Review. Standardization of Measurements for Cholesterol, Triglycerides, and Major Lipoproteins

Transcription

1 Standardization of Measurements for Cholesterol, Triglycerides, and Major Lipoproteins G. Russell Warnick, MS, MBA, 1 Mary M. Kimberly, PhD, 2 Parvin P. Waymack, PhD, 2 Elizabeth T. Leary, PhD, 3 Gary L. Myers, PhD 2 ( 1 Berkeley Heart Lab, Alameda, CA, 2 Centers for Disease Control and Prevention, Atlanta, GA, 3 Pacific Biometrics Inc., Seattle, WA) DOI: /6UL9RHJH1JFFU4PY Abstract This review evaluates the status of standardization of lipids and lipoproteins. Prerequisites and some basic principles for standardization are provided. The reference systems for cholesterol, HDL cholesterol (HDL-C), LDL cholesterol (LDL-C), triglycerides (TG), apolipoprotein A-I (apoa-i), apolipoprotein B (apob), and lipoprotein(a) (Lp[a]) are described. Brief descriptions of the standardization programs available for each of these analytes are also provided. Finally, the review addresses some of the challenges in standardizing these markers of cardiovascular disease (CVD). The standardization programs described have contributed to improvements in laboratory measurements of lipids and lipoproteins. Our intention is that clinical laboratory professionals and manufacturers of in vitro diagnostics will use these resources to standardize lipid and lipoprotein measurements. Manufacturers must take the initiative to thoroughly evaluate their products and ensure traceability to the reference systems. Standardization in the context of this discussion refers to the process by which results of clinical laboratory assays are made consistent in terms of accuracy or agreement with a reference measurement procedure (RMP). 1 The main objective for standardization is to ensure that reported results for analytes agree across measurement systems and laboratories and over time. Standardization is especially important for the lipoproteins and their lipid and protein constituents because treatment decision points have been established by expert consensus through the National Cholesterol Education Program (NCEP). 2 These decision points are based on population distributions and risk relationships observed in epidemiology studies, many of which were standardized to a common accuracy base. Standardization of laboratory measurements to the same reference system ensures that patient characterization is consistent with the NCEP recommendations. 3-6 Relatively small biases can result in misclassification of patients, making accuracy of measurements especially important. Prerequisites of Standardization Successful standardization has prerequisites. First, the analyte of interest should have an unambiguous definition. For example, cholesterol is a molecule existing either in the free form or esterified to various fatty acids. In total cholesterol (TC) analysis, the various esterified forms are hydrolyzed; consequently, TC measured as free cholesterol molecules can be defined unambiguously. By contrast, the major lipoprotein classes are heterogeneous and polydisperse collections of particles with a range of properties, in some cases overlapping. Thus, lipoproteins are historically defined in terms of separation procedures based on their physical properties, which may not correspond with functional properties or correlate with pathologies. For example, low-density lipoprotein (LDL) has been defined as the lipoprotein fraction with a density between and g/ml 7 ; however, in practice LDL is usually separated at the native density of serum (1.006 Kg/L). The d >1.006 Kg/L density fraction includes some, but not all, of the remnant lipoproteins such as intermediate-density lipoprotein (IDL). Also, some, but not all, of the lipoprotein(a) (Lp[a]) particles have a density <1.063 Kg/L and are similar but not equivalent functionally to the bulk of the LDL particles. Included in this fraction as well may be larger apolipoprotein E (apoe)-containing high-density lipoprotein (HDL) particles, which are very different in composition and function from the more common apobcontaining LDL particles. 8 Therefore, an unambiguous definition of LDL particles representing both physical characteristics and functional properties may be unattainable. As a result, the criteria used for defining RMPs for the major lipoproteins must be clearly specified and a fraction defined in terms of a particular separation method considered to be the gold standard may not be achievable by other methods. Secondly, to achieve standardization, a reference system consisting of a hierarchal framework of validated and reliable analytical methods linked by reference materials must be in place. 9 The gold standard for most analytes is a primary RMP, demonstrated to be traceable to an SI unit and highly accurate with well defined performance characteristics. Secondary RMPs are often linked to the primary RMP and should be reasonably convenient, stable, reliable, and capable of transfer to other laboratories. Secondary RMPs are ideally calibrated with a primary standard of defined content. Primary or secondary RMPs can be used to assign values to secondary reference materials where the values are then traceable to the primary standard. Primary RMPs are typically used to validate more accessible secondary RMPs, which are less demanding and more cost effective. In the absence of a qualified primary RMP, alignment to a secondary RMP is beneficial. Downloaded labmedicine.com from August 2008 j Volume 39 Number 8 j LABMEDICINE 481

