTA assay met the 1998 NCEP performance criteria for TE only in normotriglyceridemic samples. Because of a large positive bias, the PTA assay did not meet the 1998 NCEP guidelines for TE in hypertriglyceridemic samples. Validity of PTA and N-HDL Assays in the Postprandial State To ascertain the validity of HDL-C measurement in the postprandial state, blood was drawn from 32 subjects after an overnight fast and following a standard high fat meal 03 N-HDL = Liquid N-geneous high-density lipoprotein assay (Genzyme Diagnostics, Cambridge, Mass); CV = coefficient of variation.

5 Clinical Chemistry / ORIGINAL ARTICI.R ootriglyceride Triglyceride >400 mg/dl 400 mg/dl 80- I l ^ ** n = 50 r=0.96 N-HDL = REF Sy/x = n = 50 r=0.84 PTA = REF Sy/x = M-REF HDL-C (mg/dl) M-REF HDL-C (mg/dl) Figure 21 Linear regression analysis plot of the Liquid N-geneous high-density lipoprotein cholesterol assay (N-HDL; Genzyme Diagnostics, Cambridge, Mass) vs M-REF (A) and phosphotungstic acid (PTA) vs modified CDC reference procedure for high-density lipoprotein cholesterol (HDL-C) measurement (M-REF) (B) in 50 samples with triglycerides levels >400 mg/dl. The equation for the regression line and correlation coefficient is shown. Table 21 Comparison of HDL-C Determined by N-HDL and PTA Assays With Modified CDC Reference Method* M-REF* Overall (n = 181) TG <400 mg/dl (n. 131) TG >400 mg/dl (n : 50) PTA* 43.2 ± ± ±7.9 N-HDL* f 44.9 ± ± 16.4f 35.1±9.2 f 43.5 ± ± ± 8.0 t HDL-C = high-density lipoprotein cholesterol; N-HDL = Liquid N-geneous high density lipoprotein assay (Genzyme Diagnostics. Cambridge, Mass); PTA = phosphotungstic acid; CDC = Centers for Disease Control and Prevention; M-REF = modified CDC reference procedure for HDL-C measurement; TG = triglycerides. *Data are given as mean ± SD (mg/dl). ^Significantly different from M-REF method by paired t test. P<.002. Table 31 Absolute Bias Between PTA and N-HDL With Modified CDC Reference Method* Overall (n = 181) TG <400 mg/dl (n =131) TG >400 mg/dl (n = 50) PTA N-HDL PTA = phosphotungstic acid; N-HDL = Liquid N-geneous high density lipoprotein assay (Genzyme Diagnostics, Cambridge, Mass); CDC = Centers for Disease Control and Prevention; TG = triglycerides. *Data are given as the absolute bias (%). (cholesterol, 0.26 g; fat, 37 g). liable 51 shows the effect of the postprandial state on the concentrations of TG, total cholesterol, and HDL-C. TG levels were significantly higher in postprandial samples than in fasting samples (P <.005). The HDL-C measurements from the PTA and N-HDL assays showed small but significant decreases (3%) between the fasting and nonfasting state (P <.005). Discussion Currently, most clinical laboratories use precipitationbased methods to measure HDL-C. While studies have shown that these methods give precise results when compared with ultracentrifugation,20 these assays are affected by high TG levels and are unsuitable for complete automation because they require a centrifugation step. This has prompted the development of direct methods for HDL-C measurement. Initially, Sugiuchi et al14 developed a rapid assay for direct measurement of HDL-C using polyethylene glycol-modified enzymes (PEGME) and sulfated a-cyclodextrin. This method does not require sample pretreatment, thus making it suitable for complete automation. By using a Hitachi 911 automated analyzer, they found that this assay correlated well with precipitation-based methods (heparin-ca2+, dextran-sulfate/mgcl2, PTA/MgCl2) and an ultracentrifugation method. This assay Am J Clin Pathol 1998; 110: o0

6 Hubbard et al / EVALUATION OF A HOMOGENEOUS HDL-C ASSAY 120 CO i B - '«fow* v*; ,000 1,250 1,500 Triglyceride (mg/dl) Figure 31 Bias between the Liquid N-geneous high-density lipoprotein cholesterol assay (N-HDL; Genzyme Diagnostics, Cambridge, Mass) and M-REF (A) and phosphotungstic acid (PTA) and modified CDC reference procedure for high-density lipoprotein cholesterol (HDL-C) measurement (M-REF) (B) as a function of triglyceride concentration. The thinner line represents the regression of the bias vs triglyceride concentration. Table 41 Total Analytical Error of The N-HDL and PTA Assays* Overall N-HDL <42 mg/dl >42 mg/dl PTA <42 mg/dl >42 mg/dl TG <400 mg/dl 6.2f , 6.0* f TG >400 mg/dl f N-HDL = Liquid N-geneous high density lipoprotein assay (Genzyme Diagnostics, Cambridge, Mass); PTA = phosphotungstic acid; TG = triglycerides. *Data are given as the total error (%). T Meets 1998 National Cholesterol Education Program high-density lipoprotein cholesterol total error performance criteria: <13%. Table 51 Effect of Preprandial and Postprandial State on TG, Cholesterol, and HDL-C Concentrations* Fasting TG Cholesterol PTA-HDL N-HDL ±75.3 ±26.1 ± 15.0 ± 13.0 Nonfasting ±84.2 f ± 26.1 f ±14.3+ ± 12.3t TG = triglycerides; HDL-C = high-density lipoprotein cholesterol; PTA = phosphotungstic acid; N-HDL = Liquid N-geneous high density lipoprotein cholesterol assay (Genzyme Diagnostics, Cambridge, Mass). *Data are given as mean ± SD (mg/dl). 'Significantly different from fasting by paired I test, P<.005. was unaffected by hemoglobin and ascorbic acid, but bilirubin up to 40 mg/dl produced a slight negative error (<10%). Harris et al8 subsequently compared the performance of a lyophilized form of the PEGME/a-cyclodextrin sulfate HDL-C assay with a modified reference method using a Hitachi 911 analyzer. At an HDL-C concentration of 50 mg/dl, the PEGME assay produced a bias that exceeded the current NCEP guidelines in samples with TG levels between 200 and 600 mg/dl (bias = 10.4%) and in samples with TG levels 600 mg/dl or more (bias = 15.30%). However, the TE 500 Am J Clin Pathol 1998; 110: met the current NCEP target. This study did not address the effect of interfering substances other than TG levels on this direct HDL-C assay. Nauck et a l " evaluated a 4-reagent homogeneous HDLC assay comprising PEG for wrapping chylomicrons, VLDL, and LDL, antibodies specific for apo B and apo CIII, enzymes for enzymatic cholesterol determination, and guanidine salt for stopping the enzymatic reaction and for solubilizing the non-hdl lipoprotein complexes. By using a Hitachi 911 analyzer, this method correlated well with a precipitation-based method; however, this assay was not compared with an ultracentrifugation reference method. Hemoglobin at concentrations of mg/dl or more markedly increased H D L - C c o n c e n t r a t i o n s, whereas bilirubin at concentrations of 6.0 mg/dl or more lowered HDL-C measurements. In addition, the requirement of 4 different reagents hinders the feasibility of automation. Nauck et al 13 conducted a multicenter study using the lyophilized PEGME/a-cyclodextrin sulfate HDL-C assay using different Hitachi analyzers. The homogeneous assay showed good correlation compared with the PTA/MgCl 2 method, as well as with ultracentrifugation at various centers. While hemoglobin concentrations of 100 mg/dl or Triglyceride (mg/dl) 1,750

7 Clinical Chemistry / ORIGINAL ARTICLE Nauck et al" and Huang et al9 analyzed the effect of TG levels on different homogeneous HDL-C assays using Intralipid. Nauck et al" found marked increases in HDL-C results when TG levels were increased to 380 mg/dl or more, while Huang et al9 found that 200 mg/dl of TG decreased the HDL-C readings by approximately 18%. Importantly, these interference studies may not be accurate because the addition of Intralipid introduces matrix effects that are not evident when true hypertriglyceridemic samples are studied as in the present study. A review of the literature shows that up to this point, studies evaluating PEGME and polyanion-polymer/detergent homogeneous HDL-C methods have been performed on assays that use lyophilized reagents. We describe the evaluation of an HDL-C assay that uses liquid reagents in determining HDL-C and compare this method to a modified reference procedure and a PTA precipitation-based method performed on a Paramax. In our evaluation of this method, we found excellent intra-assay and interassay precision of the N-HDL cholesterol assay. The NCEP recommends using an SD of less than 2.5 mg/dl for low HDL-C concentrations (<42 mg/dl) and a CV of less than 6% for an HDL-C level of 42 mg/dl or more as guidelines for evaluating the precision of HDL-C assays. The interassay precision of the N-HDL assay (Table 1) meets the current guidelines, as well as the 1998 precision goals for HDL-C concentrations less than 42 mg/dl (SD, <1.7 