2 A third requirement for standardization is a reliable means of transferring the accuracy base from the RMP to field methods. This generally involves the use of commutable reference materials (ie, those representative of actual patient specimens). Many commercially available quality-control or calibrator materials are not commutable and hence can cause errors in the accuracy transfer. Key Concepts Relating to Standardization Before further considering the elements of a reference system as applied to the lipoproteins and their lipid and protein constituents, it is useful to review some of the principles and the terminology associated with analytical measurements. Analytical Accuracy Analytical accuracy indicates the agreement between a measured quantity and the true value. Systematic deviation from the true value is called bias, which may be proportional (ie, related to the concentration) or constant across all measurement values. Analytical Precision Precision, or measurement consistency, is a prerequisite for accuracy. It is defined as the agreement among multiple measurements of the same quantity in a sample. A method may have minimal overall systematic error or bias, but if imprecise, it still will be inaccurate for some individual measurements. Preanalytical Variations In addition to analytical imprecision, important contributions to analyte variation derive from preanalytical sources of variation including biological, behavioral, clinical (disease-induced and drug-induced), and variations from sample collection and handling. Some of these sources of variation may not be controllable, such as age, circadian rhythm, or pregnancy. Or, they are related to the clinical conditions being monitored and should be considered in interpreting results. Controllable sources of variation, such as fasting status, alcohol intake, exercise, and sample collection and handling procedures, should be carefully standardized as specified by NCEP guidelines. 3-6 Serum, usually collected in vacuum tubes with clotting enhancers, has been the preferred sample type for lipoprotein measurement in the routine clinical laboratory. Ethylenediaminetetraacetic acid (EDTA) plasma was the traditional choice in lipid research laboratories, especially for lipoprotein separations, because the anticoagulant enhanced analyte stability by chelating metal ions required for the activities of a variety of proteases. EDTA also allows prompt sample processing without having to wait for clotting; however, EDTA plasma has potential disadvantages that have discouraged its routine use. Inadequate mixing can result in microclots that could plug the sampling probes on chemistry analyzers. Also, EDTA osmotically draws water from red cells, diluting the plasma constituents, which can vary depending on sample fill volume. Because the NCEP medical decision points are based on serum values, cholesterol measurements made on EDTA plasma require correction by the factor of For accuracy transfers, such as in the Cholesterol Reference Method Laboratory Network (CRMLN) certification program, stringent sample handling is essential. Fresh human sera collected under carefully defined conditions are analyzed within a defined time period to ensure commutability. For some analytes, preanalytical variation may be a significant and greater part of the overall variation considering the greatly improved analytical systems available today. Biological variation for total cholesterol levels averages 6.1% and can be as high as 11%, even with rigorous patient preparation. 10 Such uncontrollable sources of preanalytical variation must be carefully considered in result interpretation. It is recommended that physicians base treatment decisions on results from 2 samples, collected 2 weeks apart, to minimize the affect of biological variation. 11 Matrix Effects and Commutability of Reference Materials Commercially available reference materials, which are usually lyophilized serum pools or stabilized serum-based liquid, are convenient for use; however, the manufacturing processes, especially lyophilization and spiking with materials not native to serum, may alter the measurement properties of the analytes with some assay systems. These materials are not representative of patient specimens (ie, are non-commutable ). In other words, these materials possess a matrix effect. A matrix effect has been defined as the influence of a property of a sample, independent of the presence of the analyte, on the measurement process and thereby on the measurable quantity. In this concept of a matrix effect, the sample matrix includes all the components of a material system, except the analyte itself. 9 The consequences of loss of commutability on standardization efforts are erroneous conclusions about the performance of an analytical system. Secondary reference materials, survey materials, and calibrators and control materials for lipids, lipoproteins, and apolipoproteins often exhibit varying degrees of loss of commutability that has been attributed to the contributions of matrix effects, as well as to inclusion of non-native forms of the analytes. Approaches to evaluate matrix effects and commutability have been developed, and the CLSI is currently developing a guideline for characterization and qualification of commutable reference materials. 16 Analytical Performance Goals Expert laboratory panels convened by the NCEP established analytical performance goals for measurements of the lipids and major lipoproteins based on the clinical need to reliably categorize patients (Table 1). 3-6 For example, in analysis of TC, the performance goal for total error is 8.9%. That is, the overall error should be such that 95% of individual cholesterol measurements fall within ±8.9% of the RMP value. The bias and CV targets presented in Table 1 are representative of performance that will meet the NCEP goals for total error. Table 1_NCEP and CLIA Criteria for Lipid and Lipoprotein Testing NCEP Performance Criteria CLIA Evaluation Criterion Analyte Inaccuracy Imprecision Total Error Total Error TC ±3% RV a CV b 3% 8.9% ±10% HDL-C ±5% RV SD c 1.7 at (<42 mg/dl) CV 4.0% at 13% ± 25% ( 42 mg/dl) LDL-C ±4% RV 4% 12% ±30% TG ±5% RV 5% 15% ±30% a RV = reference value assigned by CDC reference measurement procedures. b CV = coefficient of variation. c SD = standard deviation. Downloaded 482 from LABMEDICINE j Volume Number 8 j August 2008 labmedicine.com