mg/dl) and more than 42 mg/dl (CV, <4%) in normotriglyceridemic and hypertriglyceridemic samples. In addition, while the conventional precipitation method used in our laboratory (PTA; SD, 1.2 mg/dl; CV, 3.6%) met the current and 1998 NCEP precision guidelines, the N-HDL assay was superior (SD, 0.6 mg/dl; CV, 1.6%). The N-HDL assay met the current (bias, < ± 10%) and the 1998 (bias, < ± 5%) NCEP recommendations for accuracy in 131 norrnotriglyceridemic (bias, 0.9%) and 50 hypertriglyceridemic (bias, 3.3%) samples. This is in contrast to findings by Harris et al12 in which the lyophilized form of this assay had a 25% positive bias compared with the reference method in hypertriglyceridemic samples (n = 26). Thus, the liquid N-HDL assay performs better than the lyophilized N-HDL assay in hypertriglyceridemic samples. The HDL-C selective detergent in reagent 2 of the liquid N-HDL assay, which is different from the detergent in the lyophilized NHDL assay, may account for the improved performance. Our in-house PTA method met the current NCEP accuracy standards in samples with TG levels less than 400 mg/dl and in samples with TG levels ranging from 400 to 700 mg/dl but did not meet the NCEP accuracy goals in samples with TG levels more than 700 mg/dl. Because of the less-than-desirable performance of the PTA method in hypertriglyceridemic samples, our laboratory presently used the magnetic HDL assay (Polymedco, Cortlandt Manor, NY) in samples with TG levels of 700 mg/dl or more. The N-HDL and PTA assays also met the current NCEP goal for total error (<22%) in normotriglyceridemic and hypertriglyceridemic samples. However, only the N-HDL assay met the 1998 TE performance goal (<13%) in normotriglyceridemic and hypertriglyceridemic samples. This contrasts with findings by Harris et al,12 in which the lyophilized form of the N-HDL assay did not meet the 1998 TE goals in hypertriglyceridemic samples because of a large positive bias. The NCEP ATP II states that HDL-C measurement can be made in nonfasting samples provided that interpretation of nonfasting results considers a 5% to 10% underestimation of HDL-C.21 It is important to remember that the error in HDL-C measurement in the nonfasting state is an overestimation of risk. However, management of persons with such levels will depend on obtaining fasting lipid profiles. We found a small but significant decrease in HDL-C (<3%) in the postprandial state as measured by the N-HDL and PTA methods. Thus, both assays reflected the expected postprandial physiologic changes. Increasing hemoglobin concentrations had little effect on the N-HDL assay, which is in agreement with data from other studies that have evaluated similar "homogeneous" HDL-C assays.9 Also, the N-HDL assay was unaffected by increasing concentrations of ascorbic acid and bilirubin in our study. Similar results involving the resistance of the lyophilized form of this assay to these reducing agents were found by Harris et al.12 In contrast, 3 of the 4 studies Am J Clin Pathol 1998; 110: less did not interfere with this assay, bilirubin at concentrations of more than 10 mg/dl produced a negative effect on HDL-C concentrations. Importantly, the bias calculated for this assay was not determined by direct comparison with the reference method. Furthermore, the PEGME method was not compared with the reference method for hypertriglyceridemic samples. Harris et al12 and Huang et al9 were the first to evaluate the performance of a lyophilized homogeneous HDL-C assay that uses polyanion-polymer/detergent in determining HDL-C. Harris et al12 found that this assay correlated well with the PTA/MgCl, method and the reference method and showed that bilirubin and ascorbic acid had little effect on this assay. However, there was a consistent positive bias in hypertriglyceridemic samples (n = 26) using the lyophilized formulation on a Hitachi 911 compared with the modified reference method. In contrast, Huang et al9 compared this assay only with a precipitation-based method using a Hitachi 7450 and found average correlation in normotriglyceridemic and hypertriglyceridemic samples. Hemoglobin had no effect on this assay, while bilirubin markedly decreased the HDL-C measurements.