3 Considerations in Transferring the Accuracy Base When establishing RMPs that effectively function as targets for laboratory standardization, a practical challenge is interfacing the accuracy base with the laboratory community on a broad scale. Using commutable and widely available secondary reference materials is the most efficient and economical way to transfer a reference-method accuracy base and multiplies the impact of laborious reference method analyses. When commutable reference materials are not available, which is the case for the lipid and lipoprotein analytes, a thorough understanding of a specific method s characteristic matrix susceptibilities is central to establishing an effective standardization strategy. Even when a material contains matrix-sensitive components, if the RMP is demonstrably stable and reproducible, it is possible to achieve accuracy (trueness) through strategies such as set point adjustments. These adjustments are based on comparison to the RMP by simultaneously analyzing split fresh-patient specimens and reference materials. This type of comparison to initially establish accuracy and to subsequently monitor the relationship is usually only practical with secondary RMPs, although it is still an expensive and time-consuming approach. Nevertheless, it is the best means to validate traceability for manufacturers of diagnostic systems and for clinical laboratories seeking performance verification of their methods. Reference Measurement Procedures and Materials The status of the components that make up reference systems for the various lipids and lipoproteins is summarized in Table 2. The experience with TC has been used as a model for standardization of lipoprotein cholesterols, apolipoproteins, and triglycerides (TG). However, although HDL-C, LDL-C, apolipoproteins, and TG have similar commutability issues, when compared with total cholesterol they exhibit more complex, varied, and unique matrix challenges. When trying to apply this model, additional issues arise related to the definition of the analyte, modification of the analyte by processing, and non-native forms of the analyte. Cholesterol The standard for accuracy in blood cholesterol measurement is a hierarchy of methods consisting of the National Institute of Standards and Technology (NIST) primary RMP, the Centers for Disease Control and Prevention (CDC) secondary RMP, and a cholesterol primary standard with certified purity NIST SRM 911c (Table 2). The NIST primary RMP is an isotope-dilution mass spectrometry procedure (IDMS). 17 The CDC secondary RMP for TC is a modification of the extraction procedure of Abell, Levy, Brodie, and Kendall. 18 The cholesterol in ester form is released by saponification of a serum sample with alcoholic potassium hydroxide. This is followed by extraction with hexane, evaporation of an aliquot of the extract, and development of a chromophore at 620 nm with Liebermann-Burchard reagent (acetic anhydride, glacial acetic, and concentrated sulfuric acid). The method is calibrated using the primary reference standard, NIST SRM 911c. Typical performance of the method observed over 1 year is a coefficient of variation (CV) of 0.4% to 0.5%. The CDC secondary RMP is also used to measure cholesterol in the CDC HDL-C and LDL-C RMPs. Both the NIST and CDC RMPs have been identified by the Joint Committee for Traceability in Laboratory Medicine (JCTLM) as internationally recognized RMPs of higher order. 19 Comparisons of the NIST and CDC RMPs have demonstrated a small but persistent positive bias in the secondary RMP of +1.6%. 20 More than half of the bias is from cholesterol precursor sterols and phytosterols, which are extracted from serum in the secondary RMP. 21 Because the observed bias is consistent but small, the CDC RMP provides a practical basis for standardization of routine methods used in the clinical environment. Using the NIST RMP in this manner is impractical. Table 2_Lipid and Lipoprotein Reference Systems 1 Reference 1 Reference 2 Reference 2 Reference Analyte Measurement Procedure Material Measurement Procedure Material Cholesterol ID-MS (NIST) NIST SRM 911c Abell-Kendall (CDC) CDC Frozen Pools Pure cholesterol NIST SRM 909 NIST SRM 1951b HDL-C Not available Not available UC/Heparin-Mn2+-Abell-Kendall (CDC) CDC Frozen Pools Recommended by NCEP NIST SRM 1951b LDL-C Not available Not available Beta-quantification (CDC) Recommended CDC Frozen Pools by NCEP NIST SRM 1951b Triglyceride ID-MS (NIST) NIST SRM 1595 Methylene chloride Silicic acid- CDC Frozen Pools Tripalmitin chromotropic acid (CDC). NIST SRM 1951b Recommended by NCEP Lipoprotein(a) Not available Lyophilized purified Lp(a) Consensus ELISA method WHO/IFCC SRM 2B ApoA-1 HPLC-MS (CDC) BCR-CRM 393 Not available WHO Reference Reagent SP1-01 (for (primary standard only) (Purified ApoA-1) manufacturers). Value-assigned by CDC-RIA (Candidate) comparison method. ApoB Not available d = Not available WHO Reference Reagent SP3-08 (for (UC purified LDL) manufacturers). Value-assigned by NWLMDRL- Immunonephlometry comparison method. CDC = Centers for Disease Control and Prevention; NCEP = National Cholesterol Education Program; WHO = World Health Organization; NWLMDRL = Northwest Lipid Metabolism and Diabetes Research Laboratories. Downloaded labmedicine.com from August 2008 j Volume 39 Number 8 j LABMEDICINE 483

4 HDL Cholesterol The accuracy base for HDL-C is not as complete as that for TC (Table 2). Lipoproteins are a heterogeneous mixture of lipids and proteins that are not rigidly defined; therefore, significant overlap can exist in the physical properties of the major lipoprotein classes. HDL particles with varying cholesterol content exhibit a continuum of properties whether based on density, size, or electrophoretic mobility. Developing a complete reference system for HDL C will require that it be more precisely defined. Until a clear rationale or consensus for defining primary reference materials for HDL C is formulated, a basis for measurement by a corresponding primary RMP cannot be developed. The current reference point for HDL C measurement, as recommended by the NCEP Lipoprotein Measurement Working Group, is the CDC HDL-C RMP, a multi step procedure involving ultracentrifugation, precipitation, and cholesterol analysis. 1,6 This method involves separation and removal of very low density lipoprotein (VLDL) and chylomicrons, if present, by 18.5 hours of ultracentrifugation and separation of HDL by selective precipitation of non-hdl (including Lp[a], IDL and LDL) from the beta quantification bottom fraction (d kg/l) using 46 mmol/l heparin-manganese (Mn+2). HDL-C in the clear supernatant is quantified by the CDC cholesterol RMP. It is closely linked with the CDC LDL-C RMP (see LDL-C below). Factors limiting more widespread use of the CDC HDL-C RMP for routine standardization of HDL C measurements are the ultracentrifugation requirement, technical difficulty, and the large sample volume (5 ml) required. However, the method served as the RMP for the epidemiologic and clinical investigations sponsored by the CDC and NHLBI, and thus was the accuracy base from which NCEP cardiovascular disease (CVD) risk estimate decision points and population distributions were derived. Typical analytical precision of the method observed over 1 year is 1.4% CV. The method has also been recognized by JCTLM as an international RMP of higher order. In order to provide more widespread standardization, a designated comparison method (DCM) for HDL-C was developed by the CRMLN. The DCM requires a smaller sample volume (1 ml) and does not involve an ultracentrifugation step. The method is closely linked to the CDC HDL-C RMP, and the smaller sample volume requirement facilitates interactions with manufacturers. 22 The method uses a precipitation reagent formulation (magnesiumdextran sulfate) for separation of non-hdl that closely corresponds to the separation efficiency of the heparin-manganese reagent used in the CDC RMP. Cholesterol in the supernatant is measured by the CDC cholesterol RMP. Limiting triglyceride concentration in the comparison samples to 200 mg/dl eliminated the need for ultracentrifugation. LDL Cholesterol The CDC LDL-C RMP is the reference point for LDL-C recommended by the NCEP Lipoprotein Measurement Working Group. 4 The CDC reference procedure for LDL C is performed exactly like the CDC HDL-C RMP described above, but it is expanded to include analysis of the beta quantification ultracentrifuge bottom fraction (d >1.006 Kg/L) by the cholesterol RMP. The beta-quantification calculation is: LDL-C = Bottom fraction cholesterol HDL-C. The CDC HDL-C/ LDL-C reference procedure is based on the beta quantification ultracentrifuge procedure used by the Lipid Research Clinics 23 but differ in centrifugation temperature (18 C instead of 10 C) and density of saline overlay (0.195 M instead of 0.15 M). Because serum is used instead of EDTA plasma, a lower concentration of heparin manganese reagent appropriate for precipitation of serum is used accordingly. Typical analytical performance observed for the LDL-C secondary RMP at CDC over a 1-year period is a CV of 0.6% to 1.1% for the bottom fraction cholesterol and 1.0% to 1.3% for LDL-C. This method has also been recognized by the JCTLM as an international RMP of higher order. Secondary reference materials for LDL-C are available (Table 2). Establishing a system for accuracy-based target values for LDL-C analogous to that for total cholesterol is complicated by several issues. Similar to HDL-C, the basis for defining a primary reference material for LDL-C is extremely difficult to identify because, in individuals as well as within the entire patient population, there exists a heterogeneous collection of LDL particles with varying cholesterol content that exhibit a continuum of physical properties whether based on density, size, or electrophoretic mobility. Therefore, consensus primary reference materials for LDL C do not exist and, thus, a primary RMP for measurement has not been developed. The current operational definition of LDL C based on physical separations (ultracentrifugation and chemical precipitation) contains the mixture of lipoprotein cholesterol, which was defined as LDL-C by NCEP and includes some Lp(a) cholesterol, IDL-C, and the core LDL cholesterol. 4 The operational definition of LDL-C also incorporates the operational definition of HDL-C, with the tacit assumption that the specificity in the precipitation step is sufficient to adequately and consistently separate the continuum of LDL lipoprotein particles from the continuum of HDL lipoprotein particles. Manufacturers of the various clinically applied direct homogeneous methods for LDL-C, using different measurement principles that do not include a physical separation step, have accepted the challenge to account for varying cholesterol content among the heterogeneous mixture of lipoproteins called LDL-C. In principle, the commonly used Friedewald calculation 24 to estimate LDL-C from conventional lipid profile measurements represents a mixture of lipoprotein cholesterol similar to that measured by the CDC secondary RMP. Triglycerides A secondary RMP for TG was established at the CDC in 1963 and served as the accuracy base for the CDC-NHLBI LSP. It is based on the method of Carlson 25,26 and the techniques of Van Handel and Zilversmit 27 and of Lofland. 28 In 1993, the CDC RMP was modified to replace chloroform with methylene chloride. The initial step of the method combines methylene chloride extraction with silicic acid treatment. An aliquot of the triglyceride extract is hydrolyzed to glycerol using ethanolic potassium hydroxide, followed by oxidation by metaperiodate and color development with chromotropic acid reagent. The silicic acid treatment efficiently removes phospholipids and free glycerol, and it removes most of the monoglycerides and part of the diglycerides. Subsequently, NIST developed a primary RMP for TG based on IDMS. 29 The method measures total glycerides, which is defined as the sum of tri-, di-, and monoglycerides plus any free glycerol. Total glycerides represent the analytical species measured by most clinical laboratories using enzymatic methods for triglycerides. Since the primary RMPs and the secondary RMP for TG do not measure identical analytical species, a simple, direct comparison of the 2 methods is not possible. The best approximation is to compare the NIST-measured result for total glycerides to a CDC estimated result for total glycerides. The CDC estimate for total glycerides is obtained by adding the CDC RMP result, which includes triglycerides, most of the diglycerides, and a small amount of monoglycerides, to the CDC result for free glycerol. To measure Downloaded 484 from LABMEDICINE j Volume Number 8 j August 2008 labmedicine.com