8 H u b b a r d e t al / EVALUATION OF A HOMOGENEOUS HDL-C ASSAY From the 'Department of Pathology, University of Texas Southwestern Medical Center, the 'Division of Clinical Chemistry, Parkland Memorial Hospital, the ^Department of Pathology and Laboratory Medicine, VA North Texas Health Care System, and the 4Department of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, Texas. This work was supported by a grant from Genzyme Diagnostics, Cambridge, Mass. Address reprint requests to Dr Jialal: Associate Director, Division of Clinical Pathology, Department of Pathology, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd. Dallas, TX References 1. NIH Consensus Conference: triglyceride, high-density lipoprotein, and coronary artery disease. JAMA. 1993;269: Wiebe DA, Warnick GR. Measurement of high-density lipoprotein cholesterol concentration. In: Rifai N, Warnick GR, eds. Laboratory Measurement of Lipids, Lipoproteins and Apolipoproteins. Washington, DC: A A C C Press; 1994: The Expert Panel. Summary of second report of the NCEP expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel II). JAMA. 1993;269: Warnick GR, Cheung MC, Albers JJ. Comparison of current methods for high-density lipoprotein cholesterol quantitation. Clin Chem. 1979;25: Nguyen T, Warnick GR. Improved methods for the quantitation of total HDL and subclasses [abstract]. Clin Chem. 1989;35: Harris N, Galpchian V, Rifai N. Three routine methods for measuring high-density lipoprotein cholesterol compared with the reference method. Clin Chem. 1996;42: Huang Y, Kao J, Tsai K. Evaluation of two homogeneous methods for measuring high-density lipoprotein cholesterol. Ciin Chem. 1997;43: Okamoto Y, Tanaka S, Nakano H. Direct measurement of HDL cholesterol preferable to precipitation method [letter]. Clin Chem. 1995;41: Nauck M, Marz W, Haas B, et al. Homogenous assay for direct determination of high-density lipoprotein cholesterol evaluated. Clin Chem. 1996;42: Harris N, Galpchian V, Thomas J, et al. Three generations of high-density lipoprotein cholesterol assays compared with ultracentrifugation/dextran sulfate-mg2+ method. Clin Chem. 1997;43; Nauck M, Marz W, Jarausch J, et al. Multicenter evaluation of a homogeneous assay for HDL-cholesterol without sample pretreatment. Clin Chem. 1997;43: Sugiuchi H, Uji Y, Okabe H, et al. Direct measurement of high-density lipoprotein cholesterol in serum with polyethylene glycol-modified enzymes and sulfated (Xcyclodextrin. Clin Chem. 1995;41: Jialal I, Hirany SV, Devaraj S, et al. Comparison of an immunoprecipitation method for direct measurement of LDL cholesterol and beta-quantification (ultracentrifugation). Am J Clin Pathol 1995;104: Manual of Laboratory Operations: Lipid Research Clinics Program. Vol. 1. Washington DC: Department of Health, Education and Welfare; DHEW publication (NIH) Bachorik PS, Ross JW. National Cholesterol Education Program recommendations for measurement of low density lipoprotein cholesterol: executive summary. Clin Chem. 1995;41: Carey RN, GarberCC. Evaluation of methods. In: Kaplan LA, Pesce AJ, eds. Clinical Chemistry: Theory, Analysis and Correlation. 3rd ed. St Louis, Mo: Mosby; 1996:402^ Westgard JO, Carey RN, Wold S. Criteria for judging precision and accuracy in method development and evaluation. Clin Chem. 1974;20: Nauck M, Friedrich I, Wieland H. Measurement of LDL and VLDL cholesterol with precipitation techniques: a comparison with the ultracentrifugation method. Klin Lab. 1994;40; C o h n J S, McNamaraJR, SchaeferEJ. Lipoprotein cholesterol concentrations in the plasma of human subjects as measured in the fed and fasted states. Clin Chem. 1988;34: Warnick GR, Wood PD. National Cholesterol Education Program recommendations for measurement of high-density lipoprotein cholesterol: executive summary. Clin Chem. 1995;41: Livshits G, Weisbort J, Meshulam N, et al. Multivariate analysis of the twenty-year follow-up of the Donolo-Tel Aviv prospective coronary artery disease study and the usefulness of high density lipoprotein cholesterol percentage. Am J Cardiol. 1989;63: Am J Clin Pathol 1998; 110: that evaluated the PEGME/sulfated a-cyclodextrin assay found that bilirubin negatively influenced HDL-C measurements Thus, the N-HDL assay seems to be less affected by bilirubin than the PEGME/sulfated acyclodextrin assays. We found the N-HDL assay to be a precise and accurate method for evaluation of HDL-C as determined by our comparison with a modified CDC reference method and a conventional precipitation assay. Not only did the N-HDL assay correlate well with M-REF, but it also met the 1998 NCEP goals for TE and accuracy. Furthermore, the N-HDL assay was more precise and accurate than our in-house PTA method and was not affected by increasing concentrations of ascorbic acid, bilirubin, or hemoglobin. In addition to being accurate and precise, the N-HDL assay is rapid, uses a small sample volume (3 LIL), and does not require sample preparation or centrifugation, thus making it well suited for complete automation. We believe that the N-HDL assay will have an important role in routine clinical laboratories for HDL-C determination.