5 glycerol concentration in serum lipid reference materials, the CDC developed an isotope dilution-gas chromatography-mass spectrometry method. 30 In a preliminary evaluation, the estimated total glycerides for the CDC secondary RMP was compared with the total glycerides measured by the NIST primary RMP using both frozen and lyophilized serum pools. There was an average bias of +1.1% between the 2 methods. Reference Materials for TC, HDL-C, LDL-C, and TG A procedure (C37-A) for preparing commutable reference materials for cholesterol has been developed by the CLSI. 31 This procedure can also be used to prepare commutable reference materials for HDL-C, LDL-C, and TG. 32,33 It was used to prepare NIST SRM 1951b. This SRM has been value-assigned using NIST s primary RMPs for cholesterol and total glycerides and the CDC s secondary RMPs for cholesterol, HDL-C, LDL-C, and TG. SRM 1951b is the only commercial reference material available with HDL-C and LDL-C reference values. The CDC also uses CLSI s C37-A to prepare secondary reference materials for use in the Lipid Standardization Program (LSP) and CRMLN, described in more detail subsequently. These materials are valueassigned using the secondary RMPs. Lipoprotein(a) The lack of standardization of Lp(a) measurements has confounded the interpretation of results across different clinical studies to determine the role of Lp(a) as a risk factor for coronary heart disease For assays of Lp(a) mass, the major comparability problem is due to the apo(a) size polymorphism that results in overestimation or underestimation of Lp(a) values. The size heterogeneity of Lp(a) comes from the genetic coding for the loop structures stabilized by intra-chain disulfide bonds called kringle (K) domains K4 and K5. This means the apo(a) proteins produced by the genes of different individuals can exhibit size isoforms containing from 12 to approximately 50 of the K4 looped structures. 38 A collaboration between a NHLBI-supported Lp(a) standardization project at the University of Washington Northwest Lipid Metabolism and Diabetes Research Laboratories (NWLMDRL) and the International Federation of Clinical Chemistry (IFCC) Working Group on Lp(a) resulted in considerable progress in developing an Lp(a) reference system. A consensus RMP for standardizing Lp(a) measurements using a KIV9 MAb, a 40-based ELISA method was chosen by the IFCC Working Group. 39 It has been used to assign a target value to the IFCC secondary reference material 36,37 SRM 2B, a lyophilized serum pool preparation, was designated as the First WHO/IFCC International Reference Reagent for Lipoprotein(a) for Immunoassay 40 and allows calibration based on a defined molar concentration rather than unspecified mass. Other studies have shown that a lyophilized purified Lp(a) preparation appears to meet all the criteria for a primary reference material. 41 The primary goal of manufacturers of Lp(a) diagnostic test kits should be production of Lp(a) assays unaffected by apo(a) size polymorphism. Apolipoproteins A-I and B A purified apolipoprotein A-I (apoa-i) material, developed by the European Bureau of Reference for use as a primary standard, 42 has been suitable for some immunochemical techniques. However, the physical and chemical properties of apob do not lend themselves to purification of a material suitable for immunochemical analysis. A stable fraction of LDL prepared by ultracentrifugation (d = to Kg/L) that excludes intermediate density lipoprotein or Lp(a) currently serves as a primary standard for apob. The ideal approach to establish an accuracy base would be to use widely accepted primary or secondary RMPs for valueassignment in apolipoprotein mass units on secondary reference materials (SRMs) with an appropriate assay matrix. These SRMs then could be used as calibrators to transfer apolipoprotein mass units to calibration materials used by manufacturers of apolipoprotein measurement kits. This process currently is not feasible because there are no universally accepted RMPs for analysis of apoa I or apob. 43 Primary and secondary RMPs for proteins depend on highly accurate protein determinations with primary standards. The CDC has investigated and established an enzymatic digestion, liquid chromatography, IDMS method to assign mass values to an apoa-i primary standard for use as a model for apolipoprotein primary standards. 44 The IFCC Committee on Apolipoproteins, together with manufacturers of apolipo protein diagnostic kits, began a collaborative program in 1988 to produce and evaluate SRMs for standardizing apolipoprotein measurements. In early 1989, more than 20 manufacturers and several reference laboratories evaluated 26 candidate SRMs and measured 10 frozen sera with assigned values for apoa I and apob. A lyophilized serum for apoa I and a liquid stabilized serum for apob were selected as reference materials on the basis of homogeneity, stability, reproducibility, and linearity upon dilution. The World Health Organization (WHO) has accepted these 2 serum matrix based materials as reference reagents: SP1 01, a lyophilized material for apoa I, and SP3 08, a frozen stabilized liquid preparation for apob. Mass units have been assigned using standardized immunoassay techniques and purified apoa I and LDL (d = to Kg/L) as primary calibrators. 45,46 Resources Available for Standardization Standardization Programs The NCEP recommended that lipid and lipoprotein measurements should be traceable to a national accuracy base. 3-6 These recommendations include performance criteria for clinical laboratory measurement procedures (Table 1). Achieving traceability to the accuracy bases for TC, TG, the lipoproteins, and the apolipoproteins requires a unified effort involving reagent and instrument manufacturers, government agencies, and clinical professions. To fulfill its commitment to improve the measurement of risk factors associated with CVD, the CDC has focused on 1) maintaining a high-quality lipid RMP laboratory, and 2) offering standardization services to United States and international public health, epidemiologic, research laboratories, and reagent manufacturers. The primary programs available to assist laboratories in standardizing lipids and lipoproteins measurement are CDC-NHLBI LSP and CRMLN. The CDC programs differ from conventional proficiency testing (PT) programs, in which most clinical laboratories participate to meet federal and state regulatory requirements, in that they provide an accuracy base through RMPs and commutable survey samples. In contrast, conventional PT programs generally evaluate participant performance based on peer-group means at specific times. Thus, they evaluate how well a laboratory compares to other laboratories using the same analytical systems but provide no mechanism for assessing or improving accuracy. Furthermore, the survey materials used in conventional PT may not be commutable with patient samples; therefore, they cannot establish traceability to an accepted accuracy base. Downloaded labmedicine.com from August 2008 j Volume 39 Number 8 j LABMEDICINE 485

6 CDC-NHLBI Lipid Standardization Program The Lipid Standardization Program (LSP) began in 1958 because scientists investigating the relation between serum lipid concentrations and CVD were concerned about the reliability of lipid analyses within laboratories and the comparability of results among laboratories. In response to this need, the CDC developed RMPs and materials to standardize these laboratories and initiated the LSP in cooperation with the NHLBI. The LSP provides a validated mechanism to establish, assess, and improve the accuracy of analytical measurements through traceability to the CDC s RMPs and to standardize measured values of lipids and lipoproteins, regardless of the analytical systems used. 47 Laboratories are standardized through a stepwise process involving a Qualifying Survey and subsequent quarterly Monitoring Surveys. Successful participation in quarterly analytical evaluations qualifies a laboratory for a certificate of standardization. The LSP protocol is designed so that a laboratory with the NCEP maximum allowable bias and imprecision will have a 95% chance of passing the bias evaluation. Laboratories with smaller bias and imprecision will have an even greater chance of passing, whereas laboratories with larger bias or imprecision will have less chance of passing. The bias and precision evaluations are based on the NCEP criteria shown in Table 1. Today, the LSP continues to provide accuracy-based reference materials for measuring TC, TG, and HDL-C in U.S.-based and international laboratories. The LSP is available to laboratories supporting epidemiologic studies and clinical trials, lipid research laboratories, public health laboratories, and reference laboratories. The LSP participants support more than 100 studies to investigate CVD and other related factors, including diabetes, nutrition, genetics, and health issues of women and minorities. Standardization ensures credible results and comparability among epidemiologic studies and clinical trials. The standardization of measurements using processed reference materials, including those used in the LSP, can be complicated by matrix effects in some analytical systems, so successful performance in the LSP requires that a participant s analytical system exhibit negligible matrix effects using CDC reference materials. For information about participation in CDC-NHLBI LSP, see Cholesterol Reference Method Laboratory Network (CRMLN) When the NCEP published guidelines about risk factors for coronary artery disease 2 and made recommendations for risk assessment, the CDC was faced with developing an approach to standardize the clinical laboratories providing testing services for patient management; however, the number of laboratories in the United States (more than 100,000) makes standardization using fresh patient comparisons a nearly impossible task. In contrast, the number of manufacturers that provide the diagnostic methods for measuring lipids and lipoproteins is relatively small. Therefore, the CDC focused its efforts toward the manufacturers and established the CRMLN to assist them in correctly calibrating their measurement systems and validating traceability to the accuracy base through a certification program. In the absence of truly commutable materials for accuracy transfer, the recommended approach to assess how well a laboratory method correlates with the accuracy base is to directly compare diagnostic methods to the RMPs using fresh patient specimens. 48 The CRMLN was the first network of its kind to use fresh sample comparisons as the basis of accuracy validation, and therefore serves as a model for other analytical systems that require fresh sample comparisons to validate traceability to an accuracy base. 49 The CRMLN laboratories use CDC RMPs or designated comparison methods (DCMs) that are closely linked to CDC RMPs. The CRMLN consists of 3 U.S. laboratories and 6 international laboratories (listed in Table 4). The performance of all methods used in the CRMLN for certifications is monitored bimonthly through surveys involving analysis of CDC serum reference materials. The CRMLN laboratories not meeting specific performance criteria (Table 3) in the bimonthly surveys are required to improve and maintain performance for 2 consecutive surveys before resuming reference laboratory activity. Representative performance of the CRMLN laboratories for the time period between May 2002 and July 2007 is shown in Table 3. Both the RMP and a DCM are used in the CRMLN for HDL-C. All 10 CRMLN laboratories perform the DCM; 4 perform the RMP. 1. Certification of Instrument Systems, Reagents, and Reference Materials. The CRMLN uses CLSI s EP9-A protocol as a basis to evaluate traceability to the accuracy bases. 50 The comparison protocol requires analyzing 40 fresh human serum specimens covering the expected analytical range. Results are evaluated based on the NCEP performance criteria. A dated Certificate of Traceability, valid for 2 years, is issued for each measurement system (including instrument, reagents, and calibrator) that meets the criteria. 2. Certification of Clinical Laboratories. An abbreviated protocol for total cholesterol is available for clinical laboratory certification. The protocol requires that the laboratory analyze 6 fresh serum specimens in duplicate in each of 3 runs. The TC values should cover a range that Table 3_Criteria and Performance of the Cholesterol Reference Method Laboratory Network Representative Performance Criteria Inaccuracy Imprecision Analyte Inaccuracy Imprecision Average Rate a Average Rate TC % bias 1% vs. CDC CV 1% 0.1% 97% 0.3% 99% HDL-C DCM bias 1 mg/dl vs. CDC SD 1 mg/dl 0.7 mg/dl b 84% 0.4 mg/dl 99% HDL-C UC bias 1 mg/dl vs. CDC SD 1 mg/dl 0.6 mg/dl b 90% 0.4 mg/dl 99% LDL-C % bias 2% vs. CDC CV 1.5% 0.3% 85% 0.9% 96% a Rate = Percentage of time the CRMLN laboratories met the stated criterion. b These are average absolute bias values. Downloaded 486 from LABMEDICINE j Volume Number 8 j August 2008 labmedicine.com

7 Table 4_Cholesterol Reference Laboratory Network Programs Laboratory Contact CRMLN services Other Subscription Services University of Washington, Department of Santica Marcovina, PhD Manufacturer Certification: TC, HDL-C, LDL-C ReLABS: standardization of HDL-C, TG, Medicine, Northwest Lipid Metabolism and Clinical Lab Certification: TC LDL-C, apoa-i, and apob measurements Diabetes Research Laboratories (206) using fresh patient specimens Wadsworth Center for Laboratories and Robert Rej, PhD Research, New York State Department Manufacturer Certification: TC, HDL-C of Health (518) Clinical Lab Certification: TC Pacific Biometrics Research Foundation Elizabeth Teng Leary, PhD Manufacturer Certification: TC, HDL-C Clinical Lab Certification: TC (206) (x208) Erasmus MC, University Medical Center Yolanda de Rijke PhD Manufacturer Certification: TC, HDL-C, LDL-C Dutch Foundation for Quality Assessment Rotterdam Clinical Lab Certification: TC in Clinical Laboratories: standardization for TC, HDL-C, and LDL-C (lyophilized, cryoprotected materials, and fresh frozen sera) Osaka Medical Center for Health Science Masakazu Nakamura, PhD Manufacturer Certification: TC, HDL-C, LDL-C and Promotion Clinical Lab Certification: TC Canadian External Quality Assessment David W. Seccombe, MD, PhD Manufacturer Certification: TC, HDL-C DigitalPT International EQA Collaboration Laboratory Clinical Lab Certification: TC (604) H.S. Raffaele Ferruccio Ceriotti, MD Manufacturer Certification: TC, HDL-C External Quality Assessment Program in Clinical Lab Certification: TC Clinical Chemistry (PROLARIT): Italy and Turkey Fundacíon Bioquímica Argentina, Daniel Mazziotta, MD Manufacturer Certification: TC, HDL-C EQAS in Argentina, Uruguay, and Paraguay Laboratorio de Referencia y Estandarización dmpeec@netverk.com.ar Clinical Lab Certification: TC en Bioquímica Clínica (LARESBIC) Beijing Institute of Geriatrics Wenxiang Chen, MD, MSc Manufacturer Certification: TC, HDL-C chenwenxiang@263.net Clinical Lab Certification: TC includes the NCEP recommended medical decision points for cholesterol of 200 and 240 mg/dl (5.18 and 6.22 mmol/l). Acceptable performance is documented by a Certificate of Traceability which is valid for 6 months. The CRMLN currently has no specific HDL-C or LDL-C certification program for clinical laboratories. A list of measurement systems and clinical laboratories with current certification information is at Also posted is information about the certification programs and contact information for the CRMLN laboratories. In addition to certification programs, some CRMLN laboratories offer varying types of external quality assurance (EQA) programs. These EQA programs are a primary focus for many international CRMLN members (Table 4). Lipoprotein(a) Standardization The NWLMDRL is the repository as well as the distributor of SRM 2B for Lp(a). It is available upon request to manufacturers and clinical laboratories, but it has well-defined limitations because of Lp(a) size polymorphism. The reference value to SRM 2B is provided in nmol/l, and there is no correction factor to transform the values from nmol/l to mg/dl because the size of Lp(a) varies from <300 to >800 Kda. NWLMDRL also valueassigns 5 individual samples with low to high Lp(a) values and small to large apo(a) isoforms. They are provided with SRM 2B to evaluate size-dependency. After transferring the value from SRM 2B to the assay calibrator, the laboratory results of repeated analyses of the 5 samples are evaluated. Apolipoprotein A-I and B Standardization Program The NWLMDRL conducts a standardization program for manufacturers using an IFCC calibration protocol. This protocol includes procedures to evaluate dose-response linearity and parallelism, intercept equality of the reference materials, and confirmation procedures using fresh serum. The reference materials used in this program are WHO/IFCC International Reference Reagents for apoa-i (SP1-01) and apob (SP3-08). The CDC is the repository for these materials, which are available to manufacturers for use in assigning target values to calibrator or quality-control materials, and to evaluate performance of new analytical systems. These reference preparations also are available to international reference laboratories having responsibility for standardizing apoa-i and apob within their country or region. Clinical laboratories can calibrate or confirm the accuracy of their assays using 3 frozen serum pools with assigned values for apoa-i and B (traceable to the WHO/IFCC Reference Reagents) available from the NWLMDRL. For more information, contact NWLMDRL, University of Washington, 401 Queen Anne Avenue N, Seattle, WA 98109; tel.: (206) National Proficiency Survey Programs The U.S. Congress passed the Clinical Laboratory Improvement Amendments (CLIA) in 1988 establishing quality standards for all testing to ensure the accuracy, reliability, and timeliness of patient test results regardless of where the test was performed. 51 This law established standards to improve testing quality in U.S. laboratories performing measurements on human specimens for disease diagnosis, prevention, or treatment. CLIA mandated PT as a means to externally evaluate the quality of a laboratory s performance. A list of PT programs Downloaded labmedicine.com from August 2008 j Volume 39 Number 8 j LABMEDICINE 487

8 approved by the Centers for Medicare and Medicaid Services (CMS) is at Some of these programs are accuracy-based and use commutable materials while other programs use processed materials (lyophilized or stabilized liquid). Assessing accuracy for most systems is not possible when using noncommutable materials. For this reason, participant performance must be evaluated using peer grouping. It is important to emphasize that even accuracy-based PT does not provide the same information as standardization. Proficiency testing provides a point-in-time view of how a laboratory is performing, while standardization provides information over time that allows a laboratory to make adjustments to improve calibrations so measurements can be traceable to higher order standards. Challenges of Standardization Programs for Lipids and Lipoproteins Standardization programs such as the LSP and CRMLN certification programs are designed to evaluate accuracy, to allow improvements in calibration, and to establish traceability to higher order methods; however, the standardization processes do not assess robustness of methods. Thus, atypical samples, such as those from patients with elevated triglycerides, certain lipid phenotypes, disease states, or certain medications, are not evaluated. Accessing comprehensive, representative samples in order to fully evaluate the performance of a method is a challenge in most method-evaluation efforts. Therefore, care must be taken when interpreting comparison studies in the literature. Of primary importance are 1) whether the comparison method used in these studies is standardized and traceable to the recommended accuracy base, 2) whether the samples used for comparison are commutable, and 3) whether the samples used for comparison are adequate for assessing robustness of the assay. These concerns are of particular importance when considering the direct homogeneous HDL-C and LDL-C assays without an offline preliminary lipoprotein fractionation step such as precipitation or ultracentrifugation. These methods have gained rapid acceptance because of ease-of-use and often improved precision. However, achieving accurate standardization of the homogeneous methods is more challenging because the integrated separation requires that calibrators include relatively native whole lipoproteins. Furthermore, homogeneous assays may exhibit a lack of specificity in atypical samples Manufacturers of these reagents continue to modify formulations in efforts to improve their performance. However, experience has shown that changes to eliminate some interferences may create other assay-related issues such as bias. Furthermore, reagent modifications make previously published observations of performance irrelevant. Users should therefore be cognizant of the potential nonspecificity and related performance characteristics of homogeneous reagents and carefully verify their performance in the patient population of interest before use. The American Association for Clinical Chemistry s Lipoproteins and Vascular Diseases Division is addressing these issues by conducting an evaluation of several reagent systems for HDL-C and LDL-C using a variety of sample types and comparing the results to the CDC s RMP. Total Cholesterol Among the lipid and lipoprotein analytes, TC was the first to undergo standardization, is the most straightforward, and is the least problematic; hence, TC is well standardized. While some analytical systems demonstrate matrix effects on processed materials, using fresh sample comparisons in CRMLN certification programs has proven effective. 55 HDL Cholesterol The state-of-the-art for HDL-C measurement has been reviewed. 56 All automated direct homogeneous assays currently available were developed by Japanese manufacturers that have certified these products through the CRMLN. Most of these primary reagent systems are sold to other instrument and reagent manufacturers for distribution to end users. While all of the primary reagent systems have CRMLN certification, they are specific for particular instrument and calibrator combinations (see gov/labstandards/crmln.htm) and are not universally applicable to all versions, instrument applications, and reagent lots. Accuracy may vary with calibrator lot changes. In addition to requesting evidence of traceability from their suppliers, users are advised to confirm accuracy of each new lot of calibrators by performing overlap studies with previously used lots or by direct comparison with a CRMLN laboratory. When using the Friedewald formula for estimating LDL-C, an inaccurate measurement of HDL-C leads to misclassification because the risk associations of HDL-C and LDL-C are opposite, and the error is reciprocal; erroneously decreased HDL-C leads to falsely increased LDL-C. Special caution should be taken by laboratories involved with long-term studies (epidemiology or clinical trials, for example). Of particular concern are lack of specificity for atypical specimens and lack of information about the effects of freezing. Additionally, the changes manufacturers make in reagent formulation may affect results, especially a concern in long-term research studies. Routine clinical laboratories should weigh the advantages (reduced labor costs, improved precision, decreased specimen volume) and the disadvantages (lack of specificity) to decide if they will adopt homogeneous methods. When a homogeneous method is adopted, care should be taken to confirm or validate accuracy and traceability. LDL Cholesterol Most laboratories (at least 60% of those reporting LDL-C on the 2007 C-A College of American Pathologists [CAP] survey) use the Friedewald equation 24 (where TG is divided by 5) to report a calculated LDL-C. Approximately 1% of those reporting used the DeLong equation (where TG is divided by 6); however, this approach is not traceable to the accuracy base and is not recommended by the NCEP. Using the DeLong equation can lead to erroneous decisions about patient care when applying the NCEP medical decision points. Because TC, HDL-C, and TG are directly measured values, the calculated LDL-C is subject to independent errors in each of those measurements. Thus, standardization of TC, HDL-C, and TG is important for LDL-C calculation. The number of laboratories using homogeneous methods for LDL-C is steadily increasing (as indicated by the proportion of laboratories participating in CAP surveys). However, a review by Nauck and colleagues concluded that more thorough evaluation of homogeneous assays is needed before they can be confidently recommended to replace the Friedewald calculation. 7 In the interim, the homogeneous assays may be useful in determining LDL-C in patients with TG above the Friedewald cutoff of 400 mg/dl. The manufacturers that developed these method principles have certified their products through the CRMLN (see gov/labstandards/crmln.htm) and have demonstrated that the assays can meet the NCEP performance recommendations for Downloaded 488 from LABMEDICINE j Volume Number 8 j August 2008 labmedicine